up!!! _ - dtic
TRANSCRIPT
AOβA fl3 362 STANFORD UNIV CALIF DEPT OF COMPUTER SCIENCE F/S 9/2COMPLEXITY OF COMBINATORIAL ALGORITHMS. CU)APR 77 R E TARJAN N0001eβ76βCβ0668
UNCLASSIFIED STANβCSβ 77β609 NL
UP!!!_ _ _
I _
I ~~~~~ .~~~~~~~~ :11r
fl~~
FO 2
________ 2 2
I β’ I
β’ 5
~ ~1iNATIONAL BUREAU OF STANDARDS
M~CROCOP~ RESOLUTI ON lEST CH A RT
- - - - -~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~
β β-Iβ
COMPLEX ITY OF COM B I NATOR IAL ALGOR ITHMS
by
Robert E. Tarja n,
~-*~
STAN -CS-77-609 D D CAP R ft 1977 L~I~J1TL1J~~~~
AI~~ 22 ~~
~~~~~11TJ E
~:-
~~~~~ ~~
C O M P U T E R S C I E N C E D E P A R T M E N TSc hool of Huma nities and Sciences
STANFORD UNIVERSITY
>-C) S
Cβ,β’ β’ wS
~~~~~~~~~~~~~ _J
~~~~~~~~~~~~~~
!Lclacci:J ~2β.L β β
S E C U R I T Y C L A S S I F I C A T I O N OF THIS PAGE (WI,00 D~~,a ! ( r Ue r . - d )
READ INSTRUCTIONSREPORT DOCUMENTATI N BEFORE COMPLETING FORMI. REPG9I -~~tfM~~~ P 2. GOVT ACCESS ION -sC. 3.- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
A - / tji j / /I ~~~~~~~~
. /~~~~~~~~~~/ /
. /. I
4. T I T L E (aod SI~b t i t l e ) JX?.E GE. ~ β¬PO~~t & PERIOD C~~ /E fl E D ,1
tcc:c ical , ;~j~si1 l~~( 7β β - ---
C3i U i L ~ Β±i 01 ~0 L β - ,A~~Q/ .~~~L si~~ U:- ~ L I U ~~., 6 PERFORMING ORG. REPORT NUMBER
7 A UT N O R (~ ) - 8. C O N T R A C T OR G R A N T N U M B E R ( - I//
:c~~c s t ./ar :ari β’ c c ~oi~_
~~~~~~~~~~-- -.
~
-
~
~~~~~~~~~~~~~~~~~~~~~~~~9. P E R F O R M I N G O R G A N I Z A T I O N N A M E A N D A D D R E S S ~~~~~~~ ΒΆ 0 . PNOGRAJA~~ L. L..PBOIECT , T AOK
β’ A R E A A W O R K U N I T N U M B E R SStan~ or~s β’c~- ---3 .L -.β OCIj .~~C! CC ILflC C I ~par :iesF~S t an : u i c . Ca. β I
~.0 CC -
II. CO NTROLL ING OFFICE NAME AND ADDRESS . I~~. REPORT DAT EC: ICC o~ I,a2al i~esCarch / / Isj.U~fl k9
~cpai tmefl~ o~ the a.~, I~ 13. -
Ar l is i -to ss , A 1. 17 u1& M O N I T O R I N G A G E N C Y N A M E & A D D R E S S ( , f fj f f e ~ e,~t f roro ( o , r I ,~~f t , r r g ( ) f ( r ~~r 5. S E C U R I T Y C L A S S . (of β)β~~o~~~r -; 5 - , :
UC\ ~,c~, - r e sen tat i ;e : ih i l i i .1/a . . .- S. .nc1assJ~~j.~
j.ra: c-J ,~~ ccCa ~ .LCc - .L.~ . β’ . 10) , β /aβs a ii ~~~~~~~ / ~~
I 5~ D E C L A S S I F I C A T I O N D O W N R A G I N
S . . S - , S
~~~~~~~~~~~~~~~ / SCH E DU L Fa..: U-/C ~~t . ~0~ C)
6 D~S~~RI0., r~c)N S T A T E M E NT C,! :Sio F?~ po, t )
~oolCacablc wi th ocu lIrUi atj ons On i J : -!:. a .~~~h
17. D I S T R I B U T I O N S T A T E M E N T βof the &,srract entered I,-, Rk ,k 20 . ii di f fe re n t fr om F- ep.~~t I
IS S U P P L E M E N T A R Y N O β f I
: , r U i r r ,r- on r ev e ro e ,ide if r ,e ~ ~o,r or,,! ident ify 1,, h t r r r I r r r : r m t - , r
20 A B S T ~~~~CT ( I w : r f r o , e on r e r er n e si te If r,r r oe:- or:, ! ,dentif r I,: h I r , r k , r : ,mher
I t i c ~~~ - s-:a.C .i: U ~/ . s~.u r ~s c c . t i lL β . - c . i 1 :-H t ,, a t c t . cal -orit ioro; . ii .- b1 i ~}c t i Lh cc a1c~.- a r c . h o s(or -: . c l i - aat . . . - c ~ a cC cal tools ~sc u ,a :i t } 1 r . l ! ,b c i
~~O . .11/. ac.L , ia L a c - . :oβ2 c ,! r i r . c H β C . , . I c i , c .~a .c t o icica: .rcthU :orc ,i 1- .::J t~; ot an Zi.L . O i J β βiβ .5 cu,,c t t i c , t . c c V U . β : 71, c L- r c . a i c c ~~~~~~~~are ; r -~ l a o - s to oolβ.βo tools i c c t J. in thu t ied ~r i 0 100 .5 aJ . a o ithic c a: .:β u t - C l ~~~ c- c , . un ,unt o in al~ u i - i S , i sn :c l or o o l : In I Lou c -
~ c c - : cc c.a t i :c - J O 1 U D . : .ho c a l , ect.i c ri : . 1 , T o t .: Co flKi .1 r :Li.u:j :: 10β 1.0 cc r :- a : : :. .
DD . β
A~~~ 1~ 1473 E D I T I O N OF I N O V 6 5 IS O B S O L E T E
~J~~~~j~ j~~j/ / 7 /
5
SECURITY C L A S S I F I C A T~~~N OF THIS P A G E II~~~, t) o~~~~~~ r r e f
- .~~~ β ~~~~~~~~~~~~ - β ~~~~~~~~~~~ β
___i__i_ ββ .β -~~~~~~ - -~~~
__________________________________ β
C omplexity of Combinatorial ~ i c ~ r I t I c : :β
Robert ~Indro Tarj afl~ ββ
Computer Sciene~ Cep art omor tStanford β.Jiiivers.I ty
Stanford 5 California oL3-)5
A b ct r a ct .
T~H s i a i cr exanines r oLont wo rk -:a the c. :: loxi ty 01
combina tor ia l al g: r i thm :, high l ight ing the a : . r n: of thc β w o r t , the
ra t .h emat i ca . tools used , an i the i:o~ or tant results . Included a rc
oecβt i no di :cussing way: t o ::ca: are the complexity :f an aigor itboc .
caclh -do for r -v in g tha cert a in r ob lecos are very hard to colvo ,
tools useful in the der ign H good ~~g - r itbms , and re:uβnt c-o r ve:.c r st s
in alg~ ri tb r.: L r solving t or i rd reoentotive ~ r.b~ e:oo. Tho o ln a l :-oct I
suggests some ii rcct~ ons for future research .
I.
-
I .1 βrf . β β
Based on a talk presented at the Cyn i. osiuns in Honor of thu ~~ t,c:Anniversary of the Office of Naval Research, C lAN 197 Fai l ~ec t I u : : ,At lanta, Georgia , October 1 β - C - , 197:-.
Research partially :βu~ o r t t c : by National Cci on: Foundat on ~rcHM CC75 _ flf l Β°7t and by the o r f : c e of Naval Research c n t r a c~N )C1L_ 7 ..r _
~. . ~~ . Re Tr cj u ctj j n in wh le r in arl β- Β° I e r r
for any purpose of the United tt . a t u s Government.
1
5 - --.5 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~ β .
intr od.uct .LuIl.
In recent years there has been an explosive growth in research
sealing -~dth the development and complexity analysis of combinatori al
JC.~rithms. While much of thi s research is theoretical in nature, many
Ut the newly discovered algori thms are very practical. These algorit.hrr .c
~id the dat a manipulation techniques they use are valuable in both
o:r:binatorial and numeric computing. Some pr -oblLercac which at :ir:t
glance are entirely numeri c in character require for their efficient
.- - olution not only the proper numeric techniques but also the 1 r β - 1 1 ~~~ als I c e ol
data structures and of data manipulation methods. An example of such a
cublein is the solution 01β a system of linear equations when thu coefficient
. β L r l x contains mostly zeros (Tewar:-:, n [ 1973]) .
in this paper I shall survey :ome of the recent rocui t,: on
c omplexity of combinatorial algori thms, examine some of the I dear behind
hem , and suggest possible directions for future research . Cection 2 of
ho pai er discusses ways to r ::ca:u.re the complexity
:1 algorithms . Though several dif ferent measures are u :eI βui in di f fe r e n t
circumstances, I shall concentrate upon one ::.ecc: cre , the wor ;~ - - ,ca ,o e
running time of the algorithm as a function of the input slo e . Cect ion 3
t o :cu:ses techniques for proving that certain combinatorial problems are
very hard to solve. The results in this area are a natural ex~ cc i : ion ,
erhaps more relevant for real-world computing, of the incomp leteness
and undecidability results of GΓΆdel, Turing and others. ect ioI~ 14 r eoerst :
a sma.U collection of genera]. techni ques wh ich are useful
in the construction of efficient combinatorial algorithms. Cection 5
JJ: cioreo efficient algorithms for solving ten representative problems.I .
-7, - ~- 2 )~_____ c c β< βCLi 7_ _ _ _ ~~~ ~~~~~~~~~~~~
β β!~
\ t~~~ .~~~
-~~~~~~
,
~~~~~~~~
β 5 β β β 5 - .
- S
-β ---~~~~~~~~~~~~ - -~~~~~~~~~~~~~~~~~~~~~~. ββ~~~~~~~~
-
~~~~
β’ . 5
- .- - - β -~~~~~~~ - -~~~~~~~
- . β~~~~~~~~~ _ _ _ ~~~~~~~~~~~~~~~~~~~~~~~~~ β - ββ- - -
Ther e robHoccc: .1 1 lu:tr:ci.c the imp - rtance of the cc Lb iii: in Cacti ~n 14, an
thoy in clude :ooic , but certainly not all , of the corucbinat tr ial sr obl ercc c
c r whi ch good a l g o r i thm s are known. Section ~ suggests come un:βclved
c c- l ens and directj.nc for future research . The acr e n d I x c ri ta i ri: a
list or t orm in o i βogy for th :e unfamil ia r with gra~ 1i theory.
2
i - - - β r - o Ht~ Ioato cn - oL1rr~ 1LirTiTT~ti~ IdtLβI~i~~~~ -oio~tL - , -~ T ~~~~ - , β- ~~- ~~~~~~~ ______
2. Machi flc β Mt H. 1: and Io::1~ f e x I Ly M :a: βures
tn the early year: of conic ut-ing ( b e f o re c oci] afar so lence am::
recognizable as an academi c disci i iine~ , an individual confronted with a
computat i onal ~r cb lecr: was likely to ~r to ed in the fo l l ow In g wal . He -or
shed would ~-ori dor the problem for a whi Ic , f ,~~ uIa te an algorith-o c r
i ts r lut i on , and. a r , te a cc. -::c ~uter :r ogra ~o which w cold hoc e fro l ly .i rrj - erc ent
his :tl ~~ r I t h o r . To Lest, thu al guri tbm β s correctness , he a U.L J run the
program on several sets- of dat a, βd β -:bugging β the pr - gras: until it
~roduced c r-r ect output f-o r u : ,H i srI. of sanl le inc ut . To f o r t the
algorithm β s u f f i c i - n i cy , he would :r:ea: u1β~ the t in e and storad e :~
r i .c o tled by hi s c ! - orcs: t~ c r car s the sample da l i, f i t these measurements
t βurvu : (b y β;β~~~~, by J o a . :tβ :quaros f i t , or by some ::tbor :c :Hh d ) , and
claim that L }i β:se ourocs measured the e f f ic iency of th e algor I h . .
The -t r :ca β, a c k : c I t t hI s ~-cio~ ir ical a ~roaoh are obvβl d i : . The d- v e l o : c : -n
O f Vβ -:r7 : βLrgu 0 fr a : :. such as conr~ ii err and operating :y:t oco : , re ,i o . res
a .roiβh r n - c r y s t - r o c H i c method - o f checking c or r ec tn i : s. This need has led
crc co t :-r scien ~ -: St s i o deal se meth -o .i . : fo r 1crOving the c : r r U :~ n ess (an o
o t h : r Γ§r β ; er t .e s) of ~: gras:: (Floyd [ 10971, Manna 1 ~, 11 -a ra- [10 ] ) .
These methods use rcci t b :βnatical i n d u i r t i o c i to establish that cur l -e n invar:ac1l
ru l at ions ti - h U t whenever cer ta in -Β°i ni s in t he rograβ . are r β ci. 1cc i . I To r,:c ut e r
SCie r I t i ~~t s have aIm r β s : ~ -ooi math - dr (such arc β structured jβr ogra βmningβ
r cc n : t r u o t ing eacy-t- -understand and easy .tr - v ur i f y r .grrc s ( I c rchJ ,
D ij k~:tr :i , and Hoare [1972]), and hav e formul ated new rograrncning I uigria ~ e:
t. mac c t h ose :cct:th od : uasy to a~ ~- oy (w] rth [1-1 7 ] .]) . The ] circs: t of t h i s
research is to - 1 - m motr a l ,e that deal ng an algorithm and Ut er i sing a r o t β
of it s cor rec tne ss are in : ar abic art ; of t sacs- J r - ce::. Ierb . - i: sβ β ___________________________________
~~ H r r ~-efor th I shal l use βh e β to lo u t o any I nd kr i -iii-h , male d r i β uncah
L β
~~~~~~~~~~~~~~~~~~~~~~ - ~~~~~~~~~~~~~~~~~~~~~~~ ~~~
- β
~~~~~~~~
-~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~
- ~~~~~~~~~~~~~ -.
r-ec- : , ::t air, onte j f t h I s point 0f v iew 1: hI ::~ ra ( al l , h I :k ;t r a
and I [:cu-o [1972]; Di~ k :ora [197 ]).
Measuring efficiency by means ofβ ercrririca i tes ts bar: the sane
lo~ l ciency a: checking correct ness empirically ; there is no guarantee
that the result is rec r β;. mrc ble on new sets- of data. If air informed
choice is to be made coer wee r i two :,afcrlthm s for solvi ng the :aooe orob lem , :orcu
-β ore :y:teca aβci C I :Ut β rcβ :~ ion β 1 cit the algorithm s t cornilexity is cr eecied .
To toe co st u.βe:alI , thl : . i : , : oo rr:cition should U -c cro β i . I s β β i0: -~~ andunt : good
:Jc -r: i t hro: t en to- ror.β~c. good ever: if they are exl:res:eβo In J if ie rech
r ~rani n~ I :co1nia1-us- -~-r run on d.i. f t βerer:t n:achirie:. Fuirtherni , re the
β . -cr :mr e shoul d be both r - e: βi :tic ouc h iou. cep t ib l e to theoretIcal s tu dy.
~n!::plCx t v r ~~~~ :Uy , ~~~, -si ta. mInd: : those which are
st as~ ~on,ie βc enaent 1 ale size onci character is t ics of the in c βot d a t a t
- m d u i - s e w i u C . - .rc i coni c (ccc endero t upon the input data t . A ty1βiccJ
: tat io measure i s r uran. length. Irogram length in some sensuβ measures
nbc si:- .~ l i c i t y arc -i elegance of an algorithm (an algorithm with a short
c -c βr im and short correctness pro c i is s imple; an algorithm with a short
rogrcrrr and long correctness roof is elegant). This measure is m ost
a~prco rl ate if J r-ograrrolrig tin e is important or if the program is to he
run in fre -nuent l y.
11β0rn ami c c-or :m lexity measures provide information about the re: crrce
requirements of the algorithm as a function of the charac ter i s tics of
the input dsts i . iβy~βic:ol clyt amoi c mea sures are running time and :alr-:gL-
ace. There mea;:Ure : are ~ ~ roort ote if the program is to h~ run β c c : . .
Running t~~ e is usual~~ the most important factor re:t.rIc~-i c;g the - . 1 o f β
problems which can be solved by computer; most of the problem: f-c t l i
examined in Section 5 require only linear space for their s :lution .
14
- β - - - ----β- .--- ---=-- -~~~~~~~~~
---- -- - - -
_ _ _
- ~~~~
β.- ----~~
β-.- ,- ~~~~β’β’-~~~~~~
.β-.- .-~~~~~ - --. - β--- .-
~~~~~~~~~β-β-~~-. β__ _7 - ______
11 W o o β! , 1, r ~r.blorns with o i f loa r βt - c: c algorit.hβn :, so rage space :r:ay
i - : the l imit ing fe - c r . it βrage r:c ace ha: Deen used as a r eassure in
pr- oUr of the c βm o utat i oncol ito tra ct :ss ility ~ 1 certain 1 roble:r o: (see
Section 2), but c ost efficiency :tci-ot es -onjicasize runn ing t ine .
Dynamic measure: require that we spec - i fy the in at data. One
s:ibility is to assume that tb d.aΒ±a U cr a given problem size is the
w βr:t c oss ib le. A worst-ca: mea sur e βHβ running time or st rage sl ace
arc. a t uj icti crc of problem size m r dos β a er form an c co guarantee; the
I c- gras. will always recuire no more talc - r s~ aca than that specified
by the bound. A worst-case measure is in this sense not unlike a ir o o f
r -gram correctneso.
For s ~rre algorithm s a w -r s t case bound may be overly ~e:: i m ist l C;
f o r in s t a n c e , the simplex neth .o.d of linear progra~~ing (Dantzi g [ iβ ~- 1 ),
w~~i c i i bar: an 0>: : -n e ci t . iii. w - rc t β case t i m e boun d (Klce e and 2 1 nty [ l2β(2]
s can s to run much ~βa:~ or than ex-ponent l al on real-a r . Ut r c-bie ;c.s (Dantzig
3 1) . In such case: an β averageβ case or β ruc r β- senl at vu β case ma:,
give a cnβ or c real .I :t Ic to crc t . : Β° rβ c e r t a I n r i l e : β . domain: , such arc s co - f icc,
and searching (Kn ot t ; [1 ~ ] ) , c βoo rac,β β case anall, : s is aLβcco:β t : .iwcc, : ::i re
reali st ic tnan wor st -case -era. y :i a , and i n these area : much :irer :tc,s-case
analysis hasβ been i cr i e . h r -wcv βr. : rv er cr f β-r ca , β βma cd cβs I s has i t s .i!β:iwt βacr
It may be very hard to cho rc a a good r :bah . l it y mea su re . For icc r a n β - β .
assuming that d i f fe ren t art s of the cc l ci data are Inda~ u n i o r h I :
d l r c t . r i t o i u t - e d may make rhe ana ly s i s ea :iβr but. rc o ,~y to o an unreal i:βf β c
as:ucco ; t i - c rc ; furthermore evβ~n a c - at .vely s i m i l e - r I c, - r o thin may ra~ ~ Ily
destroy the independence . Wi th averageβca se a na ly a l r c on -β a ud : t l -- na l il;
run s the r i s k of t e l ng cu~~ ml sed by a very ra r e but very bad. set of i c c cot
1βi t a
5
- - ~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~ βv ~~~~ll, ~~~~~ ~~~~~~
or. _~β ,.~~~~~~-- β
- ______
:00 c. c : r - β βn: c r:co lt . :-a t~- , :β .~- : :rrr . crust be L O L C - .L U a c c r : c : n β β r β ::c. dei . :cc-
:o i s s i s : - t.~ ct n lcc β is tb .β ran 5 β- β~ β c u - . , . : n a c t s i r j e ( Co I: cocci -cobb - . [I β - i ~~~~ . βSSc Oli
1 rc ou r aβu :t ro~cI-ion Ot β :s o.βo- rβo.c β o orl i u se cr .I 11 :1 i a I co - c o o cc . Ttio cc , c-y 01β - βoc t.
a -: ,ach int : co::sβ I β i rc c r os : cr0! ,:; of :t .:rcfe cc . . , eniob nt -ic t O 1 . io an
.i rc t βgor. lice rctucβccgc s Ills c u e sos -cU re : c:. ucc u t in βsloβ f r β cc β coce the c:.s β : , s
t o i c , cell is .11: :1β- . Th .: r:ca ci: i~ ce a.i:o h :0: a c l x β. - .: : 1 :c.I te iool-
c β r ,oI : r u c β .. each able mu Ic ld, an iotc - β:β rβ . ( I-o cr s o L o u : β .: .I- β:- β]xH, β
r :uc::s β.:c , we all- -a sβt- . rac,e cells ruoci, Iβ - βrt . βOe r~β t ic .l i r c~~ mu m~~r s. ic;
omo . , βH isβ n :β.ctrlr: u -cam t o-noi s ier tt - , e .: - βssβoβ β:-t s o c r c i : i β :r t o a Il c m
cell choos e βcct hβ s - iβ 15 ccc a regi :tu~β . :-β β r - - : 0 . - c - to -s r β -~~~~
β - t c r Ice c r c - c d -
c c a :c-:rcβ.c,β: ccli ih ,-:e c Ads- β: 1β icc cc regi :t β:β, r per: β cβ: , - c oβ.ri t ism c i :
-c e r c β . t : , i c rr tic β: c o n t ents of t o β r- βui .I - : 0,: . sci .o o::ci β: c β .β t b 5 o I l - - co t s ~~1 β t a
o- -:g ster: . A :~ asβs :. os o I l e d : βic . I β - i β βc yt t : ~~ β~5 i f l β:β the : t u ~ n~~ C:
crat ion; β to be cβ .riβ I c i mit . I t s β β .LrsI βL :il con :β~~mu β: .1 r . f c o o
rer resor t : tis βs _ rsc ut . β , a:;.: the f i na l e - r o: β I ~yao-ac-i n cc β c:oc r : co r I iβ β :i Iβ. . β . c . O S
t ire β~~: ~s . . Tire detai ls of thi s- mach ci βs :β . - oie l sm- c c,uβ. c :o c c βo I Si t i icst
r o s s c - - i l β : n a.r-i rct .i - o r .β ic rio t a ff e c t 0100si .brg tIc::β: ocr Il. 0 0β - .0 aCe by more
~i .no. a c r ~ tant : ant -.-c - .
A ore . -c c . ccc β - .c sβ rn-- ich icrc ,i s :eqricβcot β β2 : t o carr . I - , ou t, - -ro e step ct
a c In: β:β . t,rci wool : has co .βEc fl don e in the :βcβu cot all nod c cβo lex .I tm of
s a r a l u i l βd r-: r β i β h c β .:, i.-o.t- I shal l ci dJ. β~c:s βHit , a r-lc here .
βrh o r e : [ rcβ acc cz : rn ao_βlc . ci β cβ ,~ deJ. r - βml he: :β. useful ft c, - r
r βoadi s f 1 c:adly measuring t h e e :β :βic ienocy H β βirmi oco.βr c o o :β.ig. rβ.i tb:::.
but It has serious -ir w c - β c . β- , 5 c r . , s~β c- cc- uomt - t.u t c e : . Usiβc c β a
s i ng iβ : .βt βc r a .ge cc i β1. can h i d -in csrt .. .I 01β c,; lc βorsie ~r :t β - f e c . ci t I
J s~~iDβi c on a roe s cii a c β :;:: c, :aoch.:~~e I , . : , ~βrβ ; out c 5: . ; :,i cI! I - i i : .2;
arallel by encoding :ev β:roi i smai.l β., βctr t β . βr : ic . t s .ccc larβg β β c c β - . 7rs β’:
ββ~~~ β
~~ . β .
-β - β- - β - - - ~~--.-- ---β - --rββ- ββ β~~
-
β s c ar 1. 1 te l : s o ales: c β β cI : : rL : . .i c s g th a t the t b~o c-la i r I r a n ircalger
β βci I ; 1001 oIl asci i to t l i β length of it-: b l c s c c r ,-; r β. . r- β.:en n β .t l on
Ito ~~c r β . : O . cor d IJIL::an [p c y ~~}) , or by reβi o c r : . L c tt: β1t βdi i c c t - r cr : beβ~.
~ su i ted isβ. absolute value by some c .cr stcmt t ic : . e: the s i ze of t h e li~ i,ot ri ot a.
ban d ic - :βa cce :~β cc- sIr m e : β βu-cβ extr βcoelm β - aβ .β.r:βu l 2 -: co~~t l β s . 1 β . ( L , l;
I - r : β cβrc : βu i t h o β.Il c c :~c a ico -o sser : . This ab i lt i to L u : ~c : β:sl f o r O s : re: β β..β c c I c c β
- s . f i .c lc : ien : t -n at co r r βc, β;s ( Knuth [ 1β j ) , r ert β o r:- :I c c g r c .o l ; - β. .rc . ( iicUL β:β. [ 1 β I -
c r h s ~ β ha: lc c ab l e s ( Knuth [ lβ7 β1J) , and the l i k e . II woo β in a βI β :o iβ..I r β . i r sg the
tf: : :rβ β i c c -c t l i m i t s of βt hi s c:io :rbilit. y seems t be a har . i ;r
- β I β:c βgocc β : t -~5 % ] . K :1:0 -gc: r ~βv and li st e c r s k i I [ l i β 3 j , tAr -cl . [~ 3
β cβ iAi: :t. βcge 1 i~~βx ] , :icc h Tar . i an [1β cβ~ } have c r - : - :ed β .:ri c-hc l c . β β ci. I sisβ I s . w h i c h
- β . - .c:es: to β. - β : c - rβ : I s Uβ; β::β:: licit re β: rcr .- r : e cl i . t in -h no as: c r o s s - r I 1.h :not I c
1: -~~sij bl Lc . I shall call o uch a :β chIne a J.iucl -:eJ β0 - r - ; :c , ac :r β cc,o. Iβhese
- , β . i , :c c β cs ao . :r r : c t ~i; o..ali the c r c ui I l lt i e s of 1 1 : 0 β : : β . β - sing I β S r f i:~ β .c
s corc h a: I i β s βso t Ucoc β Iol , .βi . β : c c β : : s l ccg fe ct o r es al β-β .f l ol β:,_c, β , β 1 β so o β .crgoic cβ: .
crc A J g ~~~ arc -Ut i L l , coo t the β,β ar c ::β.r to bβ:β :oβo- re ccj cm β :: ,orb ole t rccai.βm is
ir c e c u cβarc b ::,~~~β.βO O C 5 5 c c ccoctc n , :
Ai r ~c to.-r s β βi~; s ic : . β . Ic :c β ., βiili e : 1,1. 1 1 . -S Thr ico c β c :: :cchri c c β . (~~.: rβi ccc [lβ β~ β~~~
ha: β cccl cc: βcJ , in c orer , , t i:β: - r - - t ccii. . c : 1 β , . A T c c I i r c g :n r , c c h i : o .
cβ:r r s ..i ..i ing :5 β a tai cc . Tbβ β. 1 : β . : β - i s lv .1 .i~-β.-i into S u . . , c - c u c-c β. Sc : cc
A 1 o h ] rig c c β . f a 5 1 Il i β cr u cc , βcoer c β .~~ ,-β: :. cc I . βIi:- . c β : c rc c i r I c ββ : I ~ 5 .β~~~ .β .
a f i ni t e a lt rri al mccc -r ,β and a r ea.u w r i t e head ~~iich ccci s i : r e I n e
emi l β.i. a t I c β . β β . In erie ,. t o β : , the c :: c .c. c ~i I n c roar , r- c: ic c c~ tai β.β c e l l . aol e a
n - βw s~~nb ol in the cell (erasing ~~ - c t was there j red ~: o~; l , no vβ t~~
β.3 βwr it e head - - o ne cell :or- w βir d or ba ckwar . i c . - I . e ( :11 - , icn .d c-hang β-
+ - ~~c n t βcccc-d oco c . -:r cct β ,~~β β . The : Ieeism βfl ~j~~: ~. 0 β . c β .c - β t u -orb : 0 β β:
d βc1βccnd:: :~ri Ly on the current i c i t β.β.rna~ 0ccβ. rβ ; s ta t β and t t i β β β . c :n t β r i 01β
the tat e cell b e i n g read ; t h I s dcci c β er r i o enc led U r - βac : l c d - r c A
~~~~~~~~ β
β’:
~~~~~~~~~~~~~~~~~~~~~~~~~~~
1~ - - β- -.-- ,β.- β
_ ββ ..ββ,n,βj-w _.β~_ - ββ----_
~~~~
_-
~~~~~-
1:-r i e a~ d each t.cic to :βrc :.lβ ol in a dec is . i ri t able β which t. β s rr::: ths β~~~c β -~
01 ti le :cc c cdh i ci β . β .
Thr ic ig ro~ - βr ed hi.β ::. cβ.chircu model in lo t β , b e f o r e el. e-ctr . r , I c dig ital
5- : ft j t β..:rs cx . : s t em ; he β so::r s ~ t t o -~ t sUβs β to model comjs utational r c u s S es in
the 0 0 : 1 1 β ,:: β’ aol t i cout oi.β .c β:r βccce to cβu:y 00:0 . computer. βl riUUg :l Thrs:r β.ββ
01,01 is i:.:- β..l:quatc 5 : 1 - 0 o ar -go a r c of cc - r oc re to c βsscni:lexstl: r β :cesor : i ,, .I . t:
- ..:cβ.pL.rcitt ; curd the fact that any r:.r co. .β.. - cβ . access ::cach:Ine cars i c
.~- i s a βIβur~ c o rI n .aci i .I mc with - βrrl ~; a i olyco oucrial bl~ wβ u ~ in roco c: .icsg t ime cβ . l-:e:
t i r e i oUβirlg machine oxtr β.c c : . β.βly u:β:S βHo βor studying very .11 I c β I cult c c ::: u t β .I I :c . βJ
tasks . It is: also valuable f r :1 rI ; rig o r 1 i L-c :: : w hor e ta o β:: are the
orno β : der i c β.: , Or : for if l . β I β u c: β: I r β . β a: - β s c-t l rr g (Knut h [LΒ°75}1 .
Tn 1. wo o t wo-a soud -:5 who :1 s β~i, is . : βt.cc cc or, :or.re cr 1 tical :s er β.cc arc :
One . c . β,int S in the rucs. ng t I::.e ooβ.cwβrβs: oβ e: oilβ,β .~β:β that c r c ictr i. .r c
For i n st a u r c e , βs ri : rt . .Is , c I co Os.i :β,..c-cco l fl 10 : 1 c c : : it is: .rs β:s βuI βtcβ c βsc β. t
c -scsi ar t son : (or genβ:ral t - ~ rr c .ory dcccl 51 - βc i : β . c:easuri c:g l i i -: c β .: leo-c l I f β .βf a
~ u- si-cm by tb β. d o t ic - U a - i - c 1 : I βcc Ire , : : β it ( βill.) , II Cr I I . arid IiiL:.cu:
[ i β β f } ) . In arithi: β :β :t .,β c non βo i s - β t β c - oβ.l c r d - c β - .: , i t i s us d11 t o o~ucit
βucI (hrc r β :t io ~βs ββ r a t . i -o . rc : tur d β o a:cβ βzca: 5 : . : β . β 10-s deci s i ro : art β .;,: ; I . e . , βHs~~c
-- cc. : 0r at ] t ic or : -c β: : -cr β: I : c β i .c , β ::a β cst Ot β t i r e i c r : cu t -dat a ( I r a sot cβalcco
t r l lccrr , s i r : - β . ) . In this -n . e :ccno β β β0 O t t β S t h c β C.cr~ t e x l t β, U a c- ,:o- loc β , t1β t u e
lo: rc~~ l s u t - s - - rtc i~~ht - l l ro ir - gr:ss (-b . , β c or U - , ce~a - ILβ:β, , ~1 ~~~ I
- - β. :, β l :i tu a t c s c s : ctc.n: r β acce .. : β : : slay bc the cri t ical β c - c .1 βcc, .
r thi c- o c r I c I r - ~t 1 usβ: a : β r s t β β .β :L :e ru n n I c r ~I t I nt: cm a c β c s r β r
cac:o ,I c s β : βrc ;β a cβ .β:as ure of aig 5 r β . t h c c . l c c a :cc lexic. - , . I b i s r: , eβ :c:OLβ - is
an ti r ββ.aai . t i c O βer a wi de rangβ: of c c c c t i l c . c c t , - r I ad r h o β:β . : . i. s h ot_I l.
I ~~c - r β~ cor i _ : fac st t ac t . r . in runn l rig t O r e , tic .. β SUCh f l ; β I - r c . j ct S cot_c t. .βiβ: sec e d o
upon t i c - β exact c:. - d l , of cs. - . β . :c t scl I cc , l Ice β, β are c i t e d hard l o C β:β I c t , , . arid
L. - - ________________
β β ~~~~~~~~~~~~ ---
~~~~~-β-β - ββ
~~
- . -
they ten d, at least for large- Si sed. problems , to be washed out by
:ocoyβcco~ tot ic growth rates. To indicate functional relat ions :Lir :, I
shall use the ollowung rc β .t at .,on . Iβ.. f rβuios g are func βt βoo s -us cc
f ( c r ) is O ( g ( n ) ) β mean s f ( n ) < cg(n) for all n , where c is -a
suitable r.o:itive constant~ and β f (n l is I β . ( g ( n ) ) β means f(ro β cg(n )
ts r all n , a-bore cc i rs a suitable positive cons tant .
9
- _~~~~~O~d _ I~~~~~~~ β - β
β - β---.---- β~~~
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _
- .
β c . : I c c i ty ~fβ Tnt coot -rb so iroblems.
c i s c i re .s by ~ll Isβ:ri, ~Imc2- ] end other t β o rocrali st :, c :.a therc .00 . 0 ; βml .ans of tic-;
eβ .rr -lv tw er r f l e t t i -coci tur l , -β bsl eβi to find a :β ors.cl system which w~β ulou be ade βtuate
for ext re:: I rig and ver~ i ly i n g all matlrerc .ati c:c .i t i-uI A : . These bar β- s wer e
5 β bed Lv lb S . - - [ I d l i . who i n h i s l β c.c s cu s I r s o - . ::i5. l . β. he roes: I.he .iβecβ c ie c β .c;n:t r co t o :
that ci - - ccc e t icoct of βsro β..βU coul d be b cl i : subjec t to :::echaroβscui , v e r i f i c a t i on
and - :werfui en βug h to 1- - . Oe all the~ re:::s of β el,s β . β . c . β - β c.β.r l tt:β .co : t c . Their
i c , t e ro : t in the :β ;wcdat l -ccci: of cc .atI βicβ ,:~t . Ic s l i- ac - ,: t ed logic] βus: 1 . cc - t
the ic~::I. .I β - c c , βIi:ct is mechianical β-: β .uA c .1 cat] - cr ? β or e 1uivaient lβ~β . , ii; t
: βur rii.g- .βritbr:? β . C h a r d β . [ lad- ] . Kieeβntr [ l β 1 7 β] , 1 oct [1 .β t β r ] . βAac- . l :sg [1 β~~~- _ 7 ]
cord βj cJi~:r - ~~~rβoaβ~ dβ. β.1 :β. rcr ct_~ βiβ: c : I r ;c i t i . arrs β of an aJLg ββ.r I t I , :: . . Those dβ.d βIc β l f t i c r c
clO d allβ,β J i U : β - .- c- - βc c t i.::t r o b .:~ equiβ:a1ent~ in βI t re s β-rr :e that
a r-ocb lem is solvable acβcur β.i I : β . β .β t β. β Y e ..tef i c c l 11 . - c , f an c.c..J.g r I β . 1 c c . , O i l e r s
11 . 5 solvable accsβrd, ng to all t ic β - -c t t βie r del β i n f t i . . r r : . This r- βccu , :l.cc β ::s
a t β the nut .β. βr . 0:β an ai g.o.r~ thcβ. is usually stated. as Church β s thcm ct 5: nor :,β
c.et g .- rit}βrnβr ( in the i r i l β oruc c ai cer i se ) car r be expressed as a Tu. r ircg :c:aciβ,i:, β,.- .
and an:,- βru.r~ c : : cβcac hira.: ox~- resβ .:es an aig: ri 010.β..
βn c- c a β - - rccc-i. del β]: . .. t ion 01β an alg .riI-hm exi : te- .βt , it war 1Β° :sible
c . r cr.a thecsaβt .i c .I air : to :I,udv the . wec β of c omput at i on. Thric og pr βcve ci t hat
n . algorithm βcx i soted for - .ieterioii r in g wt ,c βI }re r a given Daring rc: βschine wi.th a
g wen input wβi 11 ever halt. Other re :ear chers d,l c ocred a cr βu c f t βer ocU such
andecidable I .rublems ( . β . .r ie c {i: . Y h ] 1 , wβrc lo h corrcso cial ic ci c o och er science
to the irneomi leteness results β. of β ].b [βoβ i. and -other: in logic . Ic βrhac : O u t
cc oo :t orr βo to t h i s rΒ°search on c rca crt : ::il ity is lua t . di.-β evi c β 5 1β 0 : r 0 β ,
bu 1I ~ Sing an ear l . ] -o r ~~ rk by 15rβ.ri,j n Davis and , r u l i ia Rotc irics - c i . t h a t Hit : β. βrt β :
t. β -rc 4 ,h i:r β:biem icc undecidable (Davd, .c , Ma t ij β .::v cc. and R ob ins-s-c o [117
H u bert β : tenth r u t - n c is to determine whether a given polynomial e~ uat oi β. β cr
has a solut i on in I
10/
_ _ _ _ - - .-. ._.
~~~~~~~, .. ~~~~~~~~~~~~~~~~~ Si 1.β...,, co. , β ____________________
- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .~~~~~~~~~~~~~ - -β-~~~~~~~~_ _ _
Two proof technique::, di agonalization and. sircouj atl . rc , ero β , he
computability theory. Diagonaliccati .or i is based on βur .: I βcc, β. :eIr β- r β e: β eβr β - c β . c -
raradoxes; Cantor [1 ~L ] used .t to r β,β.: t coao th β- :r c . or- . rβ ear nuβ c c e r , . ~~~~~
I rcI .egcβrs and G~ del used it to or ove I i i : icccomβt letc βrc β:cβ . !β- . :s ~β β . . - c . β c c i .
use it in the followi,ng way to devise an undecidable I c β c 1 - c - . t I~~ O:t β.
We are i nrterc:ted in y e sβri β . quect i sn: at cut t i re i c r t β:0 . 0: , :uc .- ic -is β 1: to
even? β r β Is: n rime?β U-u t - i β β S C we iico.βe a l~ st s r c β.- A 1 , . ~~~. . . .. Os β
a? g. . r,. βt.lcir s for answering such :c cece . n : ( U oβ -βccc:: D Y β Iii β: :t
β. Iefj r .i t I con. a: afl algor.I βtbi cc c it 1; β:β~:y I. r βs]uce . βuc βh a 15 s β. . I r c . - β~~.
Cβ orc :.slde r the set 1; of I c i tegc.βr : su b , t h a t ri is no: - I - c : - - - . β ci β
βnd only 51 β colgor i o hoc: A0 answer: β n o β ( βr d -cc rioβ - -o ,:w --s- β .~~ . all ) cc
I nil ut c c . βjβherr the cue:t I -o n ββ - n at, d enser:: . β.β 1~ , c .1 - :sc β.oc,cc d : -
since each algcr iths. in the 1,. . 1 A . , .1, . , ;.~., ... 1~..β . s cc β~s a n-n β c~~
-uo: w ec - βXi ItO . .. e βa ne ini ut ( A , is wr - rig n in : ut n ) ari d by β.Thurc i . β sc
t i c : s 1 1 cc β - , .n β. a l n r : all s- s::i tie - β... g - rI t his:, lulL cc used t i r e
s-none I n c a to ,ch . . w th e undecid.abcbl ]ty - 5 β the hall . . ng rr hβle::c 1β βr Turing
:.. :i- ,c c i i ne s .
Aimu .laOI β. r r is a cr eth d for tu rn ing one ~ r oble ccc .:r o r - c i β-cc:- :olv i ng
met -h od m t. β.- cur Slier. dn rc e we have eri e undecidable or wi crc F1 , we can
Ircβ~e another problem 1 2 undecidable by showing I βhat . i i β 1,., h ccs an
algorithm then this algorithm can be used to solve F1 . To cr cc . βcccs ]J:b
this we provide an algorithm which c βcrcvβrts an innc: tance - iiβ ~ r ~- β c
into one or more instances of pr oblem ~2
thus reducing I~~ to II
(or transforming F1
into Iβ,,, ). Similarly, to show that t w o -I βcf i r i i t i βcc:
of an algorithm are equivalent, we chc .,w how to simulate an algor ith.rco
according to one definition by an c.cl~~βr thm acc.βrci β ng β.. β t h e l i ner -
on β S nitiori .
11
I
- ~~~~~~~~~ ~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~ β,,__,_ ~~~~~~~ . . . β_, . -.--- β -β β-β-β
β. . _________________
.~~~~~~~~~~~~~~~~~~~
-- -~~~~~~~~
β~~~~~ ~~~~~~~~~
β -~~~~~~~~~~~
-β -~~~~-~~~~~~~
y ,--β -~~~~
The a cβ . bo~ !c β βcc l of ,.Ie diciβ cS i β i cir ~ - c c digItal β.o- -
.
rcβ.:: ut tn r s : rc :ade u1:: ] ble
ccc β .- c c .: leccct .,c ,t:rt ., cc c ar d β.cc- :ocut . i on of o . cβ .tl.I cated .ad ..βrI lii,β:::, ari the
- c β . β ; . 00 :β e β- c~ ut ability LβeClL βcle a c:.cβ.to βor ct β more than rnatlrerciatical
I c r0 c c β β:1 , . i: Itβ ..- O - iβ , this the. nβ:, ignores .~uec: .1 .rc: - ~O I c : - 0Uβ ..:β . : :i:eβ, which
tj:c,iSc it: r. , cc- r t o i .soβr , t i c s,β nor:β.β, ,j:ILβlt itβ, t act? cc. Ccli c β,β β. iβ bloc:.:
wlsici: -ftβS - us],β.β icrn β~β;β aig. βr.S tluc .o. :eβ.β Ot. β cmv- c c . . - β - .1 cc, t ciβ βI Iβ β . . . i c r
I r β . : : - alice , C βc::l ocrβ the β- β0:1:β .~ 1 cc c ie s- β t : , ; eβ β, g lve rc :β. - .- β - . - , :1 ri o in it
a c- .ac-,Lc-:uoβ, c cc:. :: or of volβ β.-~~~ , , cc Ow - : . c o t : . β . .βI rot . cc grc β.s. i c el th r,
coβertlo- .β: h oβ.: cnil β,β 0 cut - cot: cci β β.β:rt..I cβ βc, cur cl:: β0:0 .1 c~i β t 1 β j . e -Sigurltirc ::
a- this cβ 0 Led : βOXi S: β, _ - . b ;β.ββββββr n. onβ.β In:β: β,β e t d..I βcc β- , β βIβ β:- . : a cβd- :t- β.octi:J.lβ,β
β βccO - β!β ~le r .,0 hm : - Iβ 0 : 5 :0 i lecco.
I cc : Ic. . ce, S ~~~. 1_.i. cu:c, s-a t- : tb~, 1:1. 1 0 c.ciIoe of th is ho er: c:.e:l cm.
Iβa:le ~~~~~~ ~. β bβcato, :β, so cc g cic. -.c: co ~~~ sC~~l o . . c-olcl c β,β::rl a,.
S βci i : . CLβ β 1 :111β; sto w β that. curl. t aro β β β.c t e r c t . .eod lrco ,; less acid lo.
.i rcrn β,r :arst, a: or rβl..βr .. sl oe incrββ.so.- -:i ,.n large :r. Sol eri.β tsr β: cz:yβccci t~ h o c
- of t hc .β I m t cj s h d -cc , I . natβos the cc. β . c , c t c o : I . .c. β oβ.:i. r. The t:.rble
colsβ.: sh wc β,:sc β.t runnIng Iocβ.e grow. cxβs 1 :.: ,. -,β.β:].y it β β-ir e time bound is
- -cc - r: β .:r r t ia i . Table β .2 β:. :. cc .rstecc thβ: β- . c ci-: .1β .c~cc i,:c: of : βto:: .. - . I.β,β c,rs .βJ e
i. cs a -~ci βsen βci :. ant -of β lot - . c , . c r e - β., I:o - liiβ: nor wlt βf t l c : r c (~~β~β t S r t . β creed
of tb -c maclurid by a large U- β.,sh rβ : . ββ: β . 1 ub ,-tβur l.,iallβ, icc βlu β , c the
Sis.c0 05 β ; r u .crco :alvacβlβ our le : : I - I c e β t h . . unci gr owsβ:: .. re slowly t i c ou c
r. β: nt ;al .
Tcclclu , 3.5, β βi, c 3.2 - ugge. I a rico β : c,r βcit ct,r. tβ cb. . ,i , , t , bet ween β . all g r -
(those with worr:tβca:cc tβ,i riO ic U.I i d , β i β - v c c β. . . β .i, in the :. . - t; βU t i r e i co~ us
cccβid bad algorithms. c d,nono: I β. - β 5 ] wcic no ar- ci t ly t h e Uβ oct t . , c : re::: this
12
L .~~ ~~~~. ~~~~~~~~ ..~~~~~~1 β β
β β~~~β ,~~~~~~~~~~~~
β β
distinctio n . I :hali cal 0 a dec . olabie rile:β, t c - : t . β t β.5 1 β U i β c ia : a
cr1 al β t i me ad go ri t icβ ,: a,nd In t racta ble .β t l i e c βw . l cc . Th ie - i i , I c t j , ri
between t r a c tah l - .c and intractab iβ: ~β rob I e:ccs i s ic s lo : - : , .c- . - : :L Di β 5 c , , r ac - Ic I t i e
β βt o , since any of the common ly used o uch 1 r,β β del : cru c bce s~~~u β a t - βd by
any βthor with ..n? y a i β .lyn βccc c. βh cii loss ins run:: ing I - β . . β’ A: jab 1β:.β 1 aria 2
.chow , it 1: ci f r βeasible t . execute cxc oncen β, I a lβti :cc e alg-;ritr.β:: βci largβ.β
r β.clc : - c ns . 5-Icc: β , β c m c c c , o l c c : ch . na? n- . β.t : lern s-: are β :t:hly : lvcib I: in exj r c βn c c t i a l
t . imc β by exh au stive l.y cinecckl . cig ca:β.β; . but solving c ccii rβ : lecr,: I c c - , 1 al
I:β . - :- βecnβ.c t β nβ - rβ: :βcir c ir gr ea β -a β In c ~~r t . ~,5 :s t In. -~~r go i cβ.. β .c : rI S~
have time bounds wi le ir are :l~~,orcd a. - ofβ small d egrβ-c: ( β. (n ) c,r beβ, C I β ) .
It i s a cna ,? or ta sk β5 β 0 βcc .: tβ:xity t h~βn β,~ t - cc identify β~,c ic hi ccci β c.cra . !β c . .tβc β ,.
are tractabl β: and which cciβ; intractable.
Hartmani s. Lewi s, and rtcarns tack the U r:4. , t - β:c S war erd,~ hi I i rrg
natural intractable ~roLβ1ems (Hartmani c, Lewis, a n t β- t e a r c c s [1l~β. 5]~
1-Iartmanis and Stearns [10 ,5]). By dia~ . -n a i I c -oi cciβ over all alg βri 0.1cc-β,: wi t - h r
a given sj aβ.ce bound S1(n) , t h e y were able to Oh c . n β.j r βt lecr , : - s c -ed
in space 32(n)
but not in :jaice 21(n) , for any sc ace 5β βusc:c βC
1(n) ce:.c
S2(n) satislying lim iniβ U 1( n β 2 0 ( n) = 0 ar: : :r :β- βw I i r β .ciβ r.eo βi s .n . c β d
constraints. They proved a :Jcc:ilar but sβ z-rowhβ,I wβaI- βrβ rβ c - : β i l t c r
t ime complexity. These results ci mβj ly in art Seaβ! cu- t ha t there are cβ 01cr.:
salvable in exponential space but not in polynomial :~ acβ.c , and r. βc β.lc βc β :
solvable in ext onential time but not .I ni ~c1yβn - r o S a β. t β : cc - β .
Ofrifortunately, the intractable rc blems jr β.cluce . l by di β -,~c - :cc,li cβ crl .1 cc
are not natural ones. Meyer and Stcβckrcc- βyiβr {1β72J : r- ,βv-~β.b h i r β
intractability of a natural jr βS βl en.. They chow--S that I - ic β o r - 5 ] β ; β . 5 β
determining whether two regular ex]βre:c.c I on: with :qurir - I ng β.c - c l β . b e I lie
13
L..~~ _1~~~~~ T~~~~~~~~~~~ -, ,~ - ~~~~~~~~~~~~~~~~~~~~~~~~~~ adT~~ββc-_ β- _~~~~~~
, ~~~~~~~~~~~~~~~~~~~~~~
- β- -
s ame sot lβe-ga lβ cs c O orcc βcctc . al ~βiace (arid cer i se 000 cβnenti-.d t l c β .o)
for its : -luti err . A regular β.acpresccion is a U o~~c.β~ .c,- -o r:: I ruot~β.,i c β ro β.roc
the cyβrcbcols , , 0 , 1 , U , β’ , β ( , ) acc;rding to the fβollo βwlcβ.g
rules. Each such U - rmul a don βto: a cot . of s t r l r c g . of :c- β.iβ . - - - cool n r c , .
~.1 3 is a regβrl-ci- expression denoting the set [U )
1 is a regular cxc r -os -o I-c n denc .:-ad. ng the set [1)
~β, i s a regβu,lar cxc r e sc iorc d,cc n β.: t I n g the set win . cc s I n g].β.β elemen t
is the empty o tr .1: , c .
5.2 If A and B are regular cxβ: 005:1 ri : denoting ceβt co L ( A ~ and
L(B) , rc:ncectlaβ :ly, ther:
( A U B ) is a rβc-gui.cuβ exβrres:ol on denoting the set L(A) L~L ( B )
(A.Bβ is a regular exβfressiccc n denoting the set
jxy xc L (Aβ and y β¬ L(Bβc βj
A β i: a rio gui. - β,r β:cβcβcres :lβccn den ting the c β -l, cβ .ctcsl tlcc g
of the β o~~ ty string and all s t rI ngs U -- rccceo by o: βr :c β .tβ;~icoch I
-c cc r m ore .t r l n c c o . in L(A)
and - :t -~oc~~~e~- . r βrdd, βcci cβ.si addit .i- , r iotS rule :
3.~ I I β A is a regular c :βm r -..βs s ? on , then A2 is a regular cx: r- sad rr
denoting th e c arce set as ( A .A βl
To r β , β - that t i n - c c 1βoivalence r ~~~~~ for tw- . cueS: cxc n β s: I c . . I:
.cc tr ac t 101cc , β.I~β,- - βr and U t oacc:βoycc r used s irr,u,l crt . i . β .n . Th1 :~i - tβ .βI . -; : a
~ 0 Iyrio β.rcci al β f cc , - .β :.r,1,~-c r l h, !β.: β- whi βIβ.β. , β c - - c a Turing maclu c , - . an β. : , c s β . βc ud
an exj o-nent l βd . ::; cieβ¬c cβ. . ,uj βc. c , w β : 1 , conct ruc r . a regu.. - β..r ox-cr c : : ? βn
1~4
β~~~ - ~~~~~~~~~- β
~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~ ~~~~~~ β’
-~~~~~~~~~
- ..~~~~~
β~~~ β , .
~~~-β ~~~~ββ--~~~~e=~~ .-~~~~,
, -
βc c β: β cj t i i c ~~ t lrc c . : β i - i f . .- r t . S on of the T~~βing - . β o β h d t c β : ci t c a l a - β - cs : βof
β c βSu: β , s r β - :s? n i s oust : I-h at i t b , βn . - I L , the :- c c .c s β b a:
S c β β cc c l ciiβ,β i f the Turing c cc achnl c -:e -.5 - β o rs βt acccr I I - i β .β - I r r : r r l l.β β β i i I n t he
Β°i ace S. - nr c βS . 5~ loll wv that the β -~u.I vat -c o-c r .oi,.lβ~c - β cc
- β - c c r c :s i, β .r . : w I t h : cc cu at - i ng i, : β.~: i β . c β .r - : ( t o wi th: i c c r β- β. t yc . o l d ic c c ~ .i, -o ,s
as .j~ 1, 0 : 5 β c : β 1u β c t i - . n ccs:w β:r abi βc in -cxc β t ac t . : I βd c c mcc c by a 1 :cβi r:g :c , .β . βcl r ]n β. .
, ; ! r,cβe the !β . : - c ; β~~. h . . Low - I c , , β:,c- β .,rr , . :βo: β,ilt β! β-:~ l I β . tL -Γ¬ t ~~~~~~~~~
- - c-c i s I . : cβrh ,!o β ] i can he c β β.ltβ- : . .h . r c s .T acβ: . β ~ - Ud n t - s c . ~~; ,ββ ,
~~ , β
. cu I β:βd- :ci β β - - o r -t β l -cico f:.r iβ- βgβ,L. cir cxc c- e:s. I .n: c :u:I r β~ :β.r l r - ,β β β- c : β ; : β . . r c t l c βJ.
in , IS , -; β c c l S olve βrear:, c- β;β r d c - c- i- s c uob - . rβ -.- : c iβ , h - -;- ; β -βc- c:
β :C - β,~~ : β β - , :. i~~~~~it [ l; β. O - c e β , - - - : tha I 5 : β _ β - . β I : β.cβ β:rβ,β erc β β - β. - ::I β t . c.
β rig t ho ~ . .βaβo.β β.βcd-.βciee r i - blori I c r r e , a l a r Cs , : :β - , ,,. : β β 1] ru - c o i l s - ; :
c . r. c 1o 01 1 0 O c - β . ,.t . cβ.β:c, -co βos arc~~ , I ccβ [ 1 β ] . d , β~J β ;O β , β ,,
:βc c c t i β :. I : , βs ic β I β c : t - - c h e r :iβooβ.r . l r ig the c -c looIv β .i β;cc- .β β β i :1β: c r
cβ -:~oc c βL oc.r ox: nc:: ci: has a I : . d r _ , . β ! , d r , : tβ . β. aaβ . : :a.c e : βs rβ, s . CI. β i i β . cc βcici e rr : Sc,
5,-~~β h . ] r - c i cc i f . β β ,: :O s c t he ββa. ,! - c - I Iβ; i t β a U: :β : . βc S c fβ . - 5 , c β β S
- β.,cβ I t β !cccne t i :β ( t h e the -r β; βi f β :0 .11 β t r o t cru c - .5. β-r : w I t h noβ 1st culy β: β r β β. β cc, . ctβi
c β β :β,cs ros; β :0 0 c c- . U r : ~re : -o, βit ive yt βn: t β~ :β β β’ - βa LLβ,t : - r : .
βOiβ.i 5.~β:yβ: r [1 ~~β - sO w- . c I t h r i t the β, c r S :r - I β β - . t βccr As . βa d l a j c cβ β-: . . cc-
- β , β c :c r i t β ci β. : co s c- . β.~β a~ i- , Sg Icro . . c~ii iβHcLβc h , [ 1 - 5 ]
--wβ:βi that t e s t ing t h s - .o circularity β .f at . t r iL : t c gr: _ cs,..cβ: (a - cc b
β ri s i n g in ; c β gc-ar:ming langu age semant ics ) requ .l r - - c c x c -:: ~ β S I β d f
The idea I n all th:~β:β: proofs 1 5 the sane : one .0 βw: I. w I i ccf fi c β β β . β1~;
c~- r iv erS any co c c I u tat I β . I i with a jart i cal - cr sj ace r I β β . x ~1~β -
a,cs Sn. 4 anr - - v U tb .- gI ven c c β S I . . and β c u e ~ β β aJ.: S t S c β l0 : cβ β c c c i , . 1, -c: β :.
5β .t , β;cc rr β,;: r e c u it s to assert the cx? : t β c βrcc β βo β . 1β an ~n : l r , - t c β t - ~ l,β! e r iβ :5- β ! c: a-cl_ I in
15
___ - β.
.β-β~~~~~~~ -β ~~~~~~~~~~~~~
I
. : β L ] ~~β.~1o:: ~~~~~~ o βu
:L~, β
i: β
~~~~~~~~
β
:~~~~~~
β
.: I.; ,i : r :β .cβ rs iβ; i r :1β: β : , cβoiβ :c i,
0 Ii β: :βo cid c: ,ocoβ ,β.c βr,caOβle :~ βt. β β I cβ . . -:~~βcβ a I occluded Icc Iβ ], β β l ist β o~β rc rc c-c ,
β β . β. : β c r o c - , β c βo i; : i β . s β lc-r c β . c: . β i β i c e c o β i β ;ob le: β : i c c , ,βsβ~ L i l t ,β c β t lcβ~;i ng r o β ci β;iβl β:. If
coo lβ β. a s o -Ole::: is S crc rc cd βci a 1:0 :β cl . β - . :U e:5 .l :s , β -c, : t , ,e ci ci,β. β : β
lice-c c t i r - ;r o - Is a iyico:cU c ol.β ;o c,g t c r ~ ~β, - c β cii β t ice ci . , β.cβ c c . β cc-l β : , , . β - -
-
su c c o s e cc~ cβc:β : i , o β o . e c i : - .- β , : cβo ,l, :β β.u :::~Li-: :01 5 1 β , β.::: - β .. f , l l sβ. : C.
a givβ ;c : c~c- β : :: U cu:c cc i c : a c tabie c - I U 0. o-~ciβ. iceoβ! β ic fi :~. . c c: , c,β cc
I_ ye: . -c c : . eric. o ceco - β β b y β;c-oicI I .l t I : c e sβS- - U , ,.β, cot r,t βi .b c i : c - ,-:i c c - ~
To c cj i cβ, c : βe ,I , c , β I βcc 05 β :o lyβc: r c .l_- J β . ;β,-:: β.~ti: β!~~β β) β. c- β -.; β cc I x- .j βccc
cc c β c . β - : β - l β . : c : , βc , β i _ ,:c:- : ,:. c β, :sβ .cc β c β - t - - o β c v l r u l _ t i ,: : β β , , c i : I r c ~β - βcoy. -β.β. vac u ia:
:.L c c , . : 502. I c c : 10.: c~~u βe0: ; : ,I c c . ro o d -c- . .a β β -βc . β, β.. i. cd:β β, β U . do r o t c - c t . I i : β β
c~,aci:.: c .: accccc : , . β a rdβ:;:, ,l c β , c c .d, :β c.S . β i - β β -:βc.I .βI: s,::.c , , , . : c o- ; :ccβ β
cdc I~~h ; cau ccβ5 t h s - .β : c β.cc Β° β . Isne o β β.:v . - c β . f e c ; ,.d : 5 , -s e n β ye: β . ii β. ,s , :βo β . t i , - f l o e
( β c i β c~, . c c e l c β β : β.rir ei iv; β~~,e c-c βcc:,β . ,βle U , :c.c ce β:t cur . l : , : c r i sc l i e - f i β . I : c s u ,
arβ. .:arct of β ti:c ,c ( -c β se asβ.β ci, β, - β .c 1;,β arc acce: t l i sg a corn: u f :βsβ. - . II1L foui c β,ci:, :β
nc c β U - c t β c- al c .t I β . β :d.g :β β c . :sc: . ls.vβ.oβ c . i re β - cβΒ±cβ,uc: stait e set ccβ i βle:r . I c .
c u d f l β .β;t c βir s t, diβ:, : :r sub; . e t . βI β k a β . βr col ..β::. - c-c c , clues βS : all
β~ I r , .βt β β i β - cc β;β .- c β β I:: -,; , c βu r c s ,S ,~~Lcoc c,cc 1 , . β.βciceo c i t ti ,~ i - .-; β ofβ f c c ; \βert ,1 cβo :
βciβ-: - , t β - β -c: . Id , y .s β.~c 1. β. . - t!rβc clot:: β.cI β ~, β c : β : . - nβclclec :ic :~~.
:β .r, β r~~~cβ:i:β~,I, cally in , 1;βco β cβ ..I cd. 1 , 1 c c β and let ~~y c ,I~;β, , )t~ Li .c c β ! , , , so: β
l _ r c - : β c β c . i r blocβ : : ,J a oib β ,. - β : . r . β : - - c , β t β n i s t I c : a l _ y S r i j lyri c - l c d . t ββ:c . The
. rc a- w β :i I . . - βcc , βwβ r is . β 10β β f i r e r β . r r : , tori βc Ij c r r iβ;β - we l ch - β.rc in
- S it , cc .β β5 : , - .
~~~~~~ [IL ,~ βl ] β0 - .d I .e-cot T~D ~c c c t c l β o :n c :ββ. βrta β.n ββ li -c- β . : - ~~~~~ββ Hβ c c : . .
β. βt ~ , ] . - c - 5 71 β) - β.: c l - c t. β I - c l β - - , . A Hβ l ie- c :β. I is βf~,) _ -β β~~
iot c 1:-
1β
- ._~~~. . . - . β β~
__ . 1 __________________ _______
ppp_ -, .-β--- - . -β- - - β -β - -
~- . - β- β -.--β--β _.β - β.~~~~~~~~~~~~ -_---- β--~~- - _ - ,
β
it ;catj .:βIcs faβ - : nc- c c i i I c βs:
:Β° .β~ I is in ~~~~~~~~~
3.5 is β ~ , is I: ~~ t l c c o t l Q, is reducible o c S i c c i β .11cc βdo d. S I::, β β .
To cay tha t Q is reducible cc. . F in p - - lytc :cac β.. l i t β - cβ , ec cr : c hat
thers is :-~ (de t β β:~- ,.l β .1 c t i c ) iβ-o lIcβcc , β.rcd. βJ . - O I . ::oc β cjc~.~r , ~ 5 s β . β,β~ , i βI ,, g β cβ ::, βcc
In st ance of β 5 0 b βSβ::β, , ,β~ , wIl l :β..rr vert it ci cc l, -OO i I cs :t βci β .β - c : cc h β I c c : , I
cash: that t h e cc :wc βr βf-c t1c~β , n stance βU Q, is β β yc β β β β . 1β βc rab βcclβ,β i l 1,1cc
βurs a βer fe 10β .β i c c c t cn r s ce of F is βye: β . if Q ,1: c- . cβ s . : hole c: :- in
t . lo c s β βu c β .t F ha: a i cll,- βc :β. cβsl. a lβt i : c .e cal β n I l . β , t i , - β : , ccβ. U βesβ
Th:s: if ar βsβ; β ~~ β β.0 c , : ic-to : iβ : 1cm hoβ.: a cβlvnoc:βc,i - d β t ,i:oe a lig nβ? c i , . 2 = cβ
Cookβ s coc ci_ I cs iβ. , alt w i : t β.c , is cw that βdie ::co .1 - 1: aOl I,1tβ; 1 1 β - c c β . ? βcβ
~ re~ ,c s it i βr : . βd sccdcβulu : .15 β 2 βc c βrc; lote . 51ccβ : aI I : : βi β i β l ici ty I i β : bolero is 0
oi ,c t-encβr ine a~r c β f i cβ:z β a gi-β.β:cc logical lβ - rocui. -c 15 t ru e βcr c lca.t ccc
c~s:βS gnmotct βofβ c_h,: value. βtrue ββ β β,zc . S βt alcβ β t o t S r β - βcβ acβi βIcβS.-;:. β1 i~ ea sy
to - ic. ;βw tic if tb c r c - S lo:c. sat.i :Iβit .βs 3. 0 . Cook c i β βsβ. S ~~~. 5 5d-~β gIvi ng a
:].βfli βoiaj.β t .1 :- d.g r ,i th cc βc O βer sβen:t ,ru β.:S,I n e , f r β.ccc , cc g I v e : . c i . c , β :β.-iβ:r:βoi βnist I s β
Turing machine , a given irc~ at , ar cS a given .-lyt c c:,ial t , ir - ,e h - c c :, a iβogIccd
1 β - .r~-c sLa r ush f ina l lice c -;rcc,-,ili. a is :at i sciβdl_ c ,b le if and c:l~; a: β β i nc Ucsaβi c :g,
ccc ,cichi.ne ac: cc 1.: f loe in~ Ut w i th ,i n the ti: - .o t ;- . βucisi .
If erie iu-c - -w: a :cngle : r O l l _ c c βs, F to be T βY β cβ :c:β , : 1β: β, β , βci - β cocc i c roci-β. . -
another i r - a - I β c m Q 702 - c o rn~ lete by sh owing that Q, is in ~c Y arc , t thoU
isβ reducible icr io1~m omi al t ime to Q ro:-erty βs .5 then c β,ccllowc :βr βccc the
transitivity - 5 β polynomial-time reducibili ty. barβs [ 1β β~βsβ ) usβ., βcl t h i s
idea Icc; exh,ibil a number of natural ~~j -c-crc βs] let cβ c r - b c ] . , βT : : .
continued th is: w . βrk , and the number s.f knowr ~~~~~~ -cnc:c~ lie-I .e Fr βcS i β-c β ,: 1:
- - β
17
~Iβ1 c . ~~~~~~~~~~~~~~~~~~~ - -~~~ J
- .,.~~~~ ~~~~~~~~~---w~~~~~~~
ii x I : t i n , }cw ,d ~β :βr: (s e e cβ~r i c r c t β ,ccce- fe r n . I t c s i , ac -cd :lr ar: , .i cc [ 1 1 7 β J ;
. cicβcβ; . U . i r r , c - β . c c . cu r d t cc ciβ , ! β [ 15β I ; Ocir sβ;, : bL β : :- β :i , cc: : J aiβ~~arc [l~~cβ~~i :
ban- c [ 1 . 1 5 ] ; scaimi [10 1β!]; Sethi [l β:t β I ; and UlLr cccr [ 1 - i d o l ) . In a d u i t l s βr r
to S i ce oa t is l βic ,cOI l .ity r . Lie::: and t . h i _ β - β , βsβ: I c :. c ts: - cc :cble set r i _ i -s , thc
c β β βL βs cl cc c rotcj , β.:rcc sβ- c_ c β s - Iβ: β - β c - c c LoS e .
?Iut:gcβ coh ~sβ.oco:o~~ ic ic: c c (Ccc ~k [do 5 1 J ) . Giβ .βer i o~~ β c c c 0 : a r r . i U , is
1 iscβo, : - o i c .I c to a sub g r aj b i 05 β Ii , βS
: 0 , 1: c i βO ral ( dcccv [ lo72D . O;β,β oc , cc o:βa~5 . β5 , c c c i t s ciβ β~ βt icc : βc c
c lor d wI th 0. col cc: so that cr ta :sccta,- e c β r β U β.β β . βrβ t ,I c - - β: i β . r iβ.βsβ 0 c c β ;
5 . 0 0 0 1β:β TicS : : i-J.βIem ir . β/? 2 β c -c:~i l e t e even I f β Sc = ~~- c - c c - ,: C is
d arner (Uarsβl ; , 3 - : : , , c , , end Ut c Sccβo β β,.jei~ [ l itβ . ] ) , i , : c s c ! β , : u : it 1. 1.1 cc:
c r c::, dc c cl β c c i . b iake :r β cc r ococo β ci β the U sir c co : 1, :nβ cuc r , β e . :Oc .r1 β e (Ac el
ii:cicc:r: [1 7] t ir as f - Sc o r e 5 : a , -~~,βri β .1 c a β l I :cre cj gccrl tOo ;. cc : βr cony
lcincccβ cyβac :i wol ti : U cr cccl -r .: .
iicuado ton cyβ - βr β : ( d o n [lβj d 2 ] ) . - d c i β -.- :: c.i gr c~~bc C dot : it - c r i t : c i c r :r cycle
ci , , , c l β . asses c _ rd ue~, β:cβ ,or a vert -,;x .-x a c t l iy - β r -β.c β. β0 flnd s ~ r - -b-isβ:: 1: a
cc c c l al ca:e of the traβ,β.clling :uJ βocc :cai r i r - βbo le- cc , ( c e - s - c c β s , βon ~~ . i t
i: ~β 2 β c - rca i-.:tsβ β.βvβ;n i i β U i s c lunar (Urc r eI β .β, , β lβs , c c , arcsβs Sa c - i a n [1 T β
β i: , β ci β~βIarc [ 1 7 1 1) . -β;:ivβ β:i -βo cβJ of r cc c c , c,j , β :β c fl~ .ri cci-β.,
:u:. c o c βc5 β , 5
1β.e: cccβ , β β 5,0 - cei oS β S i n - β c cc i - .: - , : - c β : ciLco c t . β . -x: β .ct -S.β,β
~ . sarβsar cuh gra~ h (Liu acid S. βldocr:icbr ββr [1 ,β r ] ) . β βliciβ :cc cc βn c r] Si -.5 . s ic :
i t Con t -as cc Os 0 l an -c c β 5 1 5 g1β c~ c Sr aβ tO at l ois t k βccdgβ. βc?
A major ciT c- ri c r i , β βt β βo::,~ lc:-c .β.ty t .b1 , :c β. r β.β 1., to c o t β rβ : β n β. β,~i , - s i c β : β.β β 9
A r i o t βsr o ci, -c c : reach t β.β. thi : r bloc: : cc - c c I cc b e 1 o t cβy c c . i cig si - cd β c , β., - β . a
β:xlio S at a rob leco , in 719 but not in 9 . :O W β , - cc , r β β S c β: β. β. ~β c ,, y : β . c cSc et β .
0111, and So1~~iay [ l β f ( ,β . ] :~~~g β . β :S : 5 5 c c β . β diβ β.β. β c c β ri . , :at i c c is ic- i S r i 1
_____ I ____ β~~~~~~~ β ~~~~~~~~~~~~~~~~~~~~~~
- β-- - -v-flβ
re s βlβ.βicng t b c , β 9 = β 19 ? ~u - : ti - c r . c .~c_β , , c c without a iroof that 2 = β 1 2,
it is s: t csll :βru~I c_ f uJ , t o add new natural problems to the list of
~~~~~~ -conn ie-I c- one :; the large amount of time spent by bright pecβFle
fruitless:ly ccarβc}iclng :βcr ~ulyrco:sic1 :d.-time algorithms for βS7p -c βccr :~,lete
c roblerr ss is strong e~~ dence that tbc~ 7~p -com-c, lete o rcblec :c : caβe in fac t
intractable.
19
- ~,β - - - -~~~~~ -~~~~~~~~~~~~~~~~~~~~~~ I ~~~~~~~~~~~ β β -_-
~~---------- β~~~~ :.,c _ ,__~:. ~~~~~~~~~~~~~~~~~~~~~~~~~~ ., β β . β
β . ._,~
,.- β ββ β
0 . Techc . i sβ.β ,: :βc r S β.5 AJ.g βn ithcc c s.
Although many Scar er tacit combinat ;ni al i rob l βsrc c: seem to be I n t ra c t a b l e ,
many others haβsβ e good alg-snith.c βrβ.s. A :rccall n c_so,heiβ o:β d a t a ciarac ; u !at βI on
t ectnr r ?. , cue sc lβ cc rm the bas S : c βor these a igc c r c t ~c-β.c. This section β;xar β .i . nc- :
these techn i que:. w 5 r i c c c i β.rc :ut l i o s - β i n Table 0 . , . .
[ Table 0 .1
data St ru :t, c. i r β : sc .
Any al g or I t h m (good r bad β re ~u: r - .c βone r β:, 1β : data . c r aster β; . -
r eco rβ .osβ:nt tOo β2? emsont s of t i r e r :βh iccci I be :o~ ciβ ,-β . c an-U tic-s i n : c ββ -~ t I - c ,
:sm,1 βotβsd duri ng the solution r ecess . A data structure 1 s a c.oco~ o: :tc-
ob ,βeet o- :m~ u r e A 31β eiecc ,o c;t s rc- . at e.cb I n :c ec i l βi e c n w e i c . A s o . c c i c β.- - o w~ t1,
tOβs data cctru βctur βc I: a c βcβt of βler βatβS :ns U r βc : :c c lc c_ il ~β.t In g 15 ~ ele:: , c- n t c .
c β .oc - a ge.-J iso : L c c n o n t a t . . n . 1 β a g i c - o c β . data s t r s - . β t β c r o ani βt c -:r,cc1, I n:
is kn own , . r,-; can regard the crat - ens as rj ccclt i β,βe: aβ - β : , mo . o:c snt lon g
any a~g β r i t cc c : . wli , ch use s β the -lat e :t r s co : c r y . The e S βf β l c l β . c c c β ,β :1 β S h e
c β .tg cβ I t βcciro w l . 1 dβciβend a la rg e ext cc c . cc: cc tbs βo βl c β c β s i - : : β β , βrctc , t i r, oc β l Os - ,
uβi ccl βorl y ing oc et a s t r - :- - O sr β - .
Thβ:r β β Cr-: t w data sc r-si. - t : . c rβo . 1: ci ais le -h al - tSl β:r. c,r β : ba , c c l :
o . I cc l c β - .c t r c c - l , i r β,:, .. ,β c as-c-cc?,β ~?: a :, βS l e ct I : c c βof :1 ragβ.
s e t . : cc ::.c - - c c β s .β ci. - c e - i t v c -t y . βl~ c c βora L , r i o sir- .- a: c βc aβs oβoc a c _ h an
βcr r cc~β . g ,v c r , S b β β :i rnt, βor of a c βs βrage :~~- ,,, r. - β can e j l i c - .- βc- , S c e a c: β . 0:0
ir , t n c c _ c r β . -β ,β c c i i (.i ββ , β. r β, rig t ,io β β ,crn β : :s S cβ s Jc βO cc re tr Ieve the - β ,ic βcβ βr t
β;a, u- : r β β.rc , S sn aβ r β .~ - - cell . The- a βa rβy β cc β a ran ,s β acce:, β βoa c h l ne cccl
c. :β. - 1 , g c S al c :β,βs cj t β sc β , .1: re -, βo r - β.β, . ne- can β. 4 , : ci rcβa, ,β , t~ cβ :: rβ-;,
cc - c t - r , , rr ca β cβ .cβ- β , . t n , . r , , ass βS , :. .i i. , d c o , , r i c , a n aI ri,r r a β;, ( b r i c i t i : [ l β β i β .
20
_____________ I __~~ .I ,..,,T t T~~ -_-- -
~~~~ - - β - - ,
-β--,-- ,-ββ--- β - -
A I ic n l β : , ,βd ci :β, i β . C I or β -: β.- -n s i , β t ,: ~ S β c~ cci l l , β .β β ci c β 1 - 0 1 β , ii sβ β - 5β - c β s:
d l c i β l β.itJ 1:, : a :, il , cio, rβ , r β l t β β c s . β.r i , β i i a I L S: arc i d β cC β :βlβ.β l r c g ,ac~β . β1:10
c - i _ I l cc βs do: β s β - β. . β β βca l . i S β βcβ . : a r - :β c a P ci , . c β s β -
h. β : ::: :β. cnrl r . 0 :5 c : β β ~~~β β :. iia~ a β.~~~~~~~~β cc c o t s β . !β. 5 - i β β , . h β : : β , o β : c β . - β β.c
- n β s c , , : β r i r c t βo r , β β - β β β β c c . βfwo , cc β, r a t i , cn s c c β .re β s o~-~~l h S e . r c :~ 1 : 1 , 1
::,c β c : : t β cr β : 5inβ vctn ci L : β . I f l 5 s ! β t~ a resβ β.β i. ~~~ can e t b s e r :1 cβ - β.
cc I S ccβ i n ~1iβ - cc - - e l .cr β.β β.~~β.:β β . β .-β,β β- the :-ccr r β. βcc l m l βc β β c βs - c , coo i t c - :r c
11 β : β c - c- : - : . Pβi~~β:r-~ I iI ,u c 4 r , t β c s a l i n k e ,i cc , r u - t c. β- - β . 5.1: β Sea:
β β cc cβ.,,,; - - tβi rβ ββ~ β - t i β β: icc β s c : , β.β: C a : aS I s β tc - s I ciiβs ears I i c i ? a βS - - - 5 5 β β,β βsi - c d, β - -I β cc
cc , βral l , c f l . rcO -: ~~~~~~~~ i o c l S -j r - al ,~w , s c 5 cc l i n k e d :d cc : - c c . c c r β . - i cc β, β. c β .
so- , β- ; , t c cβ :β. β.β- .- , s - - S r i - c c _ c s β . ri ch t e s t ,i c . o U -r e r c a l 1β . Tiiβ~ cmβ. β.- β, - a
dckβ β .t rcβsercn~:- βc c .cci βci c i, n - r β S ri l c f l h l c β : S . β β iβ,i, βc β c l βe , βuc,l ,ββ - β- :β β
C ar d 00 nβccciarolβ , βa a. on βn r :c β, c c - βci li cnl βce β.i : 1 c c - c a r βs : .
[Figure ~. . 1 1
it is eci sβ,β t β. S c m l c β , c ~ nt an βrsΓ§ sβ and li r i l - c tn β.βl rust i s - - s s -cβ. t c , a i l
_ c t s β ss~~β d c β s - S iββ - β r ic - c i βa i lβs:qui rβ.β βcoc at c-u . m t i : - . L O r e β s , td βc~ β : c i c β- .. can : β ,
I :c; J len . βo βcct β i as ~o7cβ ectj ;βn: cf cβcxβr dj β.βsβ (see Fi~~cre .1) : t i cS . :cc :&ke - I I , S β
cc . ce :sing Ca - , β- ,β i n iar c 1βs βsiccβ¬β : : such as iβdblβI~~T 5 ah i β. β i β . A in it 3: .c : cc: ,
ccβs : j i c c i t l i s t - c r c-cs :ing c βa - i i i t y. f t c β . c β . o m n o to he imc .-::i1 Ic t.~ i r sn β i β β c c c β
β cc, e r r β ., , β β.: q - .i :nk β . ; :c 1 i βucture i n such a way that :1 c-al β- :00
mu se β .βnrc t, ani t I : m β.c , though I kn ow ut β n β s.c - c t r β thi sβ. In β .β:
Using array s and l inked structures , onc e can ircc~ I - β: c c , : comic: β ; ,t βU - β
la t a t ,r~ . β β cli β ,βS. . I shall corn s I der her e f . ve c l a s s e s ~βf S-h a s t r u c t u re s :
l i s t s : , un )rd βsred c c l s , - β rdβ :red sets , g ra ~β -1ncc , and S cβ - -
A li st is m~ ::qucc nd βe of β β - t o r r e n t s . The f i r d o e,l β s : : c β. β n : t ββs β a 1, t 1,
i t-s head ; S c β ,β i_ oct element is its tail . Sim~ its -~ β βc - ci t I - c i : β i i a li st
includ e scan n β.ng t h e 1,1st to retrieve its o1, β.βc :cβ,βnI : βI n β r i o βs β, i d , l . c g cc
-β .ββ- β - ~...~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - .β~
-
- - -~--- β ~~~~~~~~~~ β β~~~ - β β ~ - -~
--~
:c: the new head of the l ist (making the old head the c e c . -nd
β.- βs - om . :rs t) ; a~i c icng asic βrIsc , , .cro t as the- new t .a I 1, oi βcβIl t ct ing arr cl r e t r I ev in g
the inβo ad of a l is t , and delet ing and retriev ing the tai l of a l i s t .
Lis t s crc a-tsich .cc ty a :β c w u f these on erati .o- ns are i cc:: lb t , e ha β,β: ccc β.0
cramβ::. A ct ,β. βI-c is a 11:1. att ic a o l o l t i . c c, and siβ, β, e t L n al l. w:β.s -rh. y at β,he
: 0 , 0 , - I . A c c c_ c is a l is t a t t i c ad-U I 5 ? , cr a il s -a sS on ly aS t e e tad S -ir s S
ieI .ct β. , n aL s-a d on ly at tIn e head. A deccue (c ieu is t eβen d β,oo .4uβ: βse ) is a
l ist sin ~ ci β c , , ~ cc.iiti~ n . r dele t ion is βo:siβole at e i ther sol id. βone can
Icc: , y β:y βr rt :β. I -S lOe either as a circul ar array (adsir e:ses are :.-sc o st~:o
co c A-o le the s ize of the a r ray ) or a,: a s I n gly linked stcβoβ.:l β,:re ( i f c β c l e βs l O f l
c r - c c : the tal l is r r , o t n e c es s a r y ) . See Fi~~ re ii~~β~ In e i t h e r cc β .c β,c , all
ic-ora l s βno β.oxβ.:β.o~ t scanning rcc β.co ire ccc ist arβrt t ime . The ar -r - SIβ.β O c t ,r t o : e r s t a t t no
u ses no :5a, β cc :β , :r ot r I n g ~. . c I c r t e r c out r- .βA u . r c . tr i a t arc ar:. .- ur β.t chβ ot ccβ .~c--
equal t o h , cccmc, xlmarm c size :β the 1 1st be : ercc ,ancerct .y all. coOs - i to th e I t - β
[Figure 1.
Other imr .~rtan t 1.1st ,siceration : include concatenating t w . l i s t s
( c β,akiri g the head -of the second list the element Sβ ou lewt c~g the tai l~ of β
the c βi r :t ) , insert ing an element before βor after an element wβci β-se i βc a c_ i β :c
in the list is kn ., -wn , and del e t ing an eJ.omoc ent wO β se locat t -n in the lisβt
is kn wn . The :- β .o~ e ra to c-n c require a linked structure I c r t β ,c:β ir et β f βlc? β:n :
l ccc~ i so -cβ:ntati βn. A singly linked structure is sufficient lβ.-r c c l ; β s~~βo c c - i β s L , nrr
and tβ ;r i n s er t i on af ter cjn , βt Sr β , r element . Insertion bel βcrs- another s-tem β , c ,βr β . 1
arc-I d e i ct i or rβ: j~u I r so a doub ly linked structure . See Figure , β’ 3 β’ An
alternate woi y t h -j.n d.t~ - S o l e S . , . r i is t - : r , :vide each ele βcc .s ,βcr t aol : β. a f l β ,~
which is set I, β t rueβ i f the element i s to be βI e l e - t β . βc . The e t β.,βcc .er s βt I:
or o,t e cp l s ic it y dc acted until the ci βs xt s -an t r i r ough the l i s t .
[Figur e 0 .3 ]
22
-
~~~ . .β
~~~ - .
- c - . -β
Ifhco list Β°i torcition: hardest I Im I lement are , n r s e r t . cl ii; ar c j lsc:c:ent
at the k-th o- c crc it io n in a list , retr elving the element at the Sc - th position
in a li st, -S r deleting the element at the k-th position in a l ic t . It is.
possible t: toonl-,-:msβnt these ,cj- eratlons t:: run in o(log n~ ti c- c, where r,
is the size ch β the I i : β. . by using AVL trees ( Knuth [l7β~~}) - r h-~β t rees
(Aho , FβIooβcr c βS , and βJlLβccan [i971s]), which are rathe r c.ocml? cc , i ed i , ?nksr -I
s t ructures. Recently ~,i iba: , tIc fr s- t ~ ;i β.t , Fla::, and Robert : []c T7βl ] iβr β .ve
:βom,cnd a Waβ; t scarsβ,β - ut these erc,ioi -m c irr cc( i .g k~ tic:ce.
An un-o rsterso-ci set is: a coh_c β.ction of di :t ,snct β-c lecccen l .β -atth flu iru . β.:sej,
relat l sn sh ii. Basic set cr -or ati on : are addi ng an element t a cot ,
deleting an element from a set , and t e s t i n g whether an e - l ecc .en m, is tsr a
set. βInc way to rβ~β: rβ.cs so:it a set is: ic~ a si;ngly linked l i s t . AAdit .an
requdres constant t imβ.β but testing and del β t i . n r oqu . i :βe O (n~ .?me , ai r - - c β -
ci is the size 01β the set . Altern atively, if the elem ent, β01β the set are
values which can be cocβ.-cβio ared and sorted , one can represent the s-ct by an
AVL tree or a i;~ β~ tree in such a way that all three or oratIons r-; ~uire
O(log n) time (Knuth [ 1073]; Aho, ic βmr c r . ,βft , arid tJllman {lβ .-tβO]).
Another way to represent a set iso by a bit vector (Alio, βS 1 s βr β c h. and
βsJl].man [l9β14]), which is an array with one storage cell U r e ach ~os sible
β.1 βmont . A storage cell has two possible values : β.rc: . I csd . cad i rn f hat
the s o t contains the element , and false , indicating thai it J.,-es ::βt. Al_i
t hree operations require constant time using this represen ic :nti-:cn. Bit vect~βr
representation is only feasible if the number of pso s-sible β.clements is .βmcs - ,i i .
If the number of possible elements is large, one can nthndc the beh avior
o f a bit vector by using a hash table (1~nuth [1973]). A hash table consi s t s β
of a moderately sized array and a hashing function which map s each rβossible
23
~~~~~~~ --~~~~~~~~~~
- 1β~~~ β L - .
- - L β . β ::: - : c l in to an c a l m , β ca stle β, :. If an , βlLe: ccsβ c i t is resent , IS i e β . lemerot
( β . βnβ ci I- :β . . intuβr to i t ) is st rc-ost at (or near ) the address :t sccl: βied by the
hashing function. .βi , rcce t a β .- or color β - : elements may hash to β the caine address ,
:onrc- cccβcharrt ,βm must be ~β-nβovided for resolving such c- ,ollci :i cns . Hash tables
saβe used extensive ly :ir o slβ-: :, i 11cr: , and c - any Ian ers: have been aβrittsβrβ. about
th so:cc (see F~iuth [1~~~ ], I-Ion-rI : [i S P ] ) . Wi th a h:i:iβi las le , additi urs,
cci - βt.i - c i , can .! testing rβ . :i~~~cβ -: 0(n) time in the w -rst c:scc but i β.iy
c;cnct βcot timsβ on the β .β.β β -o β ~~s
Act,iΒ±tt c -c r oi . set cc sβrcit i m s are βo: es-ui if two β.;βr score set: cxi st . These
include the cii i It ty to forc : a set whi ch is t h e urr ior . , .i ni, βor β , coO t I o , , βIβ
di c β c β ,βrs βnc e of two cot-:. For most re r -resentat i .Β°rr c url.i orr, ,lr βcl-orβ:esβl, i cc.
:5151 rich L βfer cr s u - cc re , βosire t ime ro~ cr0 I :rralL I ~β- the s- icc: at β tSr β: s loe: ;. I
s e ts . Hcnwe voo r , I: the β,snlβ.βerce of ~1,ccc s-βnts is scccβssiJ.l ensuβ.~h β . sic that a b I t
βcβ se.cto r can f i t. ci,rβ.tβ.β- a βca cβm c c cotccr a rdc and tI re cβcc :ccn uS. sβ r o : .β -::sβs S t
oβ-o .s t - r operatch - m c , tS ,β:r , mod m c , I r i t β . β r c s β . c t , I cc , and di:βt β .cre-r,ce - rcβcjuciiβe c . c r s t ccc i.
tircssβ .
An ordered set is a s-ollect on of clement:, eaclo wi th an ass o ,I β.ted
rcβ.Lβcner.rc value . βiv- - ion : β.βrtan t cc oral . - as sri β.-rβ.iertcβcs set: are cccβs ., cc i ; t i s β ...
βt , emβocts in i r . βr β.:ac :ln ~,β cr- cr and se-i - -cd c m i ; the element wi th 1-s-tin lscβ.rgβ.-βct
β,βh , β.ic. A va r i c t - βi S β S . woa~βs exist to sor t n d o m e -n t : in O(n log n) t scc : β
(~~~uth [197v. ] ) ; t o β b ir β. :a~ β cβ.:β.c~ arcc s m : arc the -only a: er os . I -c-m s used to
marnipu.lat- - the va lor -cs- Si r -r n β . I (n log n) time i - c required in b .,, -th the averu~β.β
and the w orc t , case to sort (Knuth [1 7 β S ] ) . Selecting th e 1-c- I-i- largest
element r equ βi ru : - ( n ) time (Blβt,mi, β. 1. βyd, S rcstt , Pi\βsc :t, and Tarjcuo [l~ βon ]
β.:ichrsSβ.a~ β.c, S a t e rs e- mr , and Id1 j e-nger [1β cc β.~ ] ) β’A priority queue is: an - ,rdec βed c -S. on wht cli the β 11.. at cii; βiβ c c i - - β.r c:
are allowed: adding an element t . β. the β~u c u β . - , retr oivci : i g L}:β.β cβ.ir i imum-valu e
-I
L. -
- - ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-;βler,s-.βrct in the queue , and deletin g 5cc element w i. - e l- .> c a t t c .β:, is cc. am β.
t βr- .c c : c th~ qu-..ci.rc. By using blnu-::J al tree: ( Vu.iller cβLri ( l , r , β j , S r a:. [ i- i β , - ] a ,
lef t ic βt trees (Kn uth [it ] ) , or 2β 3 t rees (Am a , Hoj cr :β. :βt , co β.o βOiLman
[I s , β ! J ) O fl β c can .Lsm-lement p r t cr ity queue cm eβrat , i :,c s - t hat t i ccI re
0(log so I 5 ,. :: β: , where n is the sisoe 0t β the queu e . Ticβ:, β I βcc lcoc β u nt c aO i βon :
als o allow one to combine two queues ~x i t u a larger i;uβ.βce ( ce : tr βci rig the
smalle r queues ) in 0(log n) time.
TI β the values of the elements in an - - r dere β.i cci ar e I :ote~ βr . of
cm βle rat e sict oc e , then the ordered set operations can be so c-cOle β . ! un . β .β:ing
a k - s - a ss radi x :. -iβ t , , one can sort so integers in the range I t :
in 0(Ian+n) time (Knuth [l c βI 73 ] ) . Feter van andβ; Boas has devi sed a
method for to , , leroent ing i- ri crity queues with integer values in the rangβ.β
1 to n soo that the onue ue βon erat i on : require 0(log log n) It s i e
(van :βhsde Boa:, 1-lan:, and llt ,jkstra [ l-1β~~] ) .
A gra~ph is a set of vertices and a set of edges, each edg e a :ih r
of vertices. (~ne way to represent a graph is by a twc-dim ensional array A
c alled air adjacency matrix. The value of A(i , j ) is one I :β ( i , j ) is ccc
edge of the gr aph ; oth erw ise the value of A(i , j ) i s - sce r- β . An βaJteβrr:crt β.c
way to represent a graph is by an adjacency structure , whi ch .0: a1β. :a rr β .:.β
of lists: , one for each vertex. The list for vertex i cβ.ird cidri .β vert ex
if and only if (i , j ) is an edg e of the graph , f cc-so Figur e L~~1 β’
[Fi.gure l~. β~- ]
The adjacency matrix representation saves 5.0-ace i t β the grap h is i βm:β
( i . e . , most jco ssible edges are present); it also allows cm rs - i, t e s t th e
presence of a given edge in constant time . However , Anderaa and kosβenb-urg
conjectur ed (Rosenberg [1973]) and Rivest and Vu.illem.in [1075] j -roved t h a t0
testing any non-trivial monoton i cβ! grap h property requ ir e: c2(nβ )
*7ββ A graph propert y is non-trivial if for ar ty ii the property is true iβ-:rsome graph of ti ve~~ TEes and false for some other grap h o~ fl vertices .A graph property is monotone if adding edges to a graph does not changethe property from true to false.
25
- β ~~~~~~~~~~~~ ~~~~~~~~~~~ β - β ....
~~~~~~ ~~~~~ _ _ _ _ _ _ _ _ _ _ _ _ _ _
β~ ~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
o S i d L e β s of 0 :1β: ci- 1 , 5 crc- β O i o O β , cor al r x in i i i - β a rst so use , w h c - r β e xn .0 5 01 - ce r r u m b ic r
0 β βc- - i - t i sβ .β: In t h e - βcβ :β.: r . i~y using βin adjacency structur e , - Ic - c can search
a grur h in O ( n 4 m n ) t in . - , wlo βre ::: i s the nurcber of edge: in the gr :cc βcc;
4h us re m r e. β -r, t as-- I on by βcc adj acency st ructure ii: preferable tβcr co cci-se gr cu 1-cc .
A t ree is a gru ~ lc wi th s -ut c y cl e : . , ince a t r ee is a f r : r n }c it cain be
βoO ir e s βe r1~ s - ,5 by -so , :td , icic - .βrcc y c oms - o t s r - . A si- rc c -c o on βte t wayβ I. rec r sβsem :t
a tree is I . - .ch u e a so ot 0 β r t h e 0 , 0β - - , - - -cc :β : ~ti,c the ar -n i oS each v e r t e x
wi th res pect to- this mβ u- c . , and ct r β d c β . s i t o β m o o S - : on i r c -c o :sr r ay
(Figure ~~~~~ Thi s r en r β.βce r i t : , t i - , , βn i. cj :βs b , s - β ci: L a g a: ΒΆ i i e -~~~r β - β t 5 t , β.
be sOx b red from leav eβ: - so β- -t , wi . H, i s β 5 - c . l b - c u. βe ,I : , : iβ tc h. cc: ,c
β0 nivolving t rees .
[Fi gure h .5]
- β cr: ion .
An im~cβt βci~t and very general a].gβcri thscct c techn t que .0: recurs-I ~o c .
hecu.r sc i ofl is a method of solving a ~ r - d lem by reduc sng i β. β . ..β one r c or e
cβ stc t - r u b L em :. The sub o r .,-hi βcoc : are re-clue -on . n the same way . hβvβ.β~r l 03.1 ly
ice - subprobl ern : become s:ccaj ..l enough that they can be :. lye .1, I r oe t iy . βShe β
solutions t , - bhe smaller cuLl r ., hol e :: . . βsr t inec s β.c-oc ,.t .naJ , t . g ive c iu t β
t: the bigger :uh prohlte:r. s. until the c - iu t i ,n tβ t i c e β βr - , 5 β a c c , r β. t i
β -cc i u t e r i . As a simple exami-le ~f a recan-sionoc a lgori thm , c o t β.: Ie β~~ 5, 51 β
: β j l ,l β-w an g d efin i t i o n .f the n- tb Fibonacc i nu mber:
1~.i F ( n ) : if (n = 1) or ( n = 2 ) then 1 i - i r e F ( r i - l~ 4 r ( r n - 2 )
Using r ecursion , one can oIβtem r β tat e cdi ; r . tin -c. βc u - c l . so s o β s I S s :
than would be possible without recu .rs i I I . M any o~β ~~~~ c - . - . 1 - -s -
Inc tuding Algol, PL/l , and ~i o o k , allow rccur sciv e 1, r occur β .
2~
~~~~~ -
~~~~~ β~~~β~β
~~~~~~ _ _ _ _ _ _ _ _ _ _
ββ-- βV
whi sob -call! 5- h- .β c : ,ce l yes β) . In ci lan~~~oi 1 βe w i th β- c r 0 S I t : fac t ii t-~, sueS β. as β
F-.ob βl βl-L.Il , 0 c c - - can 1β :: β , 1β c c - c o d a recur sive :algcrol tβ Sl rr. b1.- u si : β. , -n ctack to
cI -r e Lii β generated cβ: rt βn r -1-1 c c β ,: (Ah o, Hopcroc βt , and ilLs-cu [1 i~βb ] ~~.β~-
Dynamic a r β - g r β -uβx , ,- a y (c β.- lhn ct n [ 1- 0 51] ) can be vieawc - - c β .s- a c s es -I el kind
:1β re βsβ ,lr:i- βnl in wS c~i sβic ,nne k e - ec sc t rc β .c β . β, on: the gen βcnβ t ccl c c , : ~~~
0- S I β - cc l: and
never SOlci-cc i_ S ic - βccc I cβ.cb icm:: twice . As di i exonin c le- no β the orβ n - so wh , I β c i -oar ,
βbe s:ived in thi s . i c c β . , c βmc c i icr t h e β. βorc~ utat S on of c_ Soc c r β i _ S o Fib s - c o o t ,
c s inc :her. A re -c-un - : S βcc i r - sce iaar- bci:- c - -i so 0 . J re ;u~I re: t1:c,-c so βcc 1, nal
to i_ Sic sine of F (n o t o so - ccc: ut so F(n) ; such a ~ r βcce-luI- s e n - c β -cc .. :
F(:iflβi) c- rio utal,ion , cc β F(i) for ~βac:o i in the r - n : , β - so - ccc 1 to on
A Sc ot t or way- t c rcj ute F ( r ~~ i~: to cnccc iβute each F( i~ lu - S - m cc c c cc
each value nO β t . The m o , t eff icient way I . β i cci- loo cc ec ot a !vβcc ar :mi c r e- rc e βo: .ing
-~~go-rit 1-im I: 0. β- set . on -:t i_ al Ic 0:β solut ,i cmi: to all , on! , o so I - ,β:β:, - β nc to ft II
In the tat -b e fiβur:c ~cc :Llβ. c t 5 β - βo larg sct β. β.j i O r - b βLeoc c. Gometi , c: iic .β β ne c ar .
-I i sβcard t1i- β soIl - 5 0 1 β. cc ns for ,:mccal. l sub socob l ems as: the corn iutat i - -cc :
and reβuse I- l i - c so ace I c r larger suho r -bl , eons - . One can evaluate F ( n ) i s,
ho time wit S β. two s tor a g e la-nati ons h I; using t,h t 5 1 . S i -β β . , (hIβ β . O U r ,
u s i ng a ci -c ed -c β- β-rmcn exo r- .β:c ton t s r F(n ) result: in an even :βn st fn r
o sis-~~βr t :5 β. 5 c c . )
Ioyri csc ,l c progra mcunu ng has - been ~ cββ d with gre -nt - : 0 0 C c : : cr β. a c cc-:: - β so of
combinatorial r - βt Iers :, in- .c . -cc .ng sin r βt β :c c ath r β.βoi , ,βc, ecβ ,~ (F1o-1/ βi [ I β.- 2 5 ,
context-free lan guage j a r s i n g (Younger [3. 1- 7 ] , icIa r i β..oy [1Β°β7 - β ] S , e lβS Iβ
correction in context-free languages (Ah o and le t erc cci [ 1βT β,Iβ } - , and
construction of - β . t imucn binary search trees (Knuth [1 tβl ,) , It ad [ ic c : ] ) .
Graj -h .βOear ch lng.
~β~o:S graph rob lem: require for thei r solul i - o n a :1:5 coo - ca t I so cs-e S 5~
of e~cp1oring a graph . A search i s - an examination cf t , hβ 1β~ edger of a gr oan St
using the : βn Ll. nwβhr cg r βocedure .
27
_ _ _ _ I -,__ ~~~~
- - -
_-w-β β ~~~ ~~~~~~~~~ - β~
β--β- ,- - - _β β β -
i_ c: 1 ( S a l . b β i~ Z c t t i n : ββ uβs- all ic-ige : and n-e r t ices - : the gra-rh cl ew
- p , . βrβ β ,β~ -
2 (ch - .- : c ~β a c:β w :ta ~~ . rsg ver tex) : If cc new vertex e x i s t s , halt .
(βThe -on , ri- gra~,h In-a: : β:β e - l O ex~I ore-I . ,t-he~~ βl cc , ctr- ~c:e a cc ~~w
cβ-~rt β,:X βccβ.i c :c: β.rk c. β., - i t ( - c-on - I βc β-β - S )
β en, β. ( - βsoβ: : S o r - c~ . - n .gβ β : IS β r, o 0 , - c a edges ieau ccwβsy c r - coo :11 β. rc_ice :,
g to . + -son, 2. ( AIJ of the +cr-io So recnchat le frorc c the cu .rr-:r, H
-.- βrtex h-i : Le er s ec: 1 r -~ o . - 5 , - nβ w : e , choose a new -c ,~~e lea-c l ~~
βcaalβ :d- β - m an .11 y e - :β β β . β- βa rc o d e ed ge ~ .o . If the - βth oo r er β..Io, - ., r c c ,
..1 the edge so oc β :i+ . cco βer e. I t c . : . ~ β βc . ea β5 s-t en, 3.
- β - s- is - so ]_ l c l t y c ,5~βj Sβ all v er t i c e s i n the gβ- oac io to o~β :caneciβceβd
βciβ: o- β cs- -ho t S - I β - β so coos the . β ,I c-st . - a rβ. v-or t~.- ,-: ,, e le cΒ± ci: in ste : 2. Thor : the
ccc ~~ .β .o - n- - c c . The r e t of the spascnc org t r ee is Use
c. ant β;βcr c _ βc:βs . The e-dgc~ of the spretβJ β..rig t ree axβe use e~~~eβ~ at , , en lead i _ O
c βc β cw veβott,oes when cxc s-- c in step 3 . The oro p er t ies of the soranning tre e
end βrc o r : the -r .~t eβr β β c us-cd t βc select the st~~~~ing vertex in cβtep 2 -β~tβc
i_ me -- - s~~ s to -cX 3 lore In ctep 5 . For :j me simple graph 1:rcc to lLe.rns, such ci:
c β -.rc : ~rsg - β - :5,0 β.- β β- l , C l L c ) ~~~ cc erc t s (los--cr , :-I βt :e t .s t Tarj -an [l~~ 3 c ] ) . 0-i-β.-β o r b - s o cc: β
βx β . t r a l , : ur , is :ati:i β asc t my. However , c β .r harder graph βic r β-tIe -cc :: the
β c-ct i:rc~it i orβ. .rdβ-r is βc ru sc,i cβj .
iii a de1i_ h - f ir st s-c-arcs, the so. .cge selected in step 5 is an edge -Β°ut ,
of the last explored ver β,ooβY. aci th can d,i - βiate edges. If a det t l - c - c βI rst search
is perfor med on arm ur: β.c, l O-β β c - c gr βej~h , the generated spanning tree ha: Zo oβ
rβc~.ert-j that ill nun -t r β - c - - - cr - β s cec nr ieet . βso-o rt l β.ces related in the S i
(Tarj an [lβ~ t β - I ) . , β e i - β S gurβ- ~~~~ i f :uch a search i s 1er f rlc sβ i on a
directed graph ar,S tI . - β,- β- r c. , i n- c , : are : : a β t c ~c r βasi Oβrorc. 1 t e t o as- they βsr-
marked - Id , then no fl~~r i~~t cβ edge leads: from a vertex U. a vertex which i s
-
28
-~~~~~~ 1β .β
β-_-
~~~~~~~- -
bo th ii β 5 I c e y - cr β s β - : β - oβ β ,: c β . n i - : c o o nβ - l c + t ecd in t h e , β : c β . : o - i c g S s o - - - β ( - t - c a , : , [1 - β1]
he β s β 5:1 gn.u -~ - 0 .7 . A oβ.-: β. 5c ~ 1l aβ:t ,βea1β C~ r can ice LIOn l β .cc :: - :c :c, β .S βcc cc c β
β β. -mβ aI I S : c c , exos 1 ,1 c it . :1 cisc !: S . H -re β t h e β old n-ci t - so-
511 g n s c β - I .~
In c c c β r β β + S c : β I βirr β O .β c,ββ n β .βl c , 55 β i, Igi β sβ IL- β. β β β c l β s ,β β β β. β - c o_i , c m , β- β
souL of ti cβ.,β :β ,I r , - t. ccsβsc , IL βn - - β - c nβ . - c I oc:- s β +Β°~ h
c cir t ,i t iconsc S n β . - v s : r t i β_ β e β : I n β . - v o β β c β :~ β - o c d i r c g c : - 5 . n I d : β s -_ i s β β - O r - βis ,
L I c e c t . βsrβt osβ - n βo cc-s . in ar: us-c a: 5 β - β . β β S βr u i s -.c β _I ld Lβd fβ, .0 m sr ceC ~β β ccβ β β - -
I - ic e c oins- c li -v -c l r in l w β - c β .I . h - β - - o : : - - - β.βe l : : I n a , 15 iβ . . β Z c : β . l 1:. r β l β ,.
1cc- tI c β t β n β -o β c c o ILo β\β ci S a lc β..β- , - l 0 β . β 1 βJi - βso O S - _ I c - 1, - r o n - o c t c.βiβ β I . , Z o c Fl β c r β β
0, L n β - - c + c i U r β c β,Ir β c βt cean βc- i , j β:c ,c : 5 β β. I~ :. c b- - c , - c , : Odi 0, - i rs β . β on t u e - - s c_ so _ t soc c Isc I
[ 0 - 1 β. βoccβ-
Both - , lβ.β c t , h c β I i n β _ β c etic i - r a - β , c o l c β o ββ r r βl , cc:βtr - - βSi , c i : β n r c - H β ; i~o n J β β β - cc β -
u_ c r ig ou r s βo_d , β ac- β!icl, - β5 so : - β cβ I s i c o utβ sβ b ,c-~~ β , n β β ci nβ c β β β . o ( r n o β, l β. 1:β,
eiq-l.ccβe c _ Ic c r β nβ . - r S - β -c , :β , β i - - :g . β β cβ - : : . , β.,5~~5 5 β.1~,5~ cc , , , - , c in ,β oc:5β:β i c - -n s o - so c
s β e :ciaβh , me: t S c S . , , - - β.β - r-el - it I: β βnβ:, I m i c c i o i , U r o t - tape-I - β.~l .~ βsi :e :dlβi - (h : iu c~l : ~~~~~~~
lex ,i β . - n:β β :~h,5 ββ . β. β : β .,c βc l i ( s β- t ic [1 β 5 ] : hoβ - - . β . : i so 0 β.r. c ur-c m l ,cr β cls- r ~~. βno d
.βlro r t - - : 1 - i _ I c - c - S cc -arc h βs ( 1n1 ik : tr - i [ 1 β . - J , β - ho c. β - β - i i [1 Β° β ] ) . c c c β
u: βl β: s l .
Oj I ,dcii zati cnn 5-I -I h in d :.
A large c l a s sβ. of n β - s te-cβ,, β rc β- - c i , i β - . β β l i i β β c o c , -i x i : : , I c c c c t i - βn c : a ββ s c S i cc
ii -β Lined on a cβOra~ h with we I guS eβd c - O f - β:. [ 0 i s cu r - _ I 1β; so_-c I-Ic β I 1:ββ , β β
t . } io β~ e r -ob ltom r β ci: l i n e - c o _ nβ n - l i s t - -go -n- :β- - ,- βlβ β cO - c β , , ng r u t βI , n c c. ( 1 Β° - u s i c i g I i -- β~~ 1.
Nemhauser and Garfinkel [1, Β° βdj βI , S c o t t , c β S S β . - r :e lgnr i thoβ.. t I r c e t β . β. β, o n , β rβ ci , β c s rβs - s-nβ-
-~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
~~~~~~~~~
-
~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~ - β
- _~~~~~~~~~~~~~~~~~~~~~~~~~
_ β ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- β.β
I I n t o , -: β . r or S c i t βce -t oβ : rβ - gn - βin c sc :d c c g c :,ethc βcJc ar e cj v a , i lab lo- for their s ~iut ion .
Thes-eβ βJg cr l t ints: useβ S w t o - - Icc I quo : , ~~~~~~~~~~~~ and au~~nentatio n. The most
general setting tonβ tht β .csn e S- chni ques ic c in matroid th eo rtβ.β (Lawler [19β7i_ .}),
O β - n t on e can ccnd β.βrc f β m i and cc β s - : ly the t i - - s -h i s I que s - t o grap h rotclemr wi, I S i - cit
0 0 i rc~β ~ib β i t cccci t re ,r ,1,.
h - -n , ider t S o t c J I β 1,1β ,β , of n, βic , - LI :,g, in c c ce- I, wt t [s we-s-I ~h t e - -,c i,, β.c cr :t β , a
βo, cc.x iocβcββ β -,oβ . h c ~. s~β_i O s βet - - β.5 ,i , c~,βccs g s o - r β t c e j n s add,i t ,I -onnaj, so , βn c t n - c c : a t : . The
c oil so 0 - 41 c β β β , β s- c e - t h y - I o , I c,cs Β°, : ~- usefu l I r s 5 olv ng thi s- so - - c - I β c c , .
t, t i , , c β - , β -co , βc nO , β I v w - - I β . f c c . . : : c5 , . I r β . ,- β - .β β - - I - - : n s i c - n β , O c c β a v l β . o c c t t , ,
I l gh c - β a , t csscti.U, r sg c~ ci , - I , β’ β β - β lβ:o:, c-nt β 1 β β βLc βcc : - r r i . When i:-~-cJc _J riJJig an
c l- c- i β. , , a - I - i it. t o i _ I c e c c s c β : β - O i f β s o β β - - β. ~βlt s _ a ,-f the :~ t s - e t sat isf ies
lie β . : , , β 0 01 0 i, β s s β β~ .β,~β - cc β i o β -~~ t - soβ~ et -c βc ,~rs c s-w i β, . βFIre rβ β :uJ,t -βj rt :
β - -r βt-ainlβ,β c c i 0 i , :O β ,i o β :β β i c - - c c c β I~~~, 0 , β. . β~β β ββ βq : 0 β : n - β .O e so n c - O ,I t β - β I c , β , the
sβ.d, :- β t -hβ . 11 5β -- - 0 β 5β.βo_ X l oct10 β . ; u _ c s - - l I 1β: a - - I girt β.1ββ.β so -14,-on to wt c I c i , thi s
t i _ hi - cd ~ , c c~j 11 t i - I β .β is cβ ,.S o . i i : o β u- :: -urn . r s g β. r ec r ot 1 - - oc ( Kru sβkal [l-~5β I ,
} r im [ 1 - c - I Hβ L βij k : trc i [ 1 Β° - H . Yao [I , r β~ J , h is e r , h t o r c and Tur ,l an [ 1 . T c f l .β~z,-cs i f the go- -c- , Iy c o o β S 0 : - 0 doe: r i o t nc β i a -c e -
~ t im ad, -oniu t ,i n r c , , it rcc o β
produc e solution s whl - βS ar- so le-ce t, -o ~-t~ oβcaJ, (Garey and +βohiis β01 [ li_ β.βlβ ] ) ,
and it S, s u sually eβa, I\ t o . c β , n, i - :me-r: t and fast.
In si t ua t i on: w h o - c c i_ lie- gr - - βed ,y s - c - - S i : - - ,I doesn βt work , a method of
j , - βrat l β. 0: ir o :J so cv β - o: c β βo : β : - β oβ . βc S I-os ββ: .i n i- : . The idea is to sta rt wi th army
:ulut n β-tm t - the β.corn s: S - r c ~ I a β , and look t βscr a way to au~ nent t hu weight of
c_ l i e s- iut ,i on by making ic e- oil ch βcr g β- . . The new s-solut Ion is tI β .e-n ini ~ r- -ved
S r i the same β roc ~β, arid 1- l i β β nβ .βeβ~ sβ: i s ce - i r S inued un t i l no ir~ r -vβso- :tβ. H is
J . c ::i S-ic . hinder -il l -ro b c -I o t t , β so - nc -lit! O r β . : such a locally maximal lii i .5 - β .r c
j sc also glob ally β:. :o_xcdβ:: , ic β , . 0-Oven if the s l in t , : 011 . 5 β not guaranteed S c 1
r , - ,x ,I ooβcmc , + , Si , - c - c ~~cc o β rs β 0β . β. - - - i t method coi a,y be cm good heuristic ; I β r !n:t β c r ~~- β.β- ,
~
~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~ ~~~~~~ ~~~~~-~~
---- --
5 ,in [10 O c ] 11:1: cβco- i β lj ed 5 . 5 wi t S c gcocc β. i soc ,. , β ci t : tβ - I , S r β - nβo β .β, o - t i I r c f :βsiββ. _ βooc - in
-~ c β s β - t i c s - c . The trc n\ β -I J_ Ing : a , I - β - _ I s I n β r l o l,cocc is- I o n f ir s], a : i s rS- β,, β I -oc_ c,cloc β
t hinβ . cc β.tI s oil], βcβββ . rt- β I ce~β -it β a flβ s: i s a I I I β . 1,_i, :5 cure -c , , .rc t l ,e β.-dgβ s .
i i c u :cI βt t r n : cycle onβ β β O - o , , ,β~ βcc -i - al soβ . : β β β . 1β S i c β - β n β- β.β; - LlI :β.g :of_ es βos ,s-u : :n β :
I .
i β .~~t β5: r i a t i n~ 0 - 5 t h
born e nβ , .ble:cc: require more :n~ h i :tic:it β._-~ , cβ 5tβ. cc ,a ns I β n β ccl at i , .o c i s_s -cr I:
I β o::cc lie β.-j i_ t S r theβ clβ,c~ ionβs dtt .~~ :t nβusoβl co - , d,, :cos - s:e- S - - β on β β, β ion thi c c - c e-t I , -c - c βs .
Tlβ~iβoe ac.Lvocncod teoh βsr i ques have been ievi:ed h r -I β βnl.in ~; with , 0 E on β ee-β d i v e r se
pr ob βLeo cos- wh ich re- cu , nβ s - ,.LI ,β rc os - :. I t o iii -ao -β.t, I : β cq βif .~oit :β.. Thecβ..β t β-cl r no , I ccc: β.rc
-s-t I , c,O β . ~~~j - l β .,~~~ si a. c - c c β, , n - ion Iβs c c β, c mec c :ccrc t , βc_in n linear arl- ccgeo: βson β.β..
l ath cβs-Ic,β : r - c s c , c i o βn .H a o , c th β . s i, cc! β c β.cc J,v .I rrg the c β ilcr βsi o c g, . s β s , . - , --
Iβ cn r ccidenβ a uncivc βsr s - ,,c βif - I, - , c - β ~ , O , , ccc β. I t I - -nec -I I n I t cialbIc,β ,lflto singlet onβs
ce t s. A:sc .cciatcd wi_ ti β. e t c h β - 1cccc β .~βc c c I: a βc-J oβ . cβ e icβi sb S . s-c c able S.c
-sβarrβj out βS-ir e 1 β -l ,βi , r wlng on ββr oot -cc . - rc t he s β S
thlβiion : C ornb i cβs- - I c , ,β s- β d c s - o r β β.: cc s- I~ongle c s c t . ],c:S lβ β.β r c n t5 , cs -s-id
βT i d aL β . β : 0 - t o - i l : β; t S r β β β-c -c l_ac: -c~Sβ β~β,l βlβ.,βcβcβ.cchc c in cc giver. :ei Is- i a
c o n s i s t β co s t a- c j , β .
1- Ov aluate: hβ β. n .j βc - ~5c~c value a c : s o c J - t e - i wi _ Sc or given β leone-nit.
A situation co f thi s kind ec cc ucβ c in the cc β cc ci 1:5 ,1 βc . oct F,dI βTIIA b β. β-7 . β ~1, c -sa cS
r - IQ T IVAI β.sc S fl E s-t oil ements - ( c β;.Ji cr cunci Fischer [lβ.β.o βIn ] -u n-I in several . -tOt er
combinatorial problems (1 cm r , β,a n [l β.βt β c b f l . The s-c t un ,i c ~n so-t i e-- t cc Inc
discussed in iβ-βc c t- I βr c 5 is- the :imi-ie:I such r i - I β β : : . . -1 :tl βl . β .n β -n d Fischer
[ ltc ib J proposed an a1gor !t βii o ~. fo r thi s n r - - h β , . ecβ u : ing I n - - : :0: β a d : i t s-c
- β_ β -β
-ββ-----β -ββ F ~~~~~~~~~~~ ~~~~~~~~~ -,β β ββ_-- - β -β-β- _β.-_ β
_~~~~~~
β β
s i_ r i : - , - : . :clβ βc. S-Ic [in-n :- ,β and 0-horn sβ c β .c r i !βr -at ed the set . us-β .I o n, J I β S e-os, aSs - c , S rjic .g,
- so -c : , :-u t e ccc l n icocuoc : s- .j ccnning trees and ~ rojβ-ose~1 an I soc reve l co ce c ich: si βic ing
c-ath cocc i iress . : .n on trees (β.~~o, l-Iopcroft, and Ullinan [1 1β . ] ) . T he i r r cetI βc c . βd ,
whiclr is verb ,β ,c imβl le t o program but very hard to ccn aic_ s :e , gene -ro il i βcoe cβ. t- ,β s-n
c β.u s-cob er of o t i c β . βn β problems (Tar j an [1β β1β5b}) . I shall cU scu sβs t h c _ s ooc ~ 1hβo u
and it: rc βnc :βoβ cβ:ah β Ii e running t i s - c in , β , βs c t i s orβi 5.
Pa: cctho βr n -ob len oc icc β.β c O v i n g disjoint :et: is the iβ s - l i_ lowβ or g . hu n -n ecβ e
t i le vertices cf a gra~ O i are initially β c ari-itioned into several subsets.
βs , , β,-,β I c i to c βci,rid th e coarsest i arti . t ,i sc n which is a refin ement β- β s the gIven
- rn- cS .1 β c c curd whIch is-- 1-res eo βn-- e-- βs under catj acency, In the sen se that If tow- β
β;sβri ,L ces βso ari d w are contained in the same sub:et of the cc ,r t i t l c c c r ,
cN- cO: c _ i r e s e t : A( β .β) {x (v,x β) i s- an edge) and A(w) = (βcoβ, x ) is s-uβ. e-~ige l
I o , S, - - r β , oct -n x-β5et i , ,β the c anous- :βocc-β,i_ er of times with each sub s--ct of tine partit i on .
TOil s adj acencc :β.β β c r e - se -rn -on ag s-rI 15 β1 on ir β easi ly -c so- n os~ utabbt e in O(nm ) time .
5 5 c r β ft [lyIlJ ~eβn-βised a os-c - r-c soncphi:t.Ic ated algorithm which run s in
O(m log n) i c c oorec . Gries [ iβdβI~~~] give : a n ice de scription -s-f this algorcitoincβ.
S anti, t err ret βlocc -crnerot cI sc u:e:βu,], in solving the st ate minimi zation problem
βor finite automata (Harrison [l β . s5 J ) and in t e s t i n g graph s for is-c β. rnn r iinl corn
( β1β m e l II, anβ,i Gotβ. 1 - β.~~β.β [1β T
A third r U t - s - n c so - c :L:r r:g - a good cot t on updating meth o d is 5 0 c c - linear
arrangement l re -bc I -c : c : Given ci set oS n elements and a c . ct ,t β.cct ,l β to el
subsets of the β ,-lernt-n t .cc , Can S - i ce elements be arranged in a line 5:0 that
each sub set -occur: -~~ - β o c t . 1 gu -cusc ly ? Thi tc
~ r cb l em arise: in biochemicl-ry
(Benzer [ 1Β°β,. β ] ) and in arch aeolo~~r (Kendall [11 β . βc ] ) Booth and Lueker
[1 r :- } hav e devised a s- , + ,β t h i o ) , t of solvi ng this problem in 0 ( n f r n β S βc c - ,
where m is- the total sir: β: β of the sub scβ. βt: , using a data struc t ure they
call a S - S
β~~~~~~ ~~~~~~~~~~~~~~~~~ ~~β
~~~~~~~~~~~~~~~~~~~~~ --β.- β -
- -β_β β
-- β~~~~~
-β ,β β -. - ββ -β β~-r β β ββ β β β β β . β - β β ββ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~ β 'β β β β β __
~β’luI!IIIII~
,~βc β β s - i r 0 -b ce ~~~~~~~~~~~~
i β I i c r e - arcβ tw - rnet ,ho ,,i. - ,os - β ::-ln-βJ . ng fl βcs-Ic a r,β L - le β , , , - s c β : - : . ; - I n - I or β . c c - c
shrinking, which are re lc it e-c to the algebrai c cc ,c r β .c β .cn n - : -f ,βi L. o β.J,g~:i:ra and
is β -coroncc β.orjβ h i s-c:. brie way i_ cc- solve cen βi ,a,i or tIral Ii nor l βcrc , o I,: t c .- - ce -e ons- ~. β- c - ,- n - h o,.
graph into several :ubgras i on , s- solve the r. cb leβc, on the cc ic- ~n~β : o , , , and
cocc i_- inc the :cclut l ons to gin-β β t i r e - :- loit i ~cn βsβ n the c -nc t , ir e gram Si. 115 mo:βΒ°
instances wS β. ,,-n β c thi s t , c . o i l r s ,l que is , . β- : ,sI , the :us--gn β cc i is , arcβ C. βc _ cc oβo,c c s t ,
c oni gn - ce b r , ) s -at - i c :βy i n g some c n βc n r - s o t i v l ty r βclc t ,i c , . I c βc β - r o e-n - t ,, .-
1β,- the t t β ,β:c nr . I . 1:15β . )n c β .O. I O: n β . a - _c β . e f β f l c i β . c c r c t βsoct: ; t o - j - ,- t β βr ~β l n n c the
β cβ n e - n i t : . ,~oc β .d alg- n c _ i_ t i : : : , . βs-xi s:t f . .cr a βsar i ctβs of β . comβ::β l vi, 5,β; ~
r .c hle -n cr :
I:c,r j an {ltyc: J , :i ss ~ en - : 5 β _n c . L βiβarjarr [ l973a], Sicβ s- cr c t and iβaβsβ, 0 anl [ I r s-β.o c J ,
b β .caui t [I ~βL ] , βi anβ. i - u , [ l Β° β I β . a L Tar j aYi [ l ~~0β 5 c J ) .
β. t i c - β:β β,-e cc; S . s-u βcc : 0- _ c er :5s in r, - co ,Lecc o sβ ic t _ i n r i n n k βcct s -f the gr a~ ls
β0 cc, :cIngiβ β - r β - c-c , solve the c rβ, :blej o : - or the - shrunken grap h by aβ~β~ ly-I ng
β is - β i-lea r e -c βs-orβ: β β-βi:- , and fr- s-rn this - c β Ici tion compute the :.βLuti βnc -c , the
or.l,c I oral grc i5 I,. The :iciβ ,I cr5β : ,c n i g Crab I on so s-nr c cc βccci : βt on t oil-Jon β.β. a On . c β s , - : -mc O i l s
image of the ~nβ-cs: in . -Jo.βnerally the J- :.~~t of the graph t o n be sβiiruiilβs is a
-c lon c lβ- or a union of cycle:.
3
ββ β β _ βV β -β βββ - - ββ- βv_~-
β . Ton Ti βcr - ,c t :β.β~~~~β β :β - 0 βi_ c .ccccs β
There are i β .ci c , -.ir~- .c: of so , -mb ina t . c r i s-il, n - -roblencr s β β, β , r wir I cci β . g cod algorithconc
βire ccc: we. Thi s ccci , i β cc o β:-: - c βso c ,i nes ten such n , rcb l eooc : . I have s-elected the
leo:,: -c c S- i s- t a : is s-f t i e - i so 1 0cc β. n βs -se- Β° : , t i re - r:mge- s-f t - ,cchr , i le e: l i c e β;
c,ij~ - , l c cc~,β l β s - β s .l ,_ , c c : c β βtβs wi c cl c - - . βs-lie- .l I c βi I s - nβ;- β. c - I _- c _c, β. t i l e β c β f , oc : , β tlβcβ:
cc be r eβs Ve - sOL !i S ort ,iβs i,β ofβ nβcilcicoβ:c with good nig: r β βc}c c β . . β1 an le 5.1
1,5 : tc i _ S i t I o n β .β.l lec- ,: :_cβ, β .S t l i s c t e , βSo , o β . b c s - c- : : used. in t i r e best - β~~ .;β : 1 i_ b r:,: t i n t S r - c β . .
FiI β.c β.iββ β 5.1 β - i c β - a β : ic - so so. β, o .ccc oe - n βct : in : Is - cS βS cc tc βIc c βo cc,cS cIeve d 1ββ β; βcr n β. bs-7,β β - -ccβ
th ese - r ot- I- β cc-c .
[Table 5 .1J
[F i ~~ uβc 5. 1]
hi : c r e t e i- cur ler βi r - c o n -n β .
- i_ y e -n : - β s c r oβs ic I s - c ero sβiononl_ s-βt β soS , :, βnβ ( β . s , :s1 , . . ., a 1) , the dl s-crete Fourier
crcuc: t β. so:: r -s 1 - - : , 15 t o :~~npu te the βsec t, r (h ,, , 0c 1 , . β’~ b 1) given by
n-i - ,5β , ~5 a.β~
~β . ,Li
-_β .ji c- ,~ are the (c c:snβ Itcx β) c - s - S o r o t s os-β
11= 0
β n, - β . βiβiβls -c n r o l - ,β β -,,n - ,,, :ns- i, rc c I o β'. r βc o β . , , L c r - s o c - : s - .r o β . . Ann aino r βt Oi , β c . β β , -r tine -
ti c c r e t e is- cuβ,icβnβ t r s - - o c : o soβs- 1: nc - c a l cls β a sob c - o .ct ine inc βsar I- -Us β ar β thc- :ot .ic
l β s β I I s - β - ,:. - : β β .s - i i l n g ~ol~~βcco . c:~~~I β.β c .~ β, u a t i ,n n cuc , i i . n :t βc βs - v - l c d I nns-
-ar , d .1o .t cgscr cccl β s , l,vr s-:o ,i a β. β ,i~L t i n 1 b c β c .,, β c , I r β. (~~ uth [it β.β 1, β1β . h β : :so βt .
aj β . I T β i , _ ;:scu r [ l β ~β~- ] , F-c -is - ct β r~ βun β- Sc _ t r i m [ l o n β5 ] i .
t is :t r c i βc g i r t f -sow - β:. i t o cottipu te the discrete F ourier t r :cn s-lβ -β- rclc in
O(rβ.β ) t ime , t o lc βc_ - cnn -c i βc,cβ, cc l; ~~~~ 5 1 ,βn ularized an 0(n log nβs -tics-cs-
rn - f l o e - I , ~~soj, j , c J tb - βas βc _ Sβ uβ.u βier t r c cn c c β- -n cr . They were not the fi rst to use
5 5 c c met ,I c . βd , w l r l s o h - rl~~1nated at l east as early as: Runge and K~onc .1g [1-il-U]
The t βasβ t F c_ : βr r - er trcu i s - f : nβ uses re-canβ. I βix ton cut ,L- β~a c the amount of
31~
β _~~~~~~~~~ , ~~
β’β
~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .β -
β .ββ -~~~~~~~~~~~~ r β ,_ ~, ,,.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -
:-,rβsβs ut .β.ti, c c . c β.t βcen :I ly Winsogrβda [ 19 75, 1 ilβ I rnβs- o, u-c a cβ ,cβt,iβ.oj c β :r
β.c cocil u t ing the d i s c r e t - j- βoun . I no r tn -- ins βs , - βs--coo u .. l rsg - - id ,β ββ- ( r i β s cs - β-c,~t i β s il -cor n - I cons .
This s-ctcβt.ho:,c oc oc,y be sun βcr1 nβ t , the c β ci:t F-c- βori β-r rocβs, 0 rβoc , , , r s r β .ct.ice ,
alt .S r - agO: Wi n , β:g rad has- n . - βt cj ncj β,oncec -i the vers-J .1 nβs-icr,β b O c , : β t l o c r e ot β Oils-
a,lg on c_ i_ ui : .
0-t atn i x 0-Sc c,I t i n i c l c cc r t i -c , .
bl ve -ni S cc-so n ~ n cs- ci tr icβc s- , the :β .o ,t n - c i x os- ui_t oi l ILiso at i cc ro βui ec :c is I β
is-at ne c_ O ,e . i r o:o: ct rct s-s c - i β ,:d,u s o t . flit - -ct cc_r olan d Sr i gh se-Sic-al c- ,el S βr sβco .i os-c β
ccci t n i x βs-ui_ f l li β.cc cβiti , c -n βs re coil res 0(nββ ) i _ c l c: ,e- . b t r acsc,βrn [I~ β.β oiI βVI c β tid on
wcβc,,β, to mu,,cct , c c l v two 2~~ 2 matr ices with only s even c : Ou i i _ i r i l c a t , : , and
c_ sc -c d lo is- t i c I n c a nβcccβ~irsβ ye ocratr ix mult i tβl~ c.cat ion al_gcr~I tOs coc i- cc cub sIrβ.β.; only
1o~ ,,, 70(0: c i : : , β . f l β . irβ -sumc rising rc-:c_Jt has acted as- a stimuli: f,β nβ
cs-c ub r- cce -uβ ccS r in l ,5s-: c . coo ,c le -βxity of algebraic I. ro lc ieo:o: . No - ne i_-or . a- .
wi_s e t h- s - n - ,Β°;t r cr :s-en β s algorithm is- improvable. Stnβa,:sen β s-- ;c,lgβ.β.nithrcc has
b-ocr: used to .β c:::c :ii e tncucc-itive cloc:ures- of gras-Or: ( -buur r - [los -i] , Ficc icer
βu- s Meyer [1-cl]) and t- β do c -s-nt- ext-free language ~ a1βscing (Valiβrrr t ~lJ c5 aP
β’ 2.5.1 -in 0(n ) t . oon c: .
A pr ocUie c: : related to os -- i t r lx multi : i lca tion isβ the sβlcn-rt e:t :
n - c i_ hr- . Siβs- e rr a dI rected graph with -sitive edg e di :tβunc -: , the s i n g le
scarce sohorte s-ct i ath problem is to find the minimu ms -, l i s t - c i r c e - β.1β s-c. ci g iver β.
Vertex t - - - ev cr c- I her βc, n- -- s-on The all pai r s shortest la t i c nβ s- Sd -c β is- t -
lin t the or-,ctnni mu nβ , distance between all pairs of βsc-r I ice: . D i J i - c :Β± n - I L i5~ . I
Jts-vi sceci ar t algorithm f-s-n the single :βource βi_ noble s-c whi ch re-~nb re ,c e i S O , β . r
0(n 2 ) time or 0(m log n) βti: s-e depending upon the i s-cc l β cccc β ns l . - β β . . ~~~ . - no β-
n i s - th -s- c s βococbcnβ of vertices and in the nuns-her ~c S β ni_ge- : in the g r e - So
(Johns -con [1- -r I , β ] ) . Floyd {l I β i ] gave a way of solving the all air. :n S b β
~i 5
-
~
icr β ---β - ..,
~~~~~~ . βSi ,,,,. _ _ β ~βi- β:t,_, ,, ~~~~~~~~-_
icc βs - (o: β - ,iβ β . F~β:β , S c β occ βs c [i, β i l - ,βO n on w - c d s - S c : t o β S c - : a_il n, .-a lr : ~β ,ool βs-β :β. -.c βin-;
bβ, - β βIβ .βt- -.c us i ng -,.β .( ni β β~~β ) c- s - or β i r I s - en s - : ,β.,rj d c m ; 0(or ( l β c ~β. log on β i . cg o n ) -
t~~:de- total. Avis-, Ri-ceo- b , curd is-c c [l~~β β ] βs r c s - - e i ,~~β . t cos - i- -c_ s t ,(cn2 i ~: n -
c- m i cu-i :,s -ns cβ.re- r - β ,c c ,cJ n- r , β 5 i,on βt On ic cs-c r - β~ os -s- c Ic-cc s -cβIL β,βe Os- , -.: cibi l ~ n oβ: i_ i - - c ohn -
ibi s: J, . cwn -c_ u c~.i βis- βO re -1 β 1,0 :- iβ~βis- lβo,c c cs-c : βsc _ -iβ a β, : β β- ,c c c , s - β . c .- I e βc cco :β.c Icc o β . c , β.c f β o_l
β o- :βbauatioii c cO o a S b c_ ri c a β - Is-asic .
us -c A is an cc -, ccc c: β 5 0 - , d -,s- U is- an 0 0 -β 1 c_as- β β β is- s - c β c c o: s βOc - β . irt,
c-s i s - cβ. n - ~ 1 β, - c c i : r β β.β β.β cc nβi βliβl β βs - . ;~n c- ~1 cs-n cs-i. . β:, cc coolβs--: lb .~~~st- β r u 0 β
c s-_c β-, β - - n , : Ax = L . A :t β c: c -ctac βl con e-t ic-c d 0 β - n β ccc i lcg tS cc L cβ ~~c S:nuc : .s βis
cβl ,boJrr ::t L,β.: . , F βr β . l βt O:c - c sn r d t .lnl - .βr ~i β .- β i~~, i e c _ c β c β , c , β i β. I U . :β . r β t . . c _ ic β
i, , r- β β.~β j - c m - - s β o β 1. 0 , 1, βi~~~β~~~~ - . - d β s β s - ,-β.β s-i cr βs- rI - LβJ β’ βit s-
I l i a c I : , l as - n c _ r i c- .oc , β :lc_, oβ (i. ., L icac Iso c - c -- ct- cβ - β , , β. ,, be β . :- v-: s - I . - ,
J cig~βo : c .J) and U Is al βs -n - t n i s - u n~ .si_ar (i. β’~~ 10 has- Os-u oc i_ :β . - -
be-low the -L s- ,~on i :u ) . Th or Ac-c = βcc 1: solved in t,~, - :1- sc , by solv i ng
= tβ , β.c ccj, ieβl βciβ β ro t . -o lβ;_. 0 , , β , c c c c ,tx i nng βix = y β’ c _cUr ,ccd :. c__ cic~ . -1 , :i β .
Beca_c:βs- L arc - , S β0 have cm ecial c. , so s-cc , c r nt :- ,1β, .0 o:g β u β s 1 acok: . . .lv,i . rn g cc , .n β e
vβ ny c f f i c i β,βn t : i _ Or β s - , b β -βs - oct β ,so β. β sβ 5- iu s s- l ax: -:lI c :c ,b : : β β . - j , β.,βnc is t u e f i r _ c t
β βc c β , decomposing A m t iβ.b
βth e c te c β c,irco: ,- c i t , i,orc no β A o nβ c o d e - c s β . by , c - β β β c . β 01β ~β
, cβs- Β±_c, sβ c c t i n f l . A r c-
, J erc c , t I on c c c c or : cL s I n c of c c , c - I l ~r β.~ a o ::c _ n_ L t - ci I-c - β 0 β one nβ - rβ: of A - son o f? , , so cc a
of A β’ n - f I. toe mu l t i 5 Ic: is β c c i c o - r d ~c β oc r r - β- β . β t b y , t h e - c- odd o β t e - , i fc~ β β -β,β i . l hccv n
a zen β.- j~, cc i βe- Vi oiic r β,β c i . - r : β : : - -r-: β - β c l i i _ β . . By sol ,/ s - t c o c o : t i ccal iy :15 m l y i n g
ccc c,oh row oi erβ:cti c s - i t . , - - r u co in n - tao s - n c β . on gI c c βc,l c c β . : r t - r ,l x .-\ intc ~coi
ui β; β-r t . rsc c , rigul βcr rc:rc,tm x ~1 tb ,, IβLW cd ~ β c c l , oti s : : c r 0 os - s - - β - S , i e t βIc r c a L βcβcβ i β
tr iangul ar m- - J r , , c U .βu ( βSc is- at. 1,01 = A
L .~~ ~~~~~~βββ
- -
~~~~~~~.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~ .. ~~~~~~~~~~
~~~~~~~ β~~~~~~~~~~ β ~~~~~~~~~~ -β
If A is .r . I g,?cr: clb :, β a -,t e -n ice (o n -cocc i_by n s - o ,β : . n cc , o c r b o - ., t i i e c , Ut]
-iec -n: i: osβitj on requl resc 0(n 2 ) s-~ β t e - t c and 0( n ) s- Ic :,- .. β _ c , c : n - , n r s - calβ,-β I org
and h a- .βsk :nlving neqcolre 0(n4) s- ,. c .. lO: rn - c c~, β large s~ β cL e Β°β cf eauat i om : ,
U c ~βt - β-,βer , tb - c cβ ,cclni . x A i s cβ~ β .s- βs β cc . i-or a si ars-e cc , cr r,I c-o , t i c - c βc, , cocc ai r c .c
s t o βr cigcc s - - rac e - required by Β°Iausssβi:cn elimination depend is-c a coo cm b i_ ccf cc- ,:
way on -βn t O o βs - s-es -β β c _, β n c β s -ce - icc β- cβs-ructul-,: -s-f the ccc atr s- c-s . in 5 S β .β~~~~β . C . 5 i β . β, a
0β s-OW os- eras- ., - s - n c ccciβ,β int oβ -duce new n ri βo s - s - s - - s (call ed n--I i c β b i n ) inc t o ,ds β in - i s -n , :
so , c,aliy :- s- β’ c t is- de-sβlrβrblcc to nearrange s-,he- o,:,::trix A by me an s
c: ia β - β.c oson c: c olcsc :βc β, -, β ns - cc β_ ccat i oi β~s s-cβ- t ioa β. tO β. β: βs i l lβ in care n βc _ o-sr β . i r β .g t i : : β -β. of
βIn e Is-s_ I or c r c _ I or arcβ t βe.iuce-,cc.
For tO - - s - c c _ n : se - lIc ,t s ac e-cud , toβs- re-s rs- :eocβc s-iiβs- : - iβ - β n , β c ,~ :er- β struootβu.rc
of A by a βr e - S . 0 . TOβ.- β grouc h c sontaln s β. :-n βs nβ v rβO. β:c-: 1β r nβ s-s -cOn ron β,-,β anβs - c
- _ β - I , .os-,rr o f A , s - r β . βs-ric edge (i , j ) : β s-β~~ β:rc1r o s co β :β ncr -,s-c tc cT,~ (Ui) in A .
1:β A iso c:,βicrcβ.β r . i s β , 0 sic on i β s - e s - I β sβ if A ic ccβ,c vβ.:β. - - s - r .I s o , 0 is
,Li r e - o β, β,βci. Phβ: ~~βc: Sc 0 r-cc nβ - s - β .eoic _ , A s-is-i all os-atri ,ce ~~ fβ- sos-ce- c by
, i:- , β.cl t a,rctc β ucic c βs en c β cnu t .,i n g r β .cc c u r - i - ; . clu,r- cr r co c o o β A . Iβ,β cLc r . . t i_-a. o β~g, n - Os -β . ? r β O n l β t c u c
s-- f 0 ~t mcrs-,β be ~ .s- s Icc b βs - t β.β cIa,: ci nβ - r o-c r c c , s - versl .β.n of A such that
aiβ_ cs:icic e-ls-cs-~~0 i c t S . : r β. Is : β,. : β :β b - c I , o , n . ( I t is nece :s- -~nc to I-u - -s- n O S - c t βIS . β . β
o ~-so::utat.l - s - osβ do n -ct , c O c c i _ r c a S ho .β on u s -c βs-rI co stability cβ i c -c c βlcis-c l n rat ,b on
r -βc __c. I cOt-o h if ~ i~-nβ - c tO n s- i s - S U e ic- cr c β : see s-~ rc s- βi S o β . - βoβn,i -tolt β so [lβc5β7
o i - s -wars - -n [ l βr , -~0 J ,
i ar t c βr I l β 1] ~~ c - s-c of the first to :β.cg~oe-: t U s- cc c_ cs β s - β. βn n i c o β . c s - s - s-f S O β . .:
β - , c β ; .. The i-i βs -a ha: S - - : o ,c n β . cxl c - n r c ?vo βb y Jevcl β.β-c β e r t . F n β g βs-nen βai c β : ~β . s-
c c r r c βs βnn c ng t h e r id a t ion β. cud : n - c - t w o - - n i Ilaus- Sian cβlioc .l t i - c c i on and gnuβ So β c c β on- .
cβs-nc P c - s - c - [lβrβ:~ Ha~rary [1- r .βlI : Rose [lβ~1β~-) : P s-s -c, is-n, , βcc , βc I O - : I
[ IL - -c β - ) : Icc_c S β:β [ i r β J o and Is - c s - c e and βiβ -β tr , b ccn [107β].
[Fi gur e 5.2]
2 β β
-
~~~~~~~~~~~~~~~~~ β : ~~~. . β
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~
βF β - β β~~β β β ~~~~~~~ar - βfl . - β
A: an ~βx:c cs-c lcβ ,,,c f β the ::1 iβ -ccc , c o : c c β o r S iβossib le by t aking advant c~~o.- . 1
- c ansI 5βs- , c c o c os -t ier the graph i n Figure 5 . 1Β° . l och a -c ~ k ~,y β- : fr -n O r
ar ises β in iOn β - cc io βr er βse - , - , o l βi t βc β . crc of d i f n βerential equations. Orcc ns-o.xaβ
- βS e - r is c e c Oa-.:sβ . 1 - uβ. t βi .i cβci orβ c β s . i s- ni requires O (n~ ) ~βs ace and 0(n β β ) tnmo.β 0 β .
cue-h a cc ,c.lt βs- t on , if no = k2
. f l it - band~~ 0th sso heo c β . β .o c i β , βs - a r c e cc i i c-dn β , t i on
ic- β.O U , co βs- i t s β s β1 β ore - c to o(n~~2 ) and the time f -
- 0(r.β) (Cuthill β i c - S
β- . β S?t c o.β [1 β- - β 1 , β I t - w a n _ c o n [lβ β7~ 1 ). Ge -ni - ft c [ls-7s- I β0:, : - du v e-iβcco _ c c c c -cs- er, is-et βs βs- n β
β- , β. β βS i -βc , c:Jlcd nc- - c o c c i d i s ccet ,I __-ci , t u n c i clc r e β. iu I re s- - ( c c log o r ) so ole -c and
~~( iβ β2 ) t I c : . - .- . . 1 - β t I- - cc , - c c c β t β i n , ccx l -00 cc β [ i β d β s - ] c S : - βw c d t h at , βI ~~t t i s c I o
a β - r : ~~t an t c β a c c t O o β , o c c c c c e . ,r d.β i s s -cct , r cc n c, nβ , c on ui r cc s β β I - i n β least n β .i li~~I 0 c o βs-H
- -o s -no -ut iri g i i c c c , β 00 β c_n β,β βn - -len β I rig c s - i ce -c β : β . I β: s-β Gausc scci cc_n eb b s-i cccl I βoc - Sc. k k
go β . -on s-βrct: 5,:
Nested di scce-ct ,i β . 0 0 : _ i s β a cβ c c cOβ .ls-β: .iβ ; β - β let i c - c c wh h -sh uc e - sβ Lic e β β - c βs- I c :c , t cc
I , -: β l) ~ (, kβ .11 L?rid gran -i cβ.β n β . . 1:5- c βs- 1β or I-c ~ k g r - . i grβ H o , s -c s- i tOre
I -k+ l -vertex bound ar y betwee n βI - li ens - (Figur e 5.2). 0-1 . -s- I; βc ccβ cc r,:,trlce:
which a n i s e in p ractice do not have co c β .On a n ice - , βt n β n c - c β , o β - , d c i - r oc β
ci:k whether n β:i - β c βn d,si s-re-ct ci cs-n O r s - i c any i e - t t ur:c_I]. g βcc nen β .-cl . I :cc c J c ,~ ~~~~~~ 1β,β
Lipton . R o s e c _ n - I Ta:βj :ui (Tan , i:_n [ -, β ,-β β b I d i s cos - c e - c β- β c a ,-r cc β, t o e:sO .~ - c β . I : i o . β:t . β
- .1_i s- c e - cβ S ,i - cc i i β a r lc , i tn - in βs - β lcβcnan gr am Sc: β sod βs- that lOc o c : t β βrag - β ~β o one β β 0 : . 0 ,i I i
r ) ( n l o g n ) and the r c_ uuc In g t i m e - :t,ill o(~~~~ ) . ,~ cc lc fo al βs 5 , 5 a r i se t o o
- β dm~n~ ,i β β t s , βLβl, finite - β 1 β s-c β s - n t j-rolc J coors (I t a r t ior and flcnβ- βy [~ rc ~
- 1J. c:n j aj . P 1 βw Ar i aJ . j s i β sβ.
2y:S . β- cβ.. of . 1 1 n β . - -ar e - ; β c c i i i c o n s c-rn sot in con t ext s c βt l c o βr than linear
- β t , β - - o - i β i . i - - n . I c : l : c c c c , o , t i c β β c l c o n t e s c i , 0 - cc th r cc h lc βcβ . can be I β . -r cc . β.c ic~t β - , s c c s c o c .
t e rn of c - i l i a c I , - β on : , cs-I Lb c :,j ,n incd c :iti s-n O ti s- hatβs- ng s- tβ.iI t β a cifldl β i- s 0 0-i )fl
r β~ l ac ing mul t i ~- liccit .1 β n ( S c s c β k ln β .o u :c e mid CanrΓ© [mr 1ββ3 ] ) β’ Another sic t-uat l o r
L _β -- 4 -- .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ β . -~~~~~~~~.
β.~O c βiβ , β c y o c t c s - ocrc c t β l inear equatl ons occur is in the gU ci-ci . β1 - βon c u s a ly c i s -
βI β - - os - uto r 1β- βgr β csioi :. lu βs - i - o s - c -, for i n stan ce, that wβ wi sOn c . ccc O i l y a
c -ccc i ut-~cr iβ - βgranc so that it does not recompute an ~βs-~β~ no::: i or c. u , i c ,s-:c U sc :
β,βalls i c β OS β onβ . of the variables in the cxi ression has c }i : u l g c c - i .
The l βirs -βt c it-i β in the analysioc is to re~ re - ,β n - n i t th e- i- - f r - - c . by a
5 β) , β. -in ~ rai h. -d ,cln βc_- ocr βt .e - x ~i o n the flow gr :ci O r r e-βj i- c - :o , βs - t , , a c c , , : - - 0 , - s--c 00 β
t h e _ β βcngr rrm (a :eL of n- n grocucc :t at cco c secnt c having a sc iosgl e c c io β. o n - , ~~~~
m d c_n I
a s β .i i c gl o.c ccx ,i t~ j o c β c i n t ) . βaeOn edge in the Iβlow gra) ii !, o, β; n o c e - o r Β° : ci n c u , :0
β.βs-i β c, s-n t rol I rons: t ofle basic block t - o - another. The n β-r o c bβ), ec c , s - i β - i-H
:β - - r each basic block, the set of avai lable e~~~re ss: i β.I cn s ( t i n s - c e - win - c oh don
no-c L 0 : - c S to i - c - i-cc -rio ut. eo,i βi can then b e tβonuui.ated as: a :yst cc:: c t β linear
0 u β . O i - s - n i : wi th - β no. - β.β - sr l able f o r each ba sic block. The vc c_ r .i able is cc bit
c ccl iβ , w , i t i c β n c- hit c -rc β :~~. s-ncfl ng bβ.. each program con Ic -nc - s -s--i -o s , and
c - β , s o . o t t - β - i β l l ,β - - - I, or t i c o s - c it I - as : r e - i i s -c _ c c : adtht i c - r i and c o oc _ nl t i~ U cation in
the sys - t c βm ot β c - I r i S i on s . The s-j c a r c i t y structur e cs-f the oc : cr t - r i x c -r i- es-) conri so
S o the !βIl cw gra s- Sr r ej βnesco: ting the r - βgrs - u s- c. For l l i r t i ce - nβ i.e- i- a β. iso of t O i l s β
u cs-rn ero l ,s- ti j et icβ o - , scc βt c Kildall [ 1973], l Orc lna t β i β cc r [ i l β
~], and Allen and Cocke
[ 1 ( ( c - ] .
-inc cart use standard βGauss-ian c i .0 cs-S n H .100 S o c -i c c c l s - c β β: I,, β. c - 0 : 0 β s s - c t t β
β v s -n labI L e no-c1 re-sβs-- b - 0 c c , but it is n , βI , eo r d i c e - l u _i. to S cc_ k - β r i ci~i t s s-iJnc - -f
so ca-s. . 0 ty. The fiΓΌw graphs of many computer rogrcrn:: have a β~ cc lab
n β : - -r t , v called reducibilit y, which mean s S o c essence th: rt es--cry cyc β i c c ha:
a single entry o s -m t from the starting block of the 1β. J , β gn iu : o . Aflen [1 - I f - I l]
and Cocke [.1 - T I] ] fi nβs-i formulated this- n c -i .1 s-o r c βc f reduci bili t y , re - c c β s - nt re - I
an O(nm ) β t i m e algorithm I - test for n c - W i n bi l . i t- β,β , and used thij c i - c - c t in
an O (nrn ) βtime algor ithm t β, cr glob al βl ow analysis: OS β r c c o i n , i c β t i l e giβa~ h s .
- i~l)
- ββ - β β~-ββββ βββ --fβ’ β - . __ _, _ . β --β- .-_
β _*, -β-S. ββ . . .,,-~. - β - β, - -.s---. β ~~~~~~~~ β~~β~~~_
~~j
- _~~
_ β~~~~~r
ββ ~~~~~~~~~~~~~~~~
,~ ~~~~~~~~~~~~
β β -~~~~~~~
β β - - - β β -β~~~~~~~~~~~~~~~~~~
- β ββ β .
~~~~~~~
β-. - - -
Si β . i c - i c - f t mid ~~inan [1β 1c 11 I db ,.e---, ov er ,cci cur cβ c (n : . log n ) βtime test r
iβed ue .Lb , i .01 ty, cs-u .S cli ~~~~ ian [1013 ] combined with cle-s-β t βrβ u se-- ci β t β β 3 t r e e_ c
c , - β.βo os-j r Q (β c , ,i.og n ) β t cLcne meth od β βo n β g U t c c l iβio.Β°w s-ri ca Ly : c . lsn β. I c c β β s - o .f
[1 0 - 5 ] (U s~ c , βs-β elβ t β ,d a r a t I o - n di β -r u - cl c : :c . ho . co ,I 0 0 0 0 β g h bail o βIl , cn: u c β , , c - c - c :.
wl ni ch 1: 1(1cc log n) β 1 ,1 coc o s by Β°c re s ul t - - Iβ A,ho and UlLoan [l - - T7H .
and W β - g , -can c [ lr1 ,ctc ] di :β ccovered i s - on to usβs- βs a t in c -os - c m n - c β s - cβ i c - r i I . got y- oβ β.
- e - c -, - t Sr β s - so O ( s - r 1, -g n ) -time βdl . f - s - - i 1 5 c c . Taiyi an [ 1 - t .. - } gay-c - ,β. c β( on ~( . n ) I
β c ,L o :oo. β :i.g - s - - u Sc β , f n β. :t ins-g oβ ,βcoc I Li ,S oβ; , cu r-i bat-cr ,i , β c ,;
t h n Gr c d i s - ucoβ\ - l - ; so - βJg r .S 110β ! 1 - - run in - β- (or i ( c : , , n ) ) t ic:ccc - (Ta r t s -cd [1 0 β β~~~~
β I β ,
β I t - i β β t ( n β : , n ) i s - a : βc o n o . β : . β o β o c β i . i n c en s e -1 β - βo n , c β c β r β coc u cc c β s βw n c : t c oi n βI - O t i
- c β . : - β Os - c 0-Sa β.β.ching on S t ri c cgc .
Gui. i -s β c -β :-: βcs-ic y s-ni β s- β tr - β s-Ic r 0 rig: -,β . c s c ~ ~-:ed -os- f cOO c li β - c _ ct , 1 - s - c β, ci,
0 βs-s-c βs-, a Iβi n ,I t - . aβs1 Sco ,β.L~ - c , cmi 1-inc rn .I. s Oc βI,- 0 - c - s t - In s ii c β.cn βn β x - ce -u s- β -o: a
cont iguo us sub- i n - l o n g c-I β y . I t β r o i s the length of on curd n is- t h e
1 - - cog 1 Ic cβ t β y , t i c - c a a :tra βi gict o β - n one - s - - c_ n s-i n - c thm occ o β .l,ves this yrc βhle oc i n β .
-J (nm βc t is-~~
c,. 5~ iut }c . Ico ,c r β O β I : , mid l β s-β ct t ,t [1- I βl l cicβci sed an c ( n c β m) βtim e
β Ig β r O l l s - c c S - n s- S t o r n β t t - c b e - l t i g . Tin- - i l - β o. l g - r l t h m s o β βcβ~. .- I c i in c β s I t e - s o c on
- βr c - r c t Sn g ci - i o i toc :1 I βUC i, l i O β β c β - - β nβ s - - o c t i c c ,- β a c c - -gr : .c_β t o iβ t β C β . ! f O l :ne S h o e - s o d s c
I i i ts - c i f β , ciβ . i thi n . c i n - c s- c cci. t.l:~ - Lning :z c βts c βsr:rc t , er βh β β βc,- i cc r i β : o β : . β 0 β βso β - , β 1β 5 1
0 t - β - i s - β s t e j s 5 β t i n - so - -fr-c u :,. Boyer βur l βi001β dO [l~~( 5] I r e-I β β 5 C - 0 ls-c (βs- Oc
t e t S - - i β a i go c t β S S ho β c , wi c , c 1 c . aJ.. t h βugh i t m acc an O(n l m ~ run sr s- , i n n g S β β - O o o 0c c :
w c o r o c t βcase ( βr l i i u t . O c , Monr i so , mid l o O t [ i c β β , ] ) , Iβ c - q β n.i r β c - : - - o i l - 1β - .c ( n ( 1 βg ~- β
Inc on tb -- c;-β r o n ~- :βs- , wher e ci i s the cn _ L O c c - r i β β β :1 c c β .
A gββn - r - a i _ iz a t ion of h- c i t t i -ni c oc - s - βt n β i , ,i r cg r β I t β I s t , t, β β .0 n c - : Ui , i - β c f β :
- mis-on cont igi n , , , c c f l c c β I . r c i o s g i t β I W o β s - S βr i n g s x and y . The I c c c 1. - n o ,
ii!
- ---_ .--_β~~~~~~~ TU ~~~~~~~~ ~~~ β ~~~ .- . ββ-
,~ - β - ~~~~~ .- - -
cscon t c βOi i r ng algoni~thm nc OoiLl t - - c rc -- i above do i c - c t :~~o.- cc . 5 ~ cs-n 0 lY ~ U s . cc r - i - i β s - c o d .
Karp , Miller , and Roscnh s-ng [ Iico)71? (U- . crβ i b e - -β.,i cii i o( \ o c :~ I c β. log(ncc β n ) ) βt is-s-c-
ci.gor ,stbm c βs-or longest common sub strings β . lie - m en [1 T ,β- β J β,5J c c βcen t s-cS βcc
-s-c l β.1 s-- .l bbs -cc o. n . icig tree-c I n c a :s-cW wac t i , I c h s d . t bs-βe, t O n , i i β. s- I c : : ifl o] (cr s - c : c )
t ime- . u.uc l n- - : . gh r t - [ 1 - 7 ] Ins - i s c r c - y i β.L β_cct a :βt c u t ,i 0 βbc : β.s- ,l βno s-ton i c β I - .-an d _ c s - , -i β~~c t i , . -ys
- β t β Us-i s ciTg~- n β I t i t o s - .
Iβ s - n β - - i s 1 C s-y~-cβsn c n t : .
The t r c t rn g no ml β -n β,n- nt s : r i d s-cc: is to I t ct -c n β c c o l i o n - t h e - c . r o β,gIβ,β connected
Os- . ccy~βo orients - oiβ a given directed gra : Ii wI s-h i n cent i c e s - and 0cc edges.
Thi s i . nc-h le - s- : -s -c -u n _ c in f ind ing tic β i βs- - o s - β . β .β c i , -: c-1 cm- c s of a n o s - o βc y cc β c β.en n i c
osc s-itrix (Fors~H S n β : s -n d Yoie-r [ icc β7
] ), I ns - 1.5 c o - s t cog -_ cr g , ul ;β :ub β .c icc 0 β, c r c
and tras rzc i erit s t a t e s oft a Ma rko v el s - on i r, (Fox and Landy [1, β .- 0 : ] ) , an d. I n c
:β,I rd d.sin g the - so s o , cct ivi ty .- β t sc of a set of ermutcn d i cc,, ( tue -Kay and hn-gnβ:r β:r
[1 i7 14 J ) . lang- -li t and W e - s t - s -s--h-erg [1βo c βi. j gave s-in O(n ~β) - t t c n c cilgβ.βsrs- t SL - co.
tCunro [i βl l] de-cer ibed cs-n improved :cltgorlthm od.th a r βunning t i m e c f
O(r , log n + cs-i l- . Tar jan [ILβ β7 1] s-β c s:c βiit , cβ.βi, an oiβ (n s- β~- ooc βr β 5 . 1 : : . cc β β . . c t β rj β i - O n
which us- ,c-. c - c t i c β c i m t . stcarel c -ci i ci t βos-W c i cccl bc ,c β.ini. a c βs-rue i - crβsββ. β: .1 - c- s -c Ove
this : irs -hIβ : : . .
Pla n ar it y β 1 β. :S _c.~~~~
Let G be -a grap h . The ~βlan anity testing β n oble -cc . is to , dc,c t - n-r ccc i c c e-
whether cI can he th awn Iβ on ci plane s-c U s - c -cl mc- two e Igc .βc - βnβ c- :c .
Kura,k owskc l [ l c c β~O~ r os - - i - i- cI an elegant mathemat iccc,,l ch arocet - β r , i os -u t ion cl β
j lanaiβ gras-bc , showing th at a gra ph βI i s - planar Uβ and - βr βr1 y i t β IβS d os-es
not contai.n - -n cc of the on -- f r - l i hs β Ic - ~s-i go~ 5 . βΒ° c- cs- a go βIi β:-l β:i] - sβ s - n i c o s t 0 βs- β β c d I . .
Unfortun ately , Kuratow ocki β s.c co βrβ , I S c- ri - s - r n β - - cccβs: I c be us- βl c β s : : 5 β a c r :cc βt I cal
test for planar ity. l ion . ] c _ n o t - i β cnn - β! I art t βr [ 1) - 1] r β c c c , , - S an oils - t o r i thm
Li
~~-~~~~s - E - - β β-β-~ - - ~~~~~~~~~~ β
c - i c β cs- I n S e c I s β har n an i ty by tryin g to c -r cctruc s-t a planar repre sent ation
β o s-lie ~-l r- i~ i c . They gave no t_ cne bound for the algorithm, and their
cβ - βn tat i , βn cβ - s - cn t a ln _ c cnn error : the r ot - os-ed -i lgor lthm ma~ run f e - r e v s - n.
;, - i - i : t e in [1β β .- ~- J C βrre ctlb , Iβ βro s-su,lateci tin S.: algorithm, and Ghirey ] iΒ°- s - β . ]
βci β,β c u r 0(n ) - t io c . n β. implementation c-f it . Hopcroft s-ins-i Iar :an [1y72]
e- c - - i o ,loi c β . βJ β βs e - c β s o i c β βioβ:c c-cs-f ccβ. -cc _c a:cr.o-:-n i cice Jat os , β. ,- -cs c β β ,yβ ,
in an 0(ni I -s-g n ) β t .l cs-:e- b oss- lens -co st s - s - i -cc , β. i i : Ich i wa _ c las---β. r sl coco 1.1 fled ass- . :
t β . - 0( on β. (FIo ~ er:t βt and Tanj s-cn [ 1 β .β~L ] ) .
[Fi gure 5.1]
I- c o cc i - c l , I βlvn βin , -or nd C-.βJ βcrb-cj u β.βcn [ l9c-7 ] i resented β.nn-~ti ,-:r g ood β J g c n - i s - i . c - .,
w J β O n c a n t s-r v urn g an --cci lIe-it t icore bβ.. β.nis-J. Their (U gc -n it h r r s can eas ily S.βccβ
Lei r~cn U t to ~ac,cc in 0(nβ~ t i c s - n c . Booth and Luek er [ ii β - β] sin wed S. on
to cs -c - Sc e-c e I r i ~~~~ tree data st~~ cture in c_n 0 ( n ) -time it β .ss-c-r,t βs - t i - -n c-f
4 ,t n cs - alLgonitrnr .
t-Sa .ximimi Network Flow.
Let G be a directe d gras- h with two ci cis-t ingu is-he-c i vert Ices , a
source s and a sink t . For each edge e in G , iiβt c ( - ) be a
non-negativc -β necI-volued n o s-r i an oit y. A il .s-w f on .1 is- a no s - n -n ns - -g βo t c v c c
value f ( e ) on each edge such that , oβor all vert ices- v exce - β. β. s- - sir s-i
β. in β . total fl - ow .r i edges entering v is- equal to the t e - t a l Sβlow leaβs-l ag so
β5 1ccβ value of the flow is the total flow leaving s (whi β..o ic i-c c us-il. to -
tire :. - βs-t ,ai. fi-s-w entering t ) . The majd mimi network f_ ow o r iβ, βcβ¬c cβ , isβ t o
β S β t β . c r r r c l f l c β a f low f (-c β.β ) of maximum value satisfying :β(e ~ -β.z c (e βn t β- so
coil. - -c i gec c e
Classic work by Ford and ?u.lkers-on [lΒ°7kc2] produced an cs - leg - s - c l a lgor ithm
wh i ch au~ βnentc: flow along p aths. Unfortunately, in the wcs-rs c t case t i c e - βi r
1~2
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ -- - . --
I thm requ ir e-sc ~s- xj --
- i n s - - t n t, S al L i os -rnβ. , S i no c β ,cgl r it βso or β .β.: c c - ce - , . - ii c c c β , I β ,β a-c β s-i
in j -nactice . I sis-s- lids :sic - i Karo [ 1972], by using biβ :β..it ,i,- fi r : t search
to guide the ~:eles-βt,b - c c β . c - i β cc~ Oc r c β . o o t βS I c c β 5 0 β ~. :, s- r o o o . i cn ce β. i mi 0 ( c c c o s - β ) β t c c , - β
s-βan,i ation t β the S corJβi βlclk β .βr: . βo n algoril-lic : . m dcβ: e r n I e - o n β s 11 , , S o :i n βs S c [ l β T β( β.b ]
use d U rβe .q o I βS , Sr β 1 . 1 0β: cn ca rr β}i c las icci ] rocs -ve β.i a ni a ti rn ~: ooce t hn β -ds 1- - : lCOci CV OO mi
O(n βm~ t i c oce n- - a s s - , I . The b est s-nlg . βri βt Iuc- , sβ c - c βs-sm 0 β c s - s - β c β .~ βsβ so t h I s : s - n - i d l e - o s - is
Iiiβ,β to Kars-cinβ. -v [ iβ cβc β i c 1, tino .1 cs- is- s - n - : c - i β .β β so βnig ) r , c t s - io o - c - obt :nin an
.t(n ) time boujid.
Snβ . β . h Match icr f .
If I is an urncil nβ - cc t , e ct groin So . the- gras- h β-at-eisiscg vi- βt i es- c I s . : I. c c l a d
a oc-β.ax,i :cs-.soβ, o s - .s-cr i en β of βs-ige -cβ in 0 , os-c O w β ., - h a v in g a c s-nc cc c -c o d : l o t - . l.nch
a s- c ,β. S, of edge-c: l sβ.O a ro ,ax ,i ccouio. s-catching . ,-βcs 10::: cs - βf c to c t- . r c el βs-I β. β c n . n o , - O β ti β. l s-
o r S ie-nos- is i t_ c r e s t - i -I - - I 1 _ c r β . to to l j- o , s - -: l c-- goβs-ji m: . A gnβ:β .:S s Is I - i S cin βt- ,l t e I : β
it : -,ert .1 es-c s can be I β crS I S . .o β β r i c : J ,i c n c ,. t o n c β . βtc so β Sho rt flo ~β o.ig ββ cβ:fl r n s -β β .c β Ss-
tw- - v-crβS c-ce - : in the s- as-os-c: se t .
The iβ s - c cj r βt, o te gr :1: is matching s - - I Its - β::: can I cc t s - β c u o : : β - c s - - t o β,is - :c
1 i n c s - - - βn , r βt ,c rβs-e ci β. g ri thm m t . β a on e- t one -r I - c f l- , -rc s - - r e-is-len : i n ti c S o . β O c -cl,1 n β s -o s - c c
c c βo r S t b - c c iT- β - -r βn t β (Ford and Fulkerc β~,oβs [ i- s-s - I l l ] ) ; fe-o r such :c : i β c l , β β , i S i s - β
FordβF βu.l,β.oc βr: c s-n cc, .! gc r c t hin has an r ) ( r in os l S . c::e bound. U .s-Os-os [ 1 55 ccc β .ββ.!
n β o s r s - i t s 01β .βlg- -mc nlry to cs- Sc-L aIn e- s c : _ c β o i s - I a lTh,β βt i re c s - cs - cc cc, iβ β - , ncs -sI ,ied, the-
Olu ng ar l - i n :β: β-t i : - -t . Il - s - n cuβ,,- .c βi- and Karli [ i s - ,- ββ.β- ] s-s- se c . s-oβe- :β.111, s β β s ββ,r :s- - β c - r e S s
- - - β - 1 β: - β-βcc-I s- icc] r - - β:β .c . n u~ d,a βt - β .n g c o - c I S c - Ic β t - ach e-soc, is-s β( cβ. ccc βs c c c - :
The ir alg. - r I S-ic : i s es:s:c .βnt ,i , os . ly t I c s - β c β c,c,cβ ,e asβ i l l o c . I β., c - s- ( i-lv e- rn and β , -cinβ , S βto o
[ i n , β 5 ] ) .
Berge [ lβ)5β - ] and li -rn - r o n and Rab .i r β [ 1 1 5 β . β . ] r - so - - -1 5 I rc i l an β c n s t ~ - o β cβ , c i c o g
path method can solve the max imum m o n t e - h i cig r , - i s - l c βo - , c s -n β . non- i - i c βc ,r βt S t e g i β c c c i n .
- ~~~~~~~~~~~~~~ - β-~~~~~ ~~~~~~ , β - β -β
β -β-.
Sc n,- 9, a g e-s -i β~Lg r I t :50, 0 - , β , t , - t 5β 11 on Li - - β β s i c , ,, r iβ . - sr cl t :. : . Js-o u n s -βJ. [1-c 5 ]
.1s t - c -s-vole c , t β n c 5c i n i g i tus s-is-c - β..~s- , c b o r g c a β S i , , I n β . ββ, t - - g i v c β β . c - -J y s - c - β o c i β~l~
He s-βls-I ncβd βcc 0(n β - - c - c β c- -cocci , l b β -~~}c it I s - as - t i ,
t , β ,c t - s - c - --i - s- :. o . c . β -~~~~~
β n βi thm t c - ,c in j ( r s 2 m 0 t , so ,-,β . 5, - owl - -i- [1 β.7 - j
s- si s n i c c βon [ 1 1 β . ] ,r s - . s - - - o - β c s - i . - : . β~~ , c βsβ. β 0(r s-o ) β : ,~ . cs-c- β o ,, c or ,, t c n c : ,c
βIs-Li β ,, i ] , , β 5 - i c c - o : . I nn . 1, e βrcc l t n t - i 0,1 . l i t - c c : ~β I β I - c c ; βso c βis-cd E - β.,rβo cci βs S-h is - β
. co β . t cr - t n - c . - β - β - - s of 5c c ,! on t β . -n β~~~n , β s - c , , ( n n l 2 m lug n~ - t .~~c c β β β r tg s - t i m.
β - β s - UnIt c - c .
Let ~~~~~~ Sc β β fl ,U. Β° β s - in βt β β β 0 , , es-d o β.- n c t c ~, . c , ccg s-i si ngle
- 1.- n o n e - n t . 71cc .1 - - oc i cct set cr n l o c r . β β 1 - β - 0 β i c c s-c c arry -ut a . e ~dcX c ee DO
0] s- rat ,5 u r n s β - . 10 β ScI β . β.β 5 - _ c ,,l, on t , β - W c βIβm cs- : on t 1,e s - ct , ,
n - β t β - s - s -β , 1c β .~- t βnc β - os- s -u β. β. β . I the set c -r,ta i. o s i o n g o le - os - c e -nt x
ar _co : (~~. B ) : add all . β βc - - cs - β . crs -l . . - S β s-ct B to sot A (des β s-β Is- I s- f
:s-t B ) .
The- ojec β:β.tiorno c β c,r s-β t . c:,β β.-- β.,s-βiβ .ec! -β co t , βsnβlinie o ths-t iso , each ,i o r s tn c ,set ,I
mu st be so - -ccc i lc.β.t ed i c e - fo re the next fl o s s - c is known . Assum e ion e - o β .n β,β.β -.β o , I -s -r ice
Lb jc , tic - β. cβ mu βnc-cs - c .-f cs - I -:rβjt ,i βas.β c , cn t : s-j is-s exactly o r β c s-n . β - -or βo en-at is-s-n c
(so that β-j:βt- ,-r the last uni on all c.c i β -o:, c β r i t s β are in βnc.- set ) arid ocr ~β- n
I s - c t β .β.rrrj xeci f ind -s-j erat i. c-c: (if a ~,. n , soc -n e elementcc are never 1β,s- und βs .
al .I . c:r ar cs-I Fischer [ i c c 1 . 3 rnβs c s - s - ed an algori Lions - β, S β s-n tiβs- i s i r , s-bic ccco in
wh ich each se t - is- r e - c re - s . βt i t , c β .l by a tree. Each vertex s - i β the tree re~ r e -ce -n it :
β f l β: o l e -me - oc t β’ Tic β β r , .~- S of th i ~,c t i- o c r cont ai ns the a s-ins-ce - βI β the set,, ari d each
tre e vs --rS.ex hoi s - a ; -- ir i t- - r t - - it -s , ot in the tree. Ice Figure 5.lβ..
A fβS n-S - c - ri -- βI - β o os - βnt x is n - i β s ric ed by s :Lonrt ing at the vt --r i ccc rei resenting
x and 1β 1.] -w in g arc-n i p .1 nO r: until reaching the root of the
l h I s - so - s - S c o c t s- ci s c S - S c : β . 5 c e _ c β - β’ A wi ,s-i β s - e t c A and B It ,
1414
_ _ _ _ _ _ _ _ .~~~~~~~~~~~.
___________________________ ~~~~β-β - - -
emS β .rβcc β .-d by making the m β -o βS s.f the A t ree- the ouβs -β nn t of the r oot
- s- f the B tree .
[~~~gure 5 . 14 ]
Thi s- alg s-βii-ha requires 0(nm ) time in the we-s -- ct cc β.:β. , since an
βs-os - f - so t s-mat β. β .:- - m oe -I-smβ - βs-f unions can build up a tree cc -or: 1st β cog no β a single
It β-cog p ath . β I s -IL ls--so and Fi . s-- ei c e r- o:rβs-di f ied the union so cc - - i ciβ: is- i th cs-
l i~ wtrs -s-β on ce β : Iiβ B contains ocr - -re - -cJ β . c c βcen t s than A - then the so
c s-I β the B l c-a-c is made the I t -a r ennt - s - f the root of the A to β .c β co , ins-s t O n e
nrc ocs -re A i_ c cocoved t. the old r - cco β o , as-β B . lee Figur e 5 . 5 . This wβ.c gln i i - _ co
union heur i st i c , ,i- os-1 β r β -~-~ s the algcs-rith .β.o. considerab ly c lSail-s-r and Fischer
~r~s-vect an β~ (nn β1 ~-g n βS t , ,S ns-e- bound.
[Figure 5 .5]
0-ic! n-- i β v and 0-Sc cs-0β - cc (II. , Hopcr β - IL , cur d UlL l.c:cans- [I. - - β1~] o- t i l l e d the
t ic - I i- ros -ce-.Iur t - by allis-ag, a Is-crβs-s-r I s-ti ,c calied ~
c β.tβn s-o βs-ircc ro: ss- si :β n: a fte r a
f i n - i on elecco,βnt x , all vertices - -- n the pat h f r β0 cc X t the r -s-I are
made cs - SoS -u - et c f the iβ . - - S . 5βs-ce Figure 5. - . . Th:i s. in e - ncβ o:l c-es the βS . , ooc e of
a find by a constant fact- β so h u~ ooβ.ay save-β time on later βi rs - I: .
[ Figs-cnn: 5 . -
The s-et union algorith m with cc c i t -tn s-oors -i . re c.βI os-n is very cβs-c- β,,- t cc r o βs-g ro-is -,
but β.βer 0-β i- card to ru nal y s- β- . Fischer [1β T β~~~~ ~~~~
-ve d an ~(~~1/2) a: c-cr houn d
and an i,i (m log n) lower bound on the w βso c - i - -case running t i c s - n e of the alg β.rit io cβ:.
wi th path c βc. βaj r e-c - s-s- i -n but without w c c ig l n t eo i uni on. S ~ot ens_ca [1972] 15:0 r , βv c- s-c
the upj . er boun d ts-β- 0(m log n) and this - i c determined βSb β.: running ticc oe 5 - β
within a e- , oncs c t . a r c β S . e-acβt -so Iβ β r the case when m is -- O (n βΒ° β’ 5-11th S - d I r cc s-ct 1 β .
- β βr oβs re - s-s-ion and weighted un i on , the -alg or I thm is- even inar β.βi βs - - 5, c cus s-ilty β,
Fischer [i i r o β2] proved an O(m log log n) upper bound on the running tic: , β . .
145
_ _ _ _ _ _ _ _ _ _ I ~~~~~~~~~~ ~~~~~~~~~~~~ ~~~~~~~~~~~~~ -- ~~~~ - .
~~~~~~ β β~~~~~~~~~~~
β_- β~~
I !;.βn β - β c βt - u - c β.n J β . o -,s-r c β . [ i β β~~ l i :- : c c s - - s-β -:~n S b ~ u c c - βr s-c -s-cs - s - i t - ( c_ c L~ g n c
I, t innn β .c cs-
wine - cs -β log n = m i r i [i~~~l βg β. g ... 1-g n~~~1j . 5 ai β ,ia c o [S . c- c I a ] i co , i r .ovc n n
c-h e s - o: . -.cr Sc -s - s - i s - n s- β . 0 ( 0 β , i ( c c c . n ) ) , w O n- cs - β. - -~~c - , n ) is- s-c,
c βs-mcI i s-is-al i c -v - cr : - - ~t β A-: cs- - c oc os - s -- s - or β s : c n - , - s-- s-s-S ( A β zs - ec β:co ,io , - , ( 1 β.08 βS s-i--
as c l - IL on:.
F o r i .~ Β° s-i β. 0 t h e - 0βs-i cc -c t Loon A ( i . 5 ) ccc Sβ.c o β s n , - - : icy
~~~~. - ) A ( l . 0 ) = 0
= oβ~ t β o - -
-
= A ( L - β ,2 ~ :β
r I 1
,0~~j , j b = β . l o i β I. , A ( i - j β β l i ~ O β βiβ Si ~β I . 1 β~~β 2
( , . ) β ( - - , : s - β m ioc [ I A ( i , Ion/n S ~cig r c } .~~~~~
Thβ, : z ,o β,c:_ cn β- (to β . s - ( c o c , n n 1~ is a rather c- βc~ βsLβa~ β-o n c - c 01cr such -ci s i.cmjct e
βt s -,goo βithr , . ,, ) r c - β ββa~,β ona Β° si ~β s- ,s- . i oi: K ci β S , β S c - - c - j t j s- 0 ccβ,: s o s - -a s - los - . Is - m i s - c s - [lβ β.ββ - : β.j
:5c c-wi-I S S c a t b - c r - - oc r - c~ - r si _ c βj s- s-: in : t az β.s -- β: - s - t o , - sβ:t cs-c s β - n β s o r -cc- S -c -- c
wc , l - I- , r e c l o - - - ~( co-, x ( c c ,, rβs - S t i coc- ,β. ~s- , - - - s ss- t, β:- -βs b y 1 0 - , j a t l - c cβ 000J rβ:βs β n c
a βt g c-lo s- boo . ir s- 5 β s -ct β s - c : . i . n , k-s - j cβ ,cos -~ oβ , :: cc c i c t c n c c - n c - c i i iβ . - ; c , (o c c c ( β β . o c b )
t Jj cc -~ O n t I n : rs- β n β : t s- is- t solve the set un _ cu ~ s-β βIβ1e-cβ. (Tar β ccinβs [ li β D.Iβ r c~~: ,1s -ck β. β s β -s .c. β - c u , n β t , i s-n 15 . t c I , , .-n β - β- r :c in the n - c βS . - - β
-
For -my so- β at rc clO: c i c β . β so x . L XJ - I - β in S - - : t~L - β. gre-cc S - -: - t o O - βg- - so n t βirg- β nβthan x
14 β
F_.β ~ -_β-cβ~~~ .,~ -~~
β Tl~~lβ ,,,. ,,_~~,, , ~βββ~~~~~~~β -
β
s- β’ Future Dir e β .βt,i cons - .
The field or β combinatorial alg orIthm s is tos- βs-as βs- t s- β. β.- β:-~r S n -ci
βs i ngl e- m a s e r - β so even in a single ho ok. I haβs-c t r l - - -1 s-~ i Sa t cci.
sonic -f th e- - nos a , i -so r e - s c u l l s and u,o s-d -cr l y i rng s ole-as- I s - n l b S s 1 c d l , but I t _ c - nβ -
arβ- c e - n t a i r n l y many I coc i c β so s- ant resu lts I have had t -cccnr i. t . TSr m_c ,cc, o c o βcβ,uch
on cc co βs - .c cc ,tc- ci cs -a t β r c a i alg - r I t , s - s - cc-csβ is -a β Sc β .cβccn Ion s- , rcr u chn r ββc :, a ior : t he co-
in s - h i s cc βnc t ud i rig ce-ct β cc I onβs-oI l I l k -c t s - s~~ gest five- areas I -so c β s-cO o_crβ,
r s - - - β0 β i c . βi -β - ,β. ,βi: Sri wi c β. ch rc~ at i vely l i t tle work has been i- c β . - , - Lo ut β S n
a s , , - ::, i βs-s-c cβ- -w a s ,cc as-β s-o - t βs - n s - t i a .Ly great .
5) =
Ac-sw-sn ag nβc- β n i s c q:i o t i s -u o n _ c r s-i be a cβs-cs-, s-- s.β r β cc ii s - β SΒ° s- cISc βS oc cc
βS-r io - rj . A c t h u o .igis- ctca cβsy c c .β j le hav e at 5- -: rs-c~ ted t - :,s-l-s-- β.cc t0 r -I~~ cc rub - c r c _ c v- : rv
L i t t l e ~ oβ β1- 0, 55 ca. been coca-I s-c . It se - s -co o βs tr β.at n β s - c s - c :ca Β° r c - o n β i - ca is-
; c - β . - oi -s- .l ; t o n- - e v i s - S , β . r r - o - s-f S -a-c - -c, S l i t , ~ nd Sol s - - β-aj [11β 5] , cn 1 s - , - β-,β : t s- I nat
β S I β ., - s , o o β . a l i ss-a Β° - c , . the sc t β-,,o , , I ar-J tech~n I - ,1u- -- S_ -f ~ r s- vI ng o r - L I e - c - , o c - -.r t . c - a:β.
n t - be on - -r Iul β. o , c o-siβs S :h a s- cs x . it is eves-n cβ s -c - ceβ s-- cord e
β I S i s - = βI ~~ c i β _c~~~s-ton β - e - o c , β’ be- solved w ith in the fr- s- c . on o r β :, :β fβ . - r :o,a.
s - c - - S thnoβ , cr ,β ( i i a o β t c s - a r β .i s ar 1 5 -o n r c Sβt, [ 17 β 1 ) .
~s-i so - : , c - - o c - ro t β l y , wβ s β . on cs -n w a l to , sot nothing ab o ut the r I s - L i v e
0 . . 1- β . - β rccr, I , : l s ct i c and n . -a - 5 - β t , e r c c o n S s t i c β oci βo g βn i t b c β o s- , as-s- cl s-s-n β.1β. . the
r-c s-ti o rn so hi j β : -β β w- ,β~.c β t i cs-β and s~ acc as measure s c o ol β cot s-ic I t - - c c β t . y .
r- β ,c s - , b_ c is-i hi s area wo u l d be i tcrj β rt anβt . Recent l y, H β - cr :β , . l aos-I . an si
[i ββ .1 ] w- -r cβ able t - .c Ls - o c w that any cc s-m 1-utaLan βt- go β rlng 5 (n)
i c c - β s-r i a c β s-~βt t ita j- e Tur i tog machine can be carrIed cr u l - i n S ( n ) 11ag t (n βS
- ci - β . Thus, at- co ast 0 β -r :oou i t i ta ~ -s- Tu ning cn acclβ ; βl c o - c β, c~ o s - - i l : - a , - , βso - :
β. β c , , ia t I c - r ec s-n l rc e thaiβs- S t o , : . This is the -o βII.y such m sβs-II βS ciβs β-s--nYc.
147
_ _ _ $
-~~~. -- β’ β β
~~ β
~~~~~~ β .- - βi --
~~~~~~~~
β- β β - β - - r β β -ββ - - ββ β ββ βββ- cc~~~~~~ c -β ββ ~~- , β’ β
sr i - : β -s- c , 1 nβ a - i c to βs - h - - Γ§c = 7gs-s- ? question is t consi o βi-s- r c t - i β a articul ar
~~- nβs- β c- .c-s- 1s-Sc - j r 0 lI βs-cβs- os- c β s - c ,_c fOCi β .0 s-s-las: of alg r .I βItIs-r~: as-s- s-i. to .sl r- s -w that
-ccβ,β-: g, s-lg.βn - l tSs - β: Ia s - - i s i s - ,IItc :, lt βs-s-l cIt -O s re,βs-u ,Lr e c - c ,, c - re than iβolt n n , - c β . , n~~l t ics-c.
Resujt,s ~β this ki on d has-- - - been Octal s-r od for the β.catisfiabβiIity r 0:, err
( - a , 11 [ 1 β~β5 ] 5 , Icc- c βna - s β s - Icc _ c o o s t - _ c s - c I t e sos -c t u r s - b l e o c c (dOri s-tat [ i - c - - ] ) , and the
β.c,rc βt: S s c o t s - β β. rig c-β , : - t s - c β o s - (β- t , c c βin . c, i,t [lβ s- - β S .
-~~~ - t : o r βst _ c - Is i s -β t o - βos-s-s- i d- -c βn . ,-del .c s-iβ cor,cc u t a t c i s-i s - i c - - s o
O h βs-tiβs- iβs-s- ring βc a_ Sn c c - , . is-ic cc oss- s ibi l i ty is to s tuc β s~β t O β . β : s - I , -: , - O β B oo_ can
s-r β~~βo i ts . c r - - .β ccc -,0 ng 5c-o βI t s - o ur 0 00β s- - β - c c . . For re- s-s-Its in t o t , β βcs-so - - a ,
c - s- s - .Iaβs-βagβ [1- c-: - S . A r β- β, β~β. - - 1 c nn β.1:1 i s - o h - - β, , β.-β . c~~e d -ac :s- s - n s- β, . i s-c c S c o : c r β S t . ,
I.au,. ac-S :0 1 β s - co t S - o b t a I n Β° I c e l s o - . ββ β c s- ace i radβ.c- .β.I f r β c 4 u _ t . A so t s - c .Sc,β
5~o - β_ -.cas β_ - β i n β s- - os-s-id: β s - c c t i n - β ~ - -i s - r β l e gc .n_ββ c- s - I~~ a s- 1s- nβa o i s - i l β ,β 5~~
, c c c ; o S o c a t r I βci, 1 c ;- -s i ni . (vs - s I - nc , S [1 c-~51c , is-Β° β I ) .
Ave-rag β- .β u - s - c a y _si .
Alt- h aβ :, st t β β 5 - ciβ ,c β rc cc - βc n o l nβs βctβ.- s-nlaIt a s-I s - r l t irβ o ,oc ut s I β s β . t he
β - s o - -i β: . 1 s, β c - i c , t c-ti c - S .-β βt s- β β c o n g is-a : J - . β. β - o i β~β ββ cI- β - .os-i , - -.- ici c la β.β . iS . d1v~βr s-geβ s- s -- β s - s -
a r o a _ c y c i I. j β s- - s - c t l al, v , c s -~ - s ot a n t β s- s - ioβS , s - : - 0 β s - n . . The n-,:s:,c .I. S. so Di β c rc i Β°~s.β- a 1c , n
:, . . c βs s-- [1ββ - iJ s i - - - βc - β: s-ri ββ ,,s - c , i cnnci gcβ 0t 5 o β . s β - s - c s - co s-i .β s - βtoo t h s-iβ i lace i t - s o
- c βc- β r - ~~- - β o . β c s - : β β aβi β c , y β:~~s- -I β s o β s - o S ; al~~ s o c - Sr - c: , . Sy i r s- [ l β. r~~] has ce-v s - - c d
an 0( ns- (1, βc- s- 10 β acJ- βr - c~~ - t i nts - - a_ g o n β. i ilcts β - -n the alIlβ ia ,rc sh 0- O s - - s t
s - d c , 1r β - h t - - , wO c - cβs B s - β c d β s - c β s β , gi .c o : c - - r β , s - I s - -I -b y-c r [l s-~iβ ] m o d l t βI β ..-i I s-,)
β’ β, β : i s - i s i t, Sv e s- i s - s - s - ros s in o ) ( i β s I, cg n ) av er -i~lIβ : t .Lor β .c-. ichc c rr [l - . s- ( 1
s - s - c t. - i - -cr : β - ; q . O ( n log fl - on βs ac - er - -s - c- β . t i cs-ac S rβ an sc t I c β s- ci s - s -βus-βc ,β a ig - r . S c , β,.
β{a . c [ I . I β β ) , Doylβs- and dl v- o , :t [1 β - - ] , os - cr I - t ,- i β ;th and ~c , s β. c I r S , 4 s - , , [ β - ] I cc_ ce - -
a c - i - β . r c - I Lb - b -cBs -v i c r s-I β cβ s-c t, un β o n ; βs - βg - r i t . hj ccsc 0 β -so - - βs--e r os - I i r 1-an - l i t~,β
d β : t r i c o -,r β - s -in S . -tuc lc cs-c rc β work irs- t5r 0, c βs-β-sa i.β c c - - c - : .
14 β
-- β~~~~~~~~~~~~~~~~~~~~~~~ = . ~~~~~~~~~~~~~ β~~~
β β
Gill [ s c β ) J os - s - cd Rabin [ i s - -β - ] have ro coco s-ed dti s-ot~r -: βsβ k i ln- i II.β β βs- c - so β s-ge--
cs- s--c r:o- - -iei s - ct β s- - co o i s - o x i ty, in whoi ch the ~~~gor it-ho - . c : , c Iβss - - : us-c us - β rn~ cd . oc cβ~.
cis-oicess- . i- -is - β suc in an s-Ig:βs-βi ti _ c -cr, cc c - nc s - β coca β ,- be s-isle I c s-nv LSncil the -J:g -s - - i tb:.
curβs: βs β s-c - s t βs - i Sil o . β c βftβ Iβ -βIcs-βS c β I n c β S _ c o s- n d,βs - cct βc- f the I s - s - c uS, d.ictribut l . - c c , 1 c - c o s - s -Β° ,ns-:
the os -vs - crags - c is tcc l βc - .:cc s -r u βS. over the i r i s s-nt hut over Use o .c _ ccβ IL - .l, e cc - cc c i us-. - .0 , 1 - n-s c
the algcβ.rit h.βoc s- n i a S I i s - β e - s - o c c c l s-cs-, . Ai g, cs- βitbrns cc-c l s-t c on - t e c βUn ng r I o -c s-li t y-
( f l t r c s - c c β- β.n s- ca n s - i siolos-v uCj [1- β β, β β 1, n e - s ; - l r , lβ .~:- - } , c l c s - - i l s-in s-I - . -β u-st j cints-
(Rabin [l - β.iβ J β , and, hcs - :Ssl rr g ( C a r s - o r c e - s - s -i iβ;e~~c- βoan [ II. c- β~~~ ) wb,cich cs - so β, good
ion thi nβ c o o s - c c _ c .
C, c r rc t s - s - s - nt I c r s-c β s - c β s _ c s-s- r i co A15 rβ.I t l noo , lj β o βc c i s- c β βI fs
The ch, o , βs- - : cs-f -cs-c :-,,ig nβi t hccc 10 β so - cs-n β Ice s-- c,- , 5 c - s - β s - cc β s us- s- s - o s - 0 , so - s - ho- _ cc
acy s--~ t ,βnβc,, i c ro,n ,βs rn .i ng t o t s - ce . A sitβs- l~ β s-I c s-β I i . c : oos a ,y be h e l t e r - c β . l o st ers-βoβ.:s-i..I c s - s-c e - -β
ccci os -cs - i r oLl ccc: , - S o β s-n βs a c β , β.β ,β - βc: : 1, 1 - ββ β.β β. , s - βs - , L o β rd - O nc - . β,-.βs-t th a t β s-s βs, os -F - : yc c-s- t , :-tis-β
runs-s-i m g t I c - - ic-ci t a iargβ : so e- . r i : t cs - cs l I cc cs- s- t r . If tw- aig a: : Is- a s- - c the
βncytsoβo - - rβLi c f l ,o ccc β t ics - c e - -. Sc Or es -β, the s - - c-ct ant- r ao s - t s - os - βc c: ,as- g s--es--nβs the
cIalc-: b u c ,v-s - - β c c β I . - - . ββas-β. - cs-β.r- - βc βu.l βsocaIic,βc l c 0 β s- 0_ c- I - rcβ , c l c - c β - c c c c c s - s t s - u r lβ βs-βc β. β.β s- c - ,
and S m ,:- -c s- f l β: U-H on βs - - u - c , cs-l~~o -nβ l th~n_c W,βs-_cl ,oj be a β, rss - o O βcII c-ont r i S β-u t i - - n S
t rs-i~s-t.i c βs - i - - - βroncc ut ,I mng . lois-cL cs-,s - s - c i l , v s β t sβ f cc- so s-n cs-u s-c l - - s - i β of o r I -I ts -c - cc β s - c s - c β s- nβ ,
in s-Iccr oct , iβs- β so ic - s - c ob _ c ( i~~uth [ 1 β . - β - , 1ββ β β . . l βc - cβ t l ) . A recent _co:-c as-:rl le - -I 11.1:
so-s - -se arch is- Wβ iβ cβs i cy cr- aβs: [117β:], βw4, , ch - c - ,1 ccJ a, c . l ccs β s ie -c- : β : n ; i β n O c β s - l o s - o f β
c r . l - r l t y c- c s - - c-:: :β_ c c d , βji β , ββ sβts t b β s - S ~0 c - i I c , β o o β clocj, S β β . - - r , : r , , : β . : c S tn t .i cc , i _ β
S βc - s t in no , . I , -o - i r , c s - oc β.ss-t,an c - - c s - .
L w- βi r i s - so owβ¬cr Iβ . - cs-a ds-.
Oci sβ β . c-i cc isi S 5 . 1 β β . i s- Irs , onβs β o n - - c - i t t b β.. I β . β,irn - , i c β.rv S - - on -c c- s - c , s - n β . - βi. ato β. β - β s-nβ , s -c i n c β s so- c β S O i l . I
r - , u s-t i βs- is kit o~s-βo, -~n - β c i onβc , - β . 0 _c rβ -o β : , S S O , exiH Soi ~s- go- β o;
βII,β- - r l t , lr- c β βs - r β β iffiβj β β - , β Β± : - . Fsi c- β , r β β 5.1 . u g~ c s -sβt s β c o β s - - so β s-i S β s-n ;S o s- . 1. c ,i r s-g - 0 , . n - o s .
I
L .. -.
- -
~~~~~~~~
.
F.βββββ β - β β β β ,
~~~~
- - ,--,β-β β ,β ,β, ββ _-_. ~~~~~ _:cβ - .
~~~~~~~~~~~~~~~~~~~~~~~~ .-.-.β -β,,-β .,β-,
~Cc _ c c o cs - β. s o t cs -β: be onult i~ l ied βion less IcOC . -D (n β I I β . - β s- :cs-s n SOs- os-
J I :-:rβ-s- s- -,β iβ s-ui β I em triβns-Iβ, is -co c - β . cccrn l ut cd it . lessβ i- s i β ~ o O (rs- It s-g O s - ) U s - ββ?
Can cβ ,s-Lβc 1 : c , cscc , o r e - l w βtβi-s id-on : be c β ,-a c - s --. i on βi β s - s n β s - i s - - u s - β- ( a s-) 5, 1 - : , - β 1 o c s - β i i c - e - c s - o β
I on- cβ s-s -ur n-is -.c i d cβt s- fβ,cr ccl β.β ss- β β cw o cs-βus - I β, - , βs - n β .β? : as 5 . - s β s - i n s -c (iIs s- β.s t ? s - [ 1 - 3 ]
l i o n - i l . :c β.s- 5 0 , - c βs-β :s- βs aths ( c o - n , A s- i s . acn o h ive s- I [laI d J ) , d β o , I I s β .β 5~~t Jβ~ i - 00
I t s - r i --.r c [ i β β β~~] 1 , c-s - s - n -s-i ev - s-It s-cas -,i s-n _c β sovβconss - βs- t - r I c ~u,o c : s - I c - i s - ( , .It s β o β .::ec-s- I u-r 1β i ]
1 :-n nβs- , . 0 ros-~ - βrtic-s-s- -s-I d o s s- s-a c~tructucβs--c- c arid Basic S-Ic -cs - Is - c -d c β.β.
I βS β . s- so- c-I s-is - β βi β t -s -βc }ir ; is-s-s-_ s and s-15 -c r - c l l:b.βccs - - βs c l, l c - c β, s .in i 01-s-cβs - i β c-nr c, - ~. afl, .i 5
β t o β s - s i _ cβ , βs. ;-s- s- t i , c ,, tβ - s - s - i :β c β β - : , t - β d w I th a j r - c- - β- . Sc s- - ic βs_ cs βins - s - c
β .~s s . βs -. r s - n β s - a s- β. I s-icc s - - i t t _ c - - c f - iβ β c β -β Icβ s-d s , βs -os - a ββ - c a l c u l u s : c i β t o i t β t c r nβ s - c β - - : ββ
bc s - S I β S : -n c- - β s-, s - i r s - c c - s e s-s- c - β - cc: : s - r I - n t , -, i nc-os - c - u s- soβ. :cni t at i s -βn c β-s - c - S tc cβS β .oo β c s - n : .-
β nβ s-c giβ.ββ.s-oc o r i - -s-os,? Iβ,βb,s-st cs-,cs -s - ,-c s - .βs _ cs - s - :s - iac -.a ss - s - s-βs-n β . β β β ,nn β oβ.β. I s-- s- I - -so S o - s-c : s-to βs tIer
O r a c-β s - n - β β s - , o c on: ; I oca ti.ooi β? Tb ,β. cs-n win n- -β :.βu,β.t: s-cs- β:; βs- it Y s - _ c s - s-On , s-s- s- β , β,β . 0 ~~
S I r s- so1,β I c β s o b u s s - s- - c c . βTh i s -is - s - S o - c o . β :s- β.s- , ,i β ., c , ,s- s - ro s s - ch cn ,l βt s - c s -n gβ l n g
o - lu -s-s c βs-s -oβ βlc-g c-βs - s -~ s-aβ - β? β se s - βc in s-.1 g -nβ l ths-βlc s- O , 0t o β. βs lec-s ,i
5 β .)
L . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~ ~~~~~~~ ~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~ . . .~~
β~~~~~ β β β β ~~~~~~~~~βββ
.β β~β- β
~~~~~~~~~β
~~~~~~~~~~~~- -βr ββ β ~-r - β -
l s c β s β βcβsd . i x : β I β . s - n s - β i c s- cIt : ( t O s - c t β also l β s- soge [1 0-7 β . i β .,o:oc ;o , s- - s - nd Os-ni t β; [la _c 5 1 ,
- s - os - i l o s - s - c s -n: ] l β . c β . - l s - β . I ) .
A β ~β - . 1 s-i β.β.β (V , k~ is _ cs - n oβ , I - : r c β o o - s - I r c co cci , - βli: βc g o ~β a _ c β -~ It on β
β s - s - os -i ci ce - I i l -u β - o , β :. βs- i t - i r s - c r the ~s--iges s-s-cu o s - β : - nβ .: 5 a lr . β β, . on β
s-I β d i i d i c βs c s -S βs- - ec- βt i cs - -e s - (the 011- - cs- In i s n I n - - - , - S . - β- i l . iβ s- is β .,: e β βs ac : β ,s - β - - βs-on n - - n - s - s o - c
cβ s s -nβ s - S I : t i c β , s- βt v- c - l i ce , . ca se- so-- . r β - - c - ,- m ; t - ,coi - s-c ( v. βs- , \ ( β s i n e giβa ci :1:
s-unii nβ , c c t o - ,c, β . If ( - , - , w S is an s - s - n c β- β , β,β cβ._cβ, β.cc w :c,so-.s- cit : e n , d : , l c s S : o-_cβs-,c
cs-s-β. β s-β.t, β s-r c-. .c 0 0 0 . lb s- - βcoge (a β . w~ j β . - c s - s - s . f r β aβ t - β. ~ . ( s - f ~torJis - βuct β.
4 h e β-o βmg β β coa: s- β l β β.βs-i .β β S β s β s - : on β. s . , A r; s - β o s - : is- i β~~~~( l β , : ~~) : a c c n i gnβ - s- β .
of IT if ~s- β~~~~V and ?c. β C c E . I β ,i : cs s- s - r rn l oβ.g if \β = V . A gra ~h
c - IT β = (β - , β β .ii β ) i_ c ocr c. c. c β - - c c s- c - ,β.,1β β 0 β -1 i i β thor βs- Is or coo βs- s c i r c g fr -un :. β,β
cc β. V :coβ Sβs f ? o c , I β ( : - : . c - β 1 m E β I: β on_ cn O β,cl β,β I c β c-c = f(a- ) and 1β =
sβs-s-- sos-cs-c (s-β , β n β ) β . o E . -βI s-c . n ;-s -l βl β - s - s o - c β. :~ β _ c - s - a ? β c l _ c i f β the β- o - c : I o β . g i:
- c c c β -10- 0 β . A gc β c - ; -h Iβ = (-t β .. ii - is or gus - .- s-c - I H ed cs - nc s-Is-β~j~βs s -f 0 I c β
iso a sun s- 3m -c βs - i c of a has-s- - cβs - - r s i c βi c i β β -~β β s-iβ nIT
A . s - cβ s - c β s - s- 1 β cc aβ in C I_ c~ a c β. β c - s - c β . β o c c - .Hβ cs - . ig-s- .c (β,β .a- 2 (s-β . . s - β 1 .
Thi s- ~cc t-h ,.s- cβ s-i_c t . ,β- , n , β os- , l ; . β - s e βs - (s-β . . v 2 ) ~βc β i β β β
r βs~ ar, ,i -, - c - - c j s-e~-
V s- V β’ βc s - s-c O . β c- .v c c l d cβs-li. oc- tIβn eso - β -ig β: 5 s - cs - i - S βse-iβs - I - s - e s - . βiβ}s- β: β β.0 ? , is . ioc s - c i.e
j O V~ βaβ. -ici s t- βΒ° rβ.-βI - c c - c s- -c- i t s c c i l β βLy v1 βmn.I v , c i , - β Β° I s a
coβ s- i_ cs I : β v i = βor β’ Ti n βs- c- 1β o .c β . : s - β,β,: β.s - β I s - , ,β c ,ss-β c 0 β (0 - β - .id t i s - . s- β.βiβ cβ, i O
0 0 β(0 = (V. 1- , sos - c m 00 ,_ c c-. o β, , W ) β .β~~~~ β S l o e - n c - o s - h-: i f β.-~~~~ -~ c β . cc , i βt , O , e re - βu s - a
-: s -c - t b :βs-β βs- V - on i n βI
,lβ.o , c _c β s β . ll s-β e - β . c β βs -j g s o s - i ; i c β no β. β -n r n β - e t - - β s - β i i f β I - O U- on 5 - - β ? . β n β - os-ny βs- -iβs- u- cβ: t
βccβ : : s - i s - - c r β : n β , - , . A d i r - a c t s - c l ,s-- ,u ) β, S c sβ S rs - βso;~ l~- s-β c s - c s - , n I f S ? c c s - β c -β i s c_c
O cs-H n t o β 00β βUi~ β v e-rt β β- o c - s - c : - S s - c - -i β β,β - β n β t - - s - . ~~s-e cc n- c- s - cc β β. βoonios - c - - l c - - i ( ccl βsβ. ngly
β c os o β. β - c . βt - - S . oo c - β.g r βn : ic : -s -I c g r c-0 h ~~e c s - o i l - _ c o 0 0 : s - o c o u βt,ed s-
(sO c 0 , 1 01 ,, c - lu c - - , o m ccc i - n n c s - i t . , ) . A ~r a ;h i _ c s - , β ., β c β 1 0 β i t u - O i , 00 I-c o0 β .~s - c In
h j - L - _ci β . (w i S I r βs- s- t i c- . -: cc- s - β j o - iSO - β s - n - S c cdc -~β:s β :s-~β .in~ h e cu~~ec) s lb _ c t no
_ _ _ _ _ -~~ -
β~~~~~~~~~~~~
-~~
-
β r β β~ β β β β β β β β - β β ββ
s- β . 3 s-βs~ β.β~ int - βs - βs -.βct o βXs - β s- .βj t at a _cβ c oo β s ,βc , r s- c- s- sss -i βs -Ut .
~ ~~so β_ c , β T I s - a co~~~ect e-i , u s - s - i . r β . β s - e - , I ; βs βs-, I- Hg s-h cs- s- , 1 o s - I n s r-
IOc l T C l s - c _ c . In a ar c s- ,- t ue - re is a un iqu e :i ncri_:l e a d . b _ c t β s s - c s -s - i c - s-nay lai r aβsβ
, L l st βi u u - t a- - s - r I I c - s - s . A root ed, u n d i r ec t e d βs- r e-c (T . r ) i s a βs.n ~s - os - w it h a
c - i s - t i c β s i s - s i s l r -ud βs -s - rI s-s- sβs r called the s- . - β . A r-sou-tβ.cd, olir-uctes- tm_cur It: a
s-U s - - s - s - c t β s - s - i u - r c c - ; cc T with a uni que βs-- - r s - - s - x r cuci c s-bat
(i ) c -b . c -r e i s c- s - s-s-tb f β r- co r rs - onβ t - .β. ~~iy s-s- t b -_ cr one - so s-
(i t~ each v- s- r d βs--s - s - s o - c - - s t so ?n ~s- βc exactly ~ ne edge leading to it :
(iii) r h a _ c a - β iges leading t~.- i t .
ton1 β ro . βit - s - . i . u s-n β, s - i c-β - -: d - c , o t n β - s - -,s- ( T , r ) cs - s - cc be c us -ivcs -i~t s - co? i t s - I a n β s - s-.s- . βs,
s - I so -s - c t -_ c-i I s - - -s- ~~ c- diβsβ β β o t r i g c_ c c - s - i c cigβ.s- (v, w~ s-u tht s-t v is oβ cs-t a βS s-s-ed I n c
β t I c s - β n β β r to a
β βn a c - β β , - s - . s - i s o - s - - a - c - i βs c- β - . a v - c β-, - : , w I s a -de cs-cer i lia ,nct- of a s-βer t -_ c c-s
(on i s ~~ c- a c - c c - s - c t so O l β nβ ) ii β thc . iβ - β is a ~~at ic Os-s-s-cc, v I c β~β . A v- s -sot βs-s-s
aβ i _ c a chi ld s-f v (β: U Os -lone - os -crβ s-n t ciβ a β ) if (v. βniβ) is an es-ige in tI s-s- β .
Tb βs- c-, β - i s - .sI t βIni t i βcc n s- s βuxβtβ.s-crs -d t βs- c β 0 β - . . 1, cr β s d,I roo t s- - t n I s - u - c-co β00,1β ciirec β.cIscg tie-c :
-,u-igs-βc _ c s-f s-sb- β SOβ.β β- os-c rs-l , o β: . 1111β (0 sic a~~ rβs-Us-, a , c ascrc s - inβi:β Is-βs--c s-f β IT i s - a
r o o t β:d t rβ. β.: cβΓ±s -ci ch l c c _ c .β ; a n-cs - s-Ics g co ,o βs g βs - o s - ob of ( 0 .
A i β s - nβ s - i t iβ β O s -
~
βs
k
a - -H S i s- a cc c-llectio n -o f cubs -u - I-s - s- p β
. 1 β S cu e - ir s-oh - i l. U C ,, = o ; and fl Iβ , = c?s- j f ~~~~~~~ If ~β , 1 jβJ
o cr β s -c - t i i n n s β s- f C , j I s - c a r e - o I c s -β. s - o : ,βs- o β . c of ~~~ β (anI s-f βIs- a
c-s - βs-s- β: - c n i r c - g of ~ ) i f . f - s -n - all S~ c ~i , there is s- alone ,/ β, ~ ~c β s-u s - i s- I
-
52
- I _________________ - - β’ . ~~~~~~~~~~~~~~~~~~~~~~~~~ ,.~~., - , , ,
. , - βco,, ~~~~~
β . β - β ~~~~β β- - βrβ β β ~~~~~~~~~~
-. ~~~
β β~~~~~~β.β.s-β β β~~~r β β~~~~β β ~~~~~~β β . .β~~~~~~~
-β
s β s - r β .. β ic ~β - β5
β4. AcSic β.β . n s - c n a i c O β . i_s - β .βs -Β° j . β , βnn. :o H i h b s - s - r s - sc:h cs - i s - c i t d o a u - n β . iβ reβ.s-l-:oi β,OcIilo c c , β
tI cs - S, Β°,. s - b c s - c . β β . 11 ~~ s-~~ s-:
A. V . β - . . β. o f β ~~- c s - β s t . and 1. 5 - . ST1 I,o o na r n [l , r7β S; 1 . The De s- igro cro c i Analy s i s
s- u β C io c β .s - i , O U es - β AI s-s- n I - i n s - o n : , Addi - β co βI- ,β - . : s-c s- β , i~s-β.auI c og , 0 - Sacs .
A. - . di β.- - β_c rs - I T. 0. 1 0 β s - . a [1 s- 0 β I β J . β A cβcini oc sus-c: .i.cl .βt cs-n o ,β ,s- s-nβ i~ β iββ s-c βs- s- βr : .s- s- β tO n β
~ :-a β:β:Iβ s β -lβ u- - n o t β : . - , β β s - n a β s - c - β s- o s-a r s- β . . β β.V-l 2 . t β o s-onj ut . 1 , 3 1)5 -312 .
V . s- s -c s- - an βs- i C. S. U1],nan [1 1)1. β lb s - n - 111 , - i l o g : I βor n- o s - n - .;s- β I : β ie β s I t s - a
gr c -cr~βh :. - β c s - β . - 0 . Cβv- β-nt i β . β .-0s-rs- r,uβ s- β. A β S β :s- β , β: . s - n c Theory c β -β -c :s c uti s-r g, 177-iβ5.
Fβ . s - c . d1βs o ββn 1 -- - J 0 ~~ - , - β . s - o t i β β1 ! . o βc- βwβ Os - s - s - c β . ~:~
β β
~~β 511 (0? LAO (0ot i s-- βs- s I ,
Ic β . c β . ,-βs- l , IL β c , s-cc s- -d βT. Cβ - β Is-s-c- . β A 1β. s-cβ cβ β β.sβ , cj s-s-_ctc s-βs- fl,sion β -_s,a,. , I s : r β β c c β .
Cl . 2 β S 1. ,- . 113- β -i~~β..
s- I. A: βcr1 rs- s- s--,i , , . i j s - s - ks -oo : c [1 7~β . β β H βs- c β: 1_ cs - s- s -cs - β cβ s- β: i s - n s-uβ s-n I c c β β - β
I T s - b l o c H . I - l a t i n . , to s - s o s - c s - s o .
Ac -s- :bc un- i βs --so - c _ _ c s - i . 1. V. I s-riβ t - i β [i - - a l . βc -I On ios-l β.c ? i i c c g , g r s - s - c ? r s i t β s I i β s -β ~ i -ac , β
β0 Is- cβ :. 0 ._ c l - i - S - oh. 1: , 51 1 5_ I l C I .
1β.. c-β . os -a- is- i βs usβc cc s- , c I S . s-β
. β β u - s - I [1 c : 5 j ~ β hegul ,cr s-- -βciaβ s-s- I c r - c- a~~β1ied t cs-
j at l c - s -βi n s - c h β s n s - so i t - c , , β 2. Tc βs.. S-lath s-. Aj.-~-iUs . 15, lo It -I β - .
IT . c s - h βs- β . - O i l , an _ c , , . O_ clβ c-vc iy [1β c 0 β5 1. βRelatiβs-c i . : c s-I , i β,βc -i s - c c- i t β the I = --
- 1ue st - i c , , β 0l I1A~β1 - IIl β-non~ c- s c . I , ,
- . β . iIOcI . .s- :oo - uo [11 i5 βT ] . i~ β t c β .s-β , i -, 1 r oc- gsoa βs: βss-s-i rig, Princet - - - o n T i c - s I s-β s- βs- β:i l y IT nβ - : :
F r l c , c- - β , - β. : β . . S c .
1- - - n : - - r [ ,LΒ°~5 - β . 1 . β - βsr i t b βs- I , , i βoU- -gy s - f β the gem- t ic fβ I ncβ. : l c - β : c - - β s - s c - β . - , β
I r e-c. I, o~t-. Ac-ad. 51ci. (0. 5 .A . l s - 5 , 1β 0 -i-
1 . 5β - ci β s- ββ j i β s- I J . β(0~ io - S - l i - c -r β . βs-s in [lrβc.i In theos-β, . - , 5 i -c - β’ hat β’ Ic -: β c o o . nod .
- -. -
- β’ c - , ~ β.
- - -
, c β
-
.
~~~~~
. -
~~~~~~~~ _ _ _ _ _ _ _ ___ -
r
-
~~~~~~~~~~~~~~~
i.. s-s-c- e-s-βc-~~_c ( l β c - β f S j . s- I c β s- , , β , - β o s - -y βs -I β Gral l nss -c and, it .s- A β s o l i c r s - s -, n o . , , -, . I
t n β s -e i sa . - . - β?s-s - c 1 - 1 1 - - β , 1-hcw 7 - o s - k .
β’ ~? 1s-βsrs-I β s -c :β, , 55 . J. ~- βj, β s- s- s- es-β . and A. 5 - ,. Meyer [lβs-β.β( β . ] . β A nota on the
ce-ce-rags -- t i cβ s- to cs-i βs-fl int e 5 i βans-Jt ive :Ir -:β:βsos-β - s- c , β βi -ru -ha l cal flβ.βc s- c . r cto βras-ccβ β0 β ,
~~.s β c - s c - S s - s - _ s i- C s- c - cs - s s - n t - - s - s- Cci u -c - s - u - s - c, 1 - 1 c c - c c . . f n osd - s-i β Tβ.βuβ? _ c c . l c~~,s- .
S- S . h Is-s-s . . . : t β : , : β , βi. 1 cr-s-itt , R . Riβ.β- , . s - . s-s - i-,i c β . is-a lan [ i d β s -3 , . β100β s-: b - u indc
o crβ ,:s-βl - s - s - c t l - -cr , - β J~ Lβc-ir oson -uter ail s- ,iy c s- s- cβ . β- c l - , ccu -β s - ,β 7 .
ci. S. so s - O s - u - n d b ,0 la s-do-s - iβ [io β~~β. βj . β i u - c t c i n β . g c - c - - s - is βs-- s-c - n , :-s-cu llβ;- s- s - , s - ,u5
c s - β . c β s - s - βt u - . ,i o n t - , raβ s _ l . g, βs. βo t o }i c , s-_c βs - i t βs-βcii h s - I s - β,r βls - β: u U n β n c- β 1 β ,,
βdIg . βsβi βi hsoc . c. -- s-. s- s - _ c its-c βsβ o c r β, β . β1, :s- - c cc , tβ- c i e n r u - -s-s lβcl- . β βs -β SβI s- β5.
A. h-s - r e - i ,. to :.cc~ s-i I . β - I s - u - o s - β [1 04 βi5] . βIllβ .- -, βo s- 0005 βsil os-t i os - s - H. Corn1 β1c~~~ty ac- I AOgebrβd s-
oc_cc ..i l b cβ .- βr i o I s-β: lc Ls -.βs - c c , A.o c , β.:ri c cc,o s o βIl :s-βs-βIβ.:s-β , βis β c β s- s- rb .
- β’ - βon β s-β c_ c β s _c J . , β
. -Io- - i .β-, c [ 1 s - ,7 5 1 . β c - n f β o s - s - t c t r s - c c s-I s :- ,o - c s - _ c β e ? o I β s-g βH d c c I β I β s c L β : -, β
u_cs- c us-c . . os is-β:Β°.r c β β c c , β .n :s-c lβi i t,. , β : β s-, cr , f β c β s i-I-s-search ]Insto i tut β.s-- c-_cβsJ Iβ.Is-:βsβ, ,β: is-il-
~~to Re - - s - s - s - s - I . C e n t - - o β .
? ,β.r β- βs~s- , [ IT. - !. β S . β Tb- - - s - s - ,_cJLI,β ,. I s β β so, β a s r a ct ica l ou r s - I nearl y oon tico s - s- s-Ic - , IT r ic:r.S . ty
c - n - - s β - , β . 1 . t i c s - c l β . CTs -βO β c β - β β βi β c β β β - β -
, (βcocc i rs- t er Ccience c - o s - . , o βt β s - β.βs β s- -s- βd
I 1 ; .
β’ (0. Lu . a u - k _ c r _ c s - c - o s - βI t . β, , . , βcs - s - s - 0 , β .β [ 1- β 5 ] . Fini te Gr rs -1βhs- and I c - s - l w . nc-u β: An
I c β c l ,r βs--s-i βac t io nβs ws- th A n β s IL l s-- c - t i - c , s - , SScGs- βaw- }I il l - Cs-ow cl sob s - , In . - β - β1
βso [1 β 7) j . β~~c-β- - i β β, c l n c c - i g e n c c ? s - c c f t des- Inbegrif i βec aller n - s - u - I - s - o s
- c-_ i s-β -n - r β β s - , S coh- .o β . -I s-s -ls - .l - β c - c , β βo so- .1 no- - β _s-cgew. i-Sat In. β7βs- , 25 βs β β
- O~~ βu βs β :L c - βs : .β . . . I β - c ,, s - _ c s - _ c - . β- c -cl,- - so [1βY7 β ) . -- s--cβ acc βs-nl ,Ia i. - s - s - s -- - . cocβsβoβi uβ.β.s- s-s-
nβs - i c βS s--ross β I so I cβtso l t r o t S cs -c - c o o c βcct ~c-r -s-n t .c n t i v β s c β. -βsc ocr , 51β r i o - : 0 lβ s- L nod o s β s -soy
n- - s - s β c- - , β r -~ s -β . β 1g b: s- β .c , c. uc dI Atβs-S ~~niβ. s - c rc βbc - uβsβ; 0β (βens - ,; s-g Oss - β. ~
~~~~~~~~~ , - - .-- _ -__:-s-- _β -
_.:~~~~~~~
β --~~~~~
β β. β~_ ._ β,.-β. . _ β _ β .- β -
1 . - , β c- .s-βc s- s- so β s _ cs - s - I β β. I , . β.βs- β β c -T Oβ uo [ l βs ( β :] . β n i s - c - .i β : β . - c - β . al s- la . s-ce , β has? ,
β β , uc - .βt . I - I n s . β o r - β . b i t ! , ACM C~~ ss- . Os-c ~l~bs - -my s - I β Coi cs -s ul , olg, β -s- s-c-c c βs - c - I β .
s- β IToc - βiβ. : t s - co n s-s-id : . . . Iβ s - s - H β s - n : [ 1s-)β0h 1 . β hβ nβsd l o s - g , c - .,i c o l o : s - s - t β. - IT β_ c d , j n s~s- I f - - c : . β
- β S . c- c-il ,J . Cc β . no ,~ sit . β’ β.β , C β
s - c - . f )n ~~s-β s - c - [ iβ c - 3c - β J . β Ar c un: s-βivable r e - f e c : of β .c l - s - c sn s - s - o nβ., , s--_i.β: s- r io s- os - Iβ - 1, - s - β , β
--
c - _ cs - n _ c r i e s - c - bI o s - t b. 5β c , ~ 1~5
c β ix β ital [1-,~ s-, . β i c - c - - βu c o t n n , i . nj g t i ne s tcs -is i .I l . i ty s- 1U :β, , n o : r O s β s-n s- βiβ . o , , , β
βi- .βcβinn l c β s-βl i s - - cs-sot. C iβt ( 0bβC β β 7 β β β - β , s-a: β : s - nt- -s -so Cc.i ence Dc - s - c c ., β h _ c o O β n - - n
U β A s - s - k s - [ l β β o J . β(011 ,-i-al s-cβ cc c s -os -s- c s β u i - - c - , : s- β :c .L- ,β-is- _c - I i c : c inson t i n , . β C i
s-~s- s - O s - u ) . ,. β
c . (βc- - s-ic [iβ.c-β iJ. β βT i c β.: Lβ, β Os , s I T c - β X l l Iβ βH β l b s - - o r e - c o o : n β - βs-β In g o r -c -cc - - o s - c l - u . . . β
fb i β.β0L5 , :n - β .,f I s - c _ S-I ,Ms-:,s . s-,-n lOs - cs - s -so s- I β iβ onss-; u I i l ,g . . . . I β
.1 . , βs- . (βs-c-sc-k o s - c s - c - I , . - . Reckh s [ l oβ (so 0 βII ccc e βb β u.n s s-ic -β .s-t Iβ _c,is - ,S :β. s- - s-β . - . s c c c β . - β:, β : 0 - , . --
s- c o n s - - s- and s- βu : t u βoco C c - c i s c -cu - c βlβ~
(0 . β β. LI-. s- O , β , ss - s - β . : , J . ,β,. β J u t s - -c - u s - β- ] . β s - _ c o algori t-Oβs-o I s- u s - β cβ s - r c i β . i cn β . s -
u - c _i s -s -s- i βs cc βs-i -n on β cs-s-tooβs 1βs- os- β - sr i - n β c - - c l _ c . , β I-lo s - l i ,. β- c -nβ , :. I s - -. 2 βSO~~~Ol.
β’ .1. - β -r c- roβ ,l c s - s - s - i c- . C. hs -s-_ 1i - βtc [is- -β β 1 β’ β . β c-βs-c- s - c - s - s -β s - l u - i - s - n o t algoiβ I t lon: o iβs- c- so
c-~Iβcii β i s~~~. : . o βo - β 2. 1-11 JL~~. ,,β . β 1,~
fβ Utl r, i 11 S L JSs - 0 βI . S- lcKc βe [1β β - - j . β I s-β β s - ic -,s-cing t.hs-s- h as -i dwβi It I c of s- s- 1 cs-l βs βs-
, βs - ,T~ , s-s- 5 r i C :β s: s - S Iβ, i c β s - s , β s - s o - -c β . - ~ t ,Ic hat β s - c - c - s -i. C - - c o . ;c β I . I , 155ββ ] -
. . Dahi , - . β s . I i j k s t r β c . and C . A . R. Ho~~e [i s -It β ] . β. βt s- βoi s-βd s-s-iβo,i
I ro β .gn β ocs - - β .cc s - tn β.g, Ac,oc ,s - β. l t - n , β,l s - - J r - c s - :, SleW βc - r b .
c β s- s-C s o s - I [ 11β β.β 1] . β..1 s s - c ,ββ~~ i s - β - β ~~r βs - :cns -tong cs-n d !βlx Β°, - , - oo: . β in s , i - l o ss - s - s - . n o
i T h s - v β β s o , itβ; ! s- s -a ss - . t c c - β β 1, - c , Si . (0.
55
.~~~~~~~~~~~~ s-- ~~~~~~~~~~~~~~~ - ~~~~~~~~~~~~~~ ~~~~~~~ - - - i,~ s- - β β’ - β
DO s -β, , 1 , . 7. ββ s - c β d i β s- 5 β . β7 β s- β . u- cs -i β β. f l O b I O : . fl [ 1 i 7 c ] .
Di βs -β ha_ cit o Ti s -u - s - s - s - s - u s -. t I c - c n n _ c : o u s- i , s -~ _ c β . s - no β ) s- s -Ct βs β β _ c ci 0 5 β β nlT~s- βs cr βs-cβ
s . -l s- n t. c s - . , β β - ls - : t , ?s - s -- s - o , c c s - - I T - βc H, c s -. c v β . s - l - s - ; tc:s - βn βst . s Ps-ri c i c -g n r a : i i i t I β β.βsrt I r . cβ I l c - - o :,
I . F. l I , S k β i c - β o s -- [I - c - C s - c ] . β A s - s to β.-n s- t β ,s c c βc-.c- l_ c-s-n β.,c βln -s - : -o r r c u - c-o I c . β,-dt β !s c - - b ,
1.
β1. l !j i s :t β - _ cβs - s- [1β s- c β ) . ββ b01β:.βI IT T I i O c - c o r β i r e-grcj s-c :s-: s-in β.g, 5 β s β c n c t , l - β. β β β c
. β o c I T β, -- ,-; , β .β .s 11L1 βI c , I I .
c- On . I c .l c o β s-- [ I s - l -βu ) . β ,.TJ .g β lβi t !l: cO C c- c i r s - .I a s-iβ s-n iβ s- is-leoβ, ol s- : . c _c-: . I :β:ur β ,
β β e - c - w I c - n c s - β . s - t s - s -β βs-β iβs β:I L l c : , βwc -s- s-β c - β : l , i o c , c c βt i , . .β , 0 s-βsβ. β β s - o I l s . 5 ,c - s- 50.. 11.
liItβ-lDlβi).
- . is -n e-IT s-c β β s - _ c I t . - . i s - ions -s - c t [1βββ- ] . β L,β c : β βs- u-s-cβs c - - c β. s - s -I t: β to βs- s - Os- βr sicβs ;
β.βΒ° s-- c βs β n t i cs- c oags- c r β I β l β .s .β Ic - ifs - c. I c β - . L e s-. t , -:~-c 5~
ii, _ ] S ,
.1. Us-i s- β!β ~i, 1 . A cUr~-β .s-~f c - f s-βc ciJiβSβ:c β , cs - t c - β i :-c s-β β , e β s - r - β? , . β Tuβosc ? r o , I u - β s-u i c - s - c s-nβ β. .
β β f l Ls- o o cr: u βS oc - :βcrβ , β s- ,i, , :nn s -β s - i o s -d , β,βc ts - .c :o c lI o ,Dc-d s i , . β.βs- , βd-l , , β . s β- c I β S .
I :nr li u- β,β [I s- s- c t ] . β Ic β β c β ββ I c s- ,I β β o β . l βsβ - , - cc β. , β:o- o l β : β nβ c- .β - β β, c s - nβ tn ~ o s - i .g- i βI
- cβ .β - IbI s - b β I β-β .
n. [1β .β 5 ] . β1 β c - t i c s . c- s-β - :s -βc . βs-c - i o β I . βc βs-: β:n - , ,β 2βs-s - c - s . d . bs- rt h . ~. , 1, 1 , β 5
s- s β ,jioβs ,Β° rn . L,β βc_cc - x. ~- β . l a i n [ l c r βD } . ββTb- - c-β .t β.s-i c ai J, c . o n β βsβ -:oβ , - , βc nl . oc i n , β .l β β, βlco i ,U - β
β c -_c ,β β o s - β 5β so 0 0 - - s-- s i n β S β , βIT , on I s- - i - i c - co o : . -- - 0 . c( βSs - 1 β , Sc -~~_ 5 ,
βiflOl . β β c c- I - i j l - I . β - c - n βs ti re eve-h o Is-n :β s-β cs-oβsdsc-coβ Iβ s - -s - i s - .
I Ts-Cb s - s- so _ c c . c-s-s-i-sad . b-b c s - . i l - n t . d c i . K50 .1. β . I β β ββ 1.
- . β -~rs-s - n - . 5β s- c s - I . βic- c o t , A. I i i β s - i s o [is -β -β . t , iiiββ c cβr , I-i ex.II I O s - t _ : β. 5 s - s - : l β s - β s-n.l
oc ,uj , t ,5 . -:occ β.s-. -i .I β . Ic-β .1l, -s-~ IT so β i - i β s - - c , β , I ] ,AJ: , c~ Cs - cβ s - (0,. 5 . )~~~
_ -
-
-
~~~~~~
β’
~~~~~~~~~~~~~~~~ . - ~~s - s - , ,~~~~~~~~ -
:~s-β . β c o o s -_ c - c - . II β s-nβ ,i v [ β S o c - b ] . β An β s ( r β s β~~ ) al s-~s- nβ ,I t . T u - . β β - o r β O: s-oiolc c,s-lc β, s- sβ s- c- s - s - i n _ c s - H
β i s - s g c - β n c s -βc - β s -ri , c- aβ β s - c 1-n: , - β 5β - s - n s β. fl u - es - - i βs - c , S c , : : I β tic -s-_ c _ cs - c - _ cβ s-IT β~~cn s- β’ - - c o o
β-000 o c - o - s - t β - c - nc β- c β Cno : ,o-u t s- r Cc βis - β c ss -βo- . 1,1β β t ,L β
βv - o s s - u - i s -i srr , ian [ 1 - 5 1 . β s ; - s - t s- - s- s - -rb IβLs - β .w s-s-,c -rs -o tes-t.1 os -s~ grc.ns- ci
s- β s - c ouie c β. ,bj i t- β, β . β ~~fl ββ, , -. s - s -s - β sO5 βs-~ s-. ~~~ c- β,Ai .
i i occheso [1 s - I β 1. β i β , s - β Lβiu-I- :fle -s- β of - s - -,cuj .vo a , , en ns - c - s - s-i_g oon t Β°e-c βs - , β , β Iβ s-cc-c , : lu-c ol ic:β .β
os- C is-ini s-ole - !β - .βs-o βnoβn i u t β s - , ,β - n β s , , - . . S - βi. βS.J β.,cr u - is -i - β..
βs- -I . Tiβs β β.- o s - I t β s- β , u - c : . .
5β leoβs-urn s-- c - : β c . h ew c - .rc - -c , 153-i β
1- i . J . !βoT s-ccI βs s - .s - β as-nJ A. is- . β - β s - b c βs- β [ 1 β s - i ] . β is- - . β c l c - c c β s r r c- a s - n b c - c β ,ud l i c l o i c-,β.β . l i oβn c c - c c - . i
t c- βa in cit i β.βe C l - :u_cβ .- , β β0 . 0 β . Re - β c , IEi β ,i βl hil t-h s- βc-c,O: n β s-β , , β . β-a
os-wi, s - c_u .s - c o D c s -,os - -.n Aβs-i βs- -cβ s- oh a βilbeor βs-β , 0 - β . 1 β βIt .
β-1 . . i, cc ,βLβ . i , n , [1β 1- ] . ββ - βs-nc s - n β cβsβ : βc β c - c o , β , βc-J , c cs-oc os - ,n s -βs- >c J ic,- β s-
i βrs - c ,β i β r t c - β β cβ β c- , n β ,i t i βs _ c c- β β i s -β. β I n β β, β s - - cs - t SIA n β βI u - c }u β s I C~~ b - bc - βoss- β c -β _c s β . iuβ ~~~~~ I - Ia::.
s - n : _ β β.β- c β l - ; β h s - c - .5 - s- β
Oil. i l -ye- [1 β s- c - C l . β c -s -hg- ri βtββo n .c:s- e - β β. :! o r βt β s - _ βt s β s - i n , ββ Cββso β s:,. A- I β l l 5 .
i-S. β-5. Fl β β 0 [ i s - - ~~~ β . ,,βc- s : i0~n i i n s - tcs- cβ . c- β c s - 0 s - s -os - {5 c t s - s - i c - β s - - s - s-β . :, β s- I s - s - t ? c - s -β β s - s -_I - c o b
, s - s -l c-r,, : c - ,n t - βiβ β c- - i β s - Once , ( 0 . 1. , β c- βlni β .βoirt oβs , -c- c ., A c - - n β . β-I c c - : In . C s - s - c . .
I n - s- s - i s - - . cc: cβ , F . I . . i - - βd.
c- . R. ba r s-i β .io i 1 Β° . R. ?-βnni lβo c- βs-β s β - - βs-r i [ 1 - 2 1 . Fl -ow: l n o βb o β -;. -s - β !-c ,β , l i β in c- ,c t , βn
o b l β , , c c - : β 1β .β 1 :: , IT r , L o , c - s --H - 0 : , I t . . 1.
I. Fβ- -r - ,ββc - ,he a s - s - I C . ~s - s - βS β . - r [ 1β β β] . Ccicc s; cnt er Cyst s - c - - s - f iβ.i s-n s-β. - c-,r AJ gc -c iβro s-i. e
_______ β -c - U I , t β.nglβs-w - βoβ.l CiiI β~βc , S-I . -
~βc-x βun-β 1). β1. c- -β s-o s - s - 5 c -β [1 βs-0 1. - β β - s - . β .IC:crjthrn c β s - s - I β i s - i c - s - c o t ,! c βs- I s - ig t i n β s - c - - n - cu
cc _ c l β βf l β s - I s - s -s βs- u - I s o β s - . . i - β c , c- :1 a t - c s - - s - βi :1 - u - h o c - s t - S c cncs - t -r i . βs, β β β o s - β . β. β β β - 0
- s- i - - C l .
β’ . . . I - n β s- - c - L β:: ,n - s- [ 11 , β 1. β I s - i _ s i c - o s - s - s - i s - i s - βc - s - n u n β s - s-βs- , c c , 1βs-s_s i I ,β s-c β t i c _ c . n n β s - - - , 0 s - s - s - c c
IT c- : . - β , - β β 0 s-0s-J~5 It . β β c - n c - n u t . 5 , β 5 β s - ,~_~~c-,_c
o n . I . . O s - t i c - on [ i s - i β s- j . β As - s c_ β!β!β,I β I - - βr b β s 1-,on s - c - s - n i I s - n t i s - - r n s- I s-Jo , - los- i , β o s-J~~ c β . s-b c: ,
βcc i β c β s - u - b o o s - u : β c , β β.n c ! s - i . u~5β -i n β.β~~cso - l s - s . β . . , β ,.5c- . 1 β , β I β . β’
U o~~s-1l. l [ i s -Y c β5 ] . β β i n ti n e- β s o J β l t o i t I ,s- c- s -n d c βc ci c β s - s - c i t y s - ct β L -- β βs-u: -,r - .β s- ,s- c - β : 1s-~c;
IT o β s- - ,β- - -β-,- - β n o 0 ,11 P _ co n s - no b , β - β 1 - β - ~βic nj β’ β s -c βI de -cry s-f β s. , β β s , β
A. 1 - s-b i e r β c s - s - s - f -b . C . Iβll β s- i s - s - c -so [i s - c U . β c-s-_ cs- I n - .; s o s - v s -s- i β .s - β~U 5 β V s - β ., L β β c - 0 o β c - β cc_c s- s- _cl β β,β
- ,, . s - β . s - . β . I 3 01~~1β -
H . ; β ss - c - -- β, s_ c β s - i . . C. (0 sic _c on [ i i β . β ) . β A; - c - β - β s - c t .0cc -n β s - o s -- s- βIth o:s- . 1β
s- βoc:nis- oi nr β tu - s-β,l.β.ai~~ r -oi _ s -β:cos : -s -s -n c β _ c u - n β. β O eβs-i h u h Hβ s - c ob β .β , β s- βs-i s-I . n β L t r L i c s s -,β β_ c β , n
clβ ββcccs .βl, c -c s -β c ,b βt, βs- : I t - - on I s - β !β β β C β β 0 β s - s - c s - i oβ , c - β s-O s-OβOt. iβs- s- s-OU,l t. .Oβ, - . F. s - s - s-c c . c - c e - . ,
I - - i s o β s - - . β,- s-c - - β - , - β 1β , β’
0Β° . ;OL 1- ey. D . , β . C s - - !u- βs: s~. cc - i n n L. J. It β . s- β i-s -c , c-s- β,β ββ s - β [ i _ s - -s-- i . β β os - cc β., :1:,.: H o b o
II I β ccβ . β β β β - β βs-uc5 I c s- 1 1 s - β . . β Tβc c - - c c - s o s - t I c s - I - ββc:β : s - it en β - s- c l - s - n c _ c u - 1,
I- β . R. d c c βsβ s-β , s - . ,J H is-nβs - c c , . c _ cc - c - i β -~~ ,~~ cs - iβ os-j r [ 1 - β, β - j . βdc - - o l β s~ .cnr
C s - cs -o s - i l l c l - c s - c circu I t so -I s - I t s- . H I I I c_β β : ,i . β 5 β . - β - s-Aβ β
~~~~~β s - - a β’ s- ,
.
A. b c . r~~_c [1 s- β3 j . β h β . β. ,c i . s-β s- . n d i :c β β s- H i s-s- iβ a c - s - -c- .s-IT β s-_cβ β I c - I t s - - s-. β .IT . - β c cos - c s - c - t os - c - s - sb . --
βlA S-I - . n T is-u se r . Arc-al. 1:, ~5 β 5 , βs - _ β- -
U T. bi ll, I I I [1 s - β C } . β t βon c n : u t o β c-ti on βs s-~ , C c c -os - c -~ iu-X~~iβs β u~β u c - β c o o t iii:β I s - β I β ouβioos-
o βsoc - s -βI c i s-n - c c , β T n - c - s - c . Ci s - o t i s - s-~~ o ns - ioJ I- s-b-i ~~~~~~ on βth e- -n_cβ . 0β C s-to s s - u t oloig , β1β β β5 .
l i -Cl [1β.β i ] . β I . - t β o β n _ c - - cIt , u o β . - c , β, , , o β i o s - β I J I . -uβ c - c β O s - : : , J o n β I i β I n s - s - l s - ,I c β . I - I o c t ! , β c β c β oc - ,t - ,i . β cs-
s-i n - i β~βs-r βs-:β _ co s β - - n - , β y. t erne β !, β ββ - n , - β . l , i , c- , s β t - c 5~ β 1- I . I c n t j o c - β o : c s - t ,L k s - u: : : 1 , 1 k 1
β-s-i - .
0 . ;-II . β n - I n s [ i c - - S I . β _ cn c t β t β I c , i β β s - i t β u - c s - i - . β - n i . t s - β c n - β β. l vc - s βLi c- β, - β s - d β i o c β : β c - β
t β β s t . ing c- l~ t i n β s - a a r c s - u ! . c - cs- i n , i s - s- β - . I , is - c s - l i s - i i l _ c c - n o β -, ββ rβ s-~~ii β c,s- , c
( βor n t βl c β s- c s- -b , .-:β nβs- c - c l β . . c-I c . c- s - - cs - c - - S - . β, 1 51 β , 105 10 _ ( I - i ) , s - I β s - s -i c c - β s - O h o c - β,
β5 β - , - β s - n r c - - i l, -u-j~~βl β . : s- β~~~. . 10 0 β . - - n β β.1~s - 1 h . , I r b c , - - c - β n n o
C s - β
-β -~~~~~ ΒΌ β
.- -~~~~~~~~
β . ~__ . ,β β--.s-.β _ _.-~~ ._. β __β_._.. s-β- .~_ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ β β. βββs-.,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1, . s - I c - c - i s - c u : c u - c f l b . β,β, - c - ; , n o c - ,O s - [ i β s- β - ~~.
ββ A β β cc β s-, co_c l -_c βs-s -, s - n - o s - S β sβo βt i ns - c -c β s - s - nβ β o_ βL β o s - β I βs - t , ββ
β - βc- - g l s - i c 0 1. s- βIT c- β c-,s - o cc - , l β; o, β c , . . . β β .Ih , I_ C β s- - c - .
2. βA l e- s.c [ i c - s - -s I . β 5 n β . βr β c β b L c ,i s - s -g c _ c c Os-.. ,βS , , C s-- .i. thrn Uβ-; i c - c c- s - c .β s- It , s- s - -h β s- i n _ c c ro - cβ -t s t e : n Iβ ,
iiβ. β I.
L . - l o s -lb β i: β β’ 0. I . b u - s - I r β . cβ.i c β o c β. . β β . l i on s : , nuns - c 1 . -s- s-oc-b c- β~ β t . β [ 1β . β] . β A 0 - s-- c
iβs-s- n β - :-. - n n t c s - c - .i - ,βnn s_ _cl- i b i r u - saβ h o c , , β O n β s- I l l s - s - t i . .s- _cs-~- n β s-d s - s - s - I - S Hβs:.;. 0 β!
os ( l 0 0 5 5 0 β β . t . c s - b , , Ia) s - c c β. ββ - n β .
5β . l , β β βo β β s- . c β β I l β - β . L β β - β β β ITs - βIβhβs- s- β c β . I t -, , i , β o ,I , , . β β -
- - -y [ I I T . i . β ,~~~ c i β :- β β β ci n β , I us-cs β s-s- o s - i βs-β o . : I Iβ, c β . - β β c~~β0 s - β . β - s - c e , β c -βt . c
Os - β i O u - _ c β ~ i.b,be -_ _ β Lc_ bc- c]. :. - . , As -β c s - s s - β s - : o ,n - o ,s i s o β . . . β . 1, c β, - i 0 , . 13 β
s- β . }is-s-βcβI c s-n o LβL 0C51. l c , ! n β - - - tuc u. I cβrn t~ s-β - β s - - s i t e - h I s -sc β. β -s-ro d Air β., s - _ O s - β I βso β . soy,
Sβ .β Is β s- s - -cβsc: i i , β, , β- β
~β β. -
c - β . . I 0 β. c - -~~~~~-_ c 5 , cc-s- cl 1. Β°- . Loi s- - c t [1 1β . β β b n s - s - β β s c - β r βc o - β:c nc β - s o - o s-ill ,β On c- no on β s-or
β’ - -si β n c - c s - , -- , β t ,β.5βc~ . c β . βOc -βs- β1IΒ°~ I
ic: s -c βc ::~s-cs- βb, , . . 0. - c- -s-oS , - - -s- s - - i . o β .. , β n - c - s - - c , , It -- s- - . βIt βI t . - , β: 1 5 .1 - s - c - U s - s - i C .
s - - s - n - s β β s - s - . b s - o . s - cs - β, c 0 0 0 0 β β .,s-βsβ :: - β - β s-.β , 1β ou o r-~s - β s- - β ~ β β , . β i i β s-cβ . βI t β !I LHo , βs-β - . :s- β .
- c β s-i . , , ,s-w β- βit β - β3 3 β s βS .
c l β s- β β s: coc_β c- i : s- s- iβs- - I E . . . β s - β s - , s - βc c : l~β 5 ] . β c - - c - s - n t i o β β β . β β s- β- i t β β . I a - nj c oos-os- l - s - : - c i β βy
cc- I β c-~ s-β0 β .1 u-h o :. :, ββ A s s -n c : . Its-s-u- β β . I s - . C β s - . - I β β , iS- β 5 β s - -c -
l i i i β β - i s-Cβ . β 1 β s - s o s c . β !s-β - c - , - o , o β s- β i o c - , , β 0β c s , ~ s- s- o s- s - I - c c- ,
1ββ β . β c- s -s - c c u--n b c _ c s - n c - o s - t i c t - s- .β ,i . . 1 β β -s - , s s - s s - β - β 1 r a β . β o ,β β :, ____
- I β: ,~
.1 . fl n β los-, c s - . (0. . β .
β - - cβ t ,IT 0 . . β i c - - n β’ i~ I - I s - , c- Is- c- β J . ββ - β - βc-~s s - c -.β: . . n , - , -c u β s -Soβ Ic βs- iβ
so β. -1β0β_ci β ..s- β S i o , . b I - - I I : β - βiβ β n - , β - - cs - s . It I n n i β. β - L β - o c . c - β s - β. β β iβ s - i :. - β -
:β:
is- . β - ~~~ c- c - ,
itc - c-~ rio β - O β t βS βs-βI . β c - _ c β . ic- ig cs-l~ c-s- β j~~ft ~s- i βs- -iβ c : β j c c . i o c , ,i : b n c s - β 0 : tc nt e . , b n o s-c-
f , T o i , i c- . βc - c s - n . co o l s - - c β . ,β di- β s o ; - β - l o s - s -β l n b c - - c c c - i n s - I -
β - β β c i , S no . . Ks - l i -n v ,i c u rs - i
0 0 , . β β - I . . , c-βc - , β O l i iβ s- c- I so , , ; β β w - - β . 1- β β1 ,
I,,, β-β-βc-β~β β β β~~~~~~~~~Β°-β
~~~~~~ββββ- - β~~~~~~~~~ ~~~~~~~~~~~ .T~
____________ __________-
β
-
5 /βs-s o s- β 0 [It J u - c βi β I s i t s - c - , s - I β
- o s - c l co n lcc . n oj is- l .,t e β βs- s - c - s I c . , β c- iβd, - . . s- . sO . . , 2β.. ) β
C. Ho O - β c β β ββ . . βc - . . I β s- b ., cu - i s - I IL. tc s cj,_ b ss.ns- S [1- A l l ] . β ITs -i t . oi o : oe βv er c u c .- os - c - ce c_ c o - c
l βs- o I L s - c oon: . β - c c : β . i- lu - -u - c . li - lI -S E Ic s - - I c s-β_c ss - r βs- .noj , β n s - :: . s-cc -n βs- Us - s- s o β
s- I β s- - c - o s - c l - - u - f l u - l u - s - s β β . 57β β It .
iIos - s-c Iβ ,s - s - βt co o . : I t . βlβ j β,I c _ c o [l _ s-~β2 j . β i i c c _βs - c .or.I ty H - c ob, ns- g .i n s - β β i. , c; s- oc t β s - o s - :
- β s-si c O O s - c - .1 :s -n ;c t rc i ct , - β : , β β- βs-soβ. β o β c , I nβ s - o n c e - c u - b u g 1, \βoiβcu-s-β 1 β Fs - s-us - i .l. o , βc ,β O s - ,
s- o s - βt o - ,:. I I 0 βs- βt i o - c i i _ i c _ c , n . s-ILβ ,: β . .
J. II. lLβ.c cn- :1 s-u s-s - I 1β . βs-I. Tβ s-i βlari [i0 βi , β b c s - j . β boS v.T β .ib n s - β 0 s-I s-dβ-s-ni s-i o s - s - n t t n .- o s- o s- cc β s - c β s -
cs-c -c s-c β s - c - i s -cs - n i :. β C ci,β, β . β s-s-c c, ; j O . , 13-3βi S-o-
il . [s- β5 β u - s - β βt ur d R. . β s βs-so s- s- an [1- c-73b1 . βA V log V aigs- n b ths-:. fc c-i-
-coo rc bs - , i s- src , 0 β βt i- ,i u - - - r s - nc-.β s- ,oted 5β ianlar L βsros -I s n c , ~,Iβ. - β . , c β β β s-s- O , t , β,β : t βs-O:
βu-I
s- s - n c β β
J. I I. T i s - j u - c - β β and i- . . . βI s- s -nβ3an β [l0 βsβc β3 c [ . β s- βs-i,s- β . -nβ , l thc- :s- 5 , 1 , 1 β :
s-1s- β c - s - I t ins - -sc c cclβ giβo s -~ ls - c . on c c .I ; d c c c l c H β s - - co Os -c . s - - b lo , β β. β ,,
J. Hoβs c rc -. c βS . as-i , l I- , . Toc,r , 0 s-s-rn [ l- β(L ] . βflβ I i βt βb u - i s - s - n t 1-_c c-cul tβ,β I es_co l i c, β ,- - ,β β , 1,
E. :1 β s - u - r o o t and . β. s - . AUs-oouc [ 1C C ] . β β u- c- nβs log n Ug -nith c :. s - o r
J u - t e cβ ct ,:i s- c c - ; ~ β.,cI . i s - c β b blc_ β C o o β s - I c - c , β I roc. - t i n iβ_cmual Pr is -n cs - cI onn Cs - - s - cs -β. s-n β l o s - f
Cci s-s-noes- ins-I βs - β . 0 , ,, 11 ββls-b. β .
- . S b β s - c .cr s - . - I s-β_cβc c -,i -β. I . il _ os - so c - s - s [1β β7~ [. - β . β c- βt merging cs-I Cs- r.0 i-thu- ., β C I s - s - b b s-β.
β I S - cs-u -s βs c s - c . - β β 1, _ u -
β’ s - . I ll_cs -h . III [1β βi s 1 . βTI β - β. equivalence s- - so - h β . lu-os- β, s- or nc -. ini , cu - s-c u -c on βs -,-:: β n βs. wit I ,
βi o l . s - r , β c l - I - c s - n Hβ- s - i β s - S . 1 coJ,βs-βi c β. c:.,i ,βc, l in 1- s - io u- , β A - - c - s -Cs -rITe - al . R e ; , -oc tc β 0 3 β c
C o:oo βs - β , - s - Cci c-,:rr ce U β s - I β’, Co - n o n e - I l 1 bloc - i oc r: 11 5β .
s- β.. it s-ni [1β s- s- β j . β u - c -s s - i n c . β s-i rs- 1l Iis - i. Uc - :t ol u- t n β- β :. β 11,-s--Is- - , . Coon1ut. 5, c-~ lβ~
β>0
L , . o β s-~~~~~ :c s s - _ . j o~~~~~~ s-, s-s- ,. ,,~~.A . Iβ _ I ~~~~iβ~~ - -β
~~~~~H. . . --- --β--
β .ββ-β β- β s-~~~~- ~~~~~~β β
5.5 . cis:oiy c .iβ.I , c- β . F. 1s-s-β s - s - s - s ,~ and, 7. C. Rounds [19β~5] . βTh e l o s - i r i s - s - c l βs s-j,1~,
βs-s--si . . oo n s - β s - n i b sc-I. s - c βs-no n lecβ.s.b ty of the circul arity ~ r- - his -sn:β. c u - n β c-ih-ncl. βu -ut s-
β~~
βcβ s_ c.. β s ,o s - c - β. , β clβ s-c-ross- :,. 0111-S 11. β , β s- β ,I?7β β 7 0c-
I . 0 0 . , - o l i n : . -s - i n [1 7 7 1 . β - β I s - b e - I es-nt algoritlu-ross- 5 β s-s-iβ shortest I a t . in s β β bn βs cs- a rcs - .
c , s - t , w c β -s - βi s - c , β s - . c-Its-l b- S . 5 . , 1β13 .
C -rn~s-c [10 1.]. βds-u-uiβ:ive undec.idabilit y β βus- e,~ :-:it i β βs o , β cΒ°usner.I cons-
lβs- . I-I. Ear s- [ IT , βs- [ βA ] . βIclc - βo-ducb.iciiity s-s-βss-s-s-ng cornnb ol nn cs - i. -r . l al s r -Ui u - c - , :, β s- s-s-c-cs-i 1βs-c-s .1 on:
βof clβ -so so ; c i βso (bβ ooc-o s-s-o utatcions- , i - I . - . 11,111cr -c-s-s-d . 5β,β. l i o o s - c ,cs - b c - c, ,-s - β , eds.,
11cc-s-us-cc I n ββ: : , flew β ri βs-, o-5β l -It .
I β . Ear [h ocβs I . β s- s-no S- is - c -β s-s-cs-mi c-_ct sc _ c.: β-nal 0,_cs-Os-I O lu-s-sHy DI ce-s -os -c - cnna t - s- i βia i
I βsβ s- 1-s-~β c 0β0 . β Octβ s-c-cs-rh : 5, 1.s-5_ β
0~ β . bias-A . R . . β-Soi l βL c - β . so , ~_ c c - c - I A. I β’ c-Β°βsβ β β . β s i c β . s - r j []U , βl β] . β β b s - i~, s-id βi dc-ocn s-t .b. fb . u - cn cc i ccn o
β - I n βs-c c - s - s -b c-s- s a t .Ls - on c s - c βi c , . s ton - ,b o β sc: , s- l βs-_c , s -c , and arrs-nc- c- . β , β I n - - c . Oβ βO)β._clβ_c } c s-~c-5~~I
on The-s-cry of Cocos ss -c -β.. b n βsg, I.. βs - - Il . β-
A . V. K uir : s -u - s - cβ - -: [1 i L ] . β u- s - - s - s - s - c b s - s - b s - n j - ti βs- c ooβ, :βs-.c iunnal 1βls-cw in a s - s - c - s - is - c -- s-s-rh b_cs- t i n s -
metho d of c r e s - β hs- β - -s-c , β. β βs . 5 - I s - t I c . - - -Is-i . 15, i t t β i t
0. 1 . K β s - s - s - - I-c-Il [ 1 c c 9] . β l r c - e .I c - ic β s - s - c e - s o b - r i s - c - s - :, int e~~~al s - β 0 n βs - : b o : , 0 _ c c - s - i sβ s-s - iβ I βs- i b -
in βs-r u - I . βc β s - βl -gs-s-, β b. aoncβ 11cc . 1. 1. 5cc - Cc. ~ 0β , 5 c --β 5β β 5ββb -
. -β- . K e s-~ c - ,nJ5~ [ I L - s- 0 ~~]. β AS s- βH list ing s applied to, dos-ta f1,,s-w cu βss -s-L~,βsi c , β
Cons-β. is-It-cord, See nrA dA I S C~c-ri5p. on P r inciu lc- c of iβrognocs-c~s-oi.nng Lβs-s-rgs-us-n~s- c - β.
is-β- It .
I. s-β . Kildall [1β r~β3 1. β A uni fied aj s-yroach to gl- β- b o s -.l r- . c- gns- βc-s-cn - i s - ,Acj s - s o s - ) i c , - β
Conf. R ecoi r d , Ad -I 15βs-s-ni . o s- is - I r b o c u - i . pies of Pr s - -g rc.cn o β.m,i n : s - s - i : β . 2 , s - l s -n β o~s-1s-β . β’ -
- b , jjj cβ β’
β1
-
~~~~~~~~~~~~~~ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.β --.-β .~~β -.-β β ,~~- β ββ--β. β -β--- . .β -v- __~~β
___ β β -β Iβ
β β ci c s- β i s - u-i . ββ s - i s - -β.β gβ, β. ooβ, β~~~ do . β β
-s- I. - . .β.β.lc o β I β
c -_c - s - i c -. β . . . CIβ , β ~ h o t , s c - i . . s-βs - s - : r . β β - , , o l βc o β β .. β., s - . . LI β -
c-,β s-βiβ._ s- s- [1, βiβ . ββ I βs- c -is - s-rs -o s- J , n β s-s -c c - n i β . - c-i - s - s ons - - s-b s - os - - ) l β c β β β.,i β β ,,s- . 0 β . S β c β β βs- Iβ ,Oβ , --
β’ :0 β0 112, 00c- _ 7 J _ )~~
βj o c - s - T i . β . TI c β ,β. , e - β . u - IT β IIβ i - n. iβs - cs - - - o s - c β s - c c - . : s- . βcis-s- ,:c - . β lOs-n o 0β .β -
5\ - t i c - O s - β , . , . I s - s - o s - i l : ,β . β β . , -
bIts - . s-c- - s - s - s -β~β β s - c - ~.i t c - β 1β l β s-β c-s- l1Oi ,OβOOOO.oO, , β , β. 10.0β s- ,, -s-o β o n ,_ o n β . Lβ β - β β β
c - - c - β β β .~o . . A x I s - c - n β . β βc- β5β . , , β - , :, E s - c c - o . 5 - l ot . β
. . βs i s β H .s- . β c- s - I, 1 0 0 , β,,Lβ-o i β s - s - l a > β β: - β C β s - c c s - n β . on - , β βs-~ βoo ββ β) o ββ . . β co , ,i β - c-
β’ . βu -s 1 β r . Ton ic β - 1 β .1 β. β . >
β . o β β~~i β,c-o: .oβ i ni g , β.βoIβ.io , ,s-. .0 : - o s - - -β I c on
- s- . β , 5: 0 β
- β β s - - s - s i β β: . l u - s o - β , s-
β . - Iss β s- .
β’ . hoc- . . . β , β c- no β’ c β . , c _ c o s- , . β . I ~~β.
, ], s- .ββ~ ] . ββ i β s-c- β β., s- βc s - βO -s-n
- - β I n s- β c , β β. β s - i s - β . g : , - β . β , , I T~ ~ c β s - s - - s- - t I c - β .
β’ a. is-s o s - c : os-c s-, A. CCI, β β o . o c . c -e - - , I1 c - c - 1 . β s - i - .β β s - c : , s-c- - , -s- β. s - . L s - β .βs - c c -, s - β l β A . .: os- .1βs- -
u β . , . i,βs-c c . β c . c .β . β c- lge-cβ. i β -o c : . β T β β s- βcbu s- βis-βai 11, - I onβs- ,~ i 7 c - L 1~ β β β c- β β .
s-is- 1 - r n s - - . o f l β s -n i . S I c - _ c β β β s - - n s - o o s - β .~~: β
β1. s- I- .s- Ls -c : β.g β xβ β : ~1 c - s -β J . ββ - β s -s - c - th β.s- β c 5 β s - . , - n β s-- , βII , l b ) Β° β I l c -. Oiβ . β s-is- s -s - i β. b b s - c t . . c ol . o .
K s - I c β . cg- - -onβc- β- s-s-s - n β s - β- . 1. β 5:1 . c n s - : k l i [ i c - c - - βA . - β n c - t i- .β iβ β β ββ .o β i .β s-s-i :0 β s-us-
I s- s- - c r . Iβs- - o h i o . - β βcc . βIβ c- βc- , c - o n : β. . β1 C β s - . A L , ββ , β
β’ β’ β β.β~~~~β~~~
, , s c - β . [ 1 - l s - - 1 . - β c - s - c- i n s - S i c - s - - l e s t :1 c _ c n s - > .ing , β s - O s - ,- s - i β c _ s - IT β c c - β u - _ c o bs-
7. K s - s Os - - β5 β I ~~s - β .β - 1 1 0 β j i β c - β 1 ,~,j s - - :0 β . s - c - . β the s - s - c s - I s - s - n c β s- s- β.t o c-β 5 . - b - β . - β
β’ ββ: 1:β . A βs- - u- β: . . β- - β.β β . ( .
c-β, [ I ββ . s- , c - s -> I β : β I β β1. βc c - .- - i s - c sβ β . i β s - s - - c ,β β s-, - .s - - β β β , -s- .β c - s - s - i βons- > , s-~~ 5 s - .
c β , β β 0 β βs- S β s-A bc- c β - β~~
β β β c- s - c i . C j β I 1 .
~~~~~~~~~~~~~~~~~~ ~~~~~ -
~~~~~~ β s-~~~~~~~o n _ _ r~~~~~~~~~ l i H β - -~~~--~~~ - - β
β β-β--- -, s - -- β β7 β ~~~~~~ β
c -β . Β°β . l s - c -. t β ,-d s - n > β Loon 1. lβ β : s - s b β .c ~~~~ s - s - b c -c- s- . β. . β b c s - , t s β o s - t β o n : β β. 0 β ,: s - - o s as-o s- b- s - t o I ,
11:1: , E,i s - β s c- : } s - a i βs- . s _ c c β . i 7,1 s - s - c s -- - c ,, l l s - s - β: - s- s- βc- - s - .
A. It- β - - β β. , S b . s-- os - . βoci I . - . β n s - i~s- β cnu ccc [1 - β 7 β’ β s- i s- βs-Is-β.,s-βb 5 0 0 0 . 0 0 β :β βons -β,~~~β 0 ,β
- βI β n β β : U : . β 1, 1. - 0β ,: . β β
~Β°- s-L, 1~~
β β, oY, - - -nβ o βiat i β .s- n βi c- , Iβ~~c. s- c βI s,,,s - c , : I .
I s - - . H Rs-c- , :n s , ββ. β , , , , . 1., s . c -n , and β β n ~c - s -. ss - :i . . .. β βiβc-β c s-β s- βS. 21βsβ β . , β .
I S o β s [1 β . s - 5 5 . ββ c β,,c- c : , o β_ cu - cβ c- s -I c - ohs- s-os- os-I 1-h β..- r oson β β i - s - cβcC c s - s - l u - s c , , s - t o o o r s . - β . β
s-0 s-I l ,β:ci . cu so - - s- - I , . J . T , I ,~ - 1 s - _ 0.
β’ s-β . 5A s-.n β s-~ . f - . 1,1. - . 5 - o I L sβ s- Β° n β : c - β [ i - - c - I . β .β s- s- s - i s - - c dβs-I-s-ctS s- 0 0 0 . ; β s-Os - c - u - β137-:β:
- 1 a s- liβ s-u- s i . β Dc1 0. oo β I, l β β n β l s -β β s - l :β - s - s -j i . ITs - C s -. iβi O > β 0 . s-It β.-0.β βo β. s- lo on βs, . βs-of Tβs - - s - c o .
1111. β- β-- s - o s - co oβs- [1 - β s -A ββf βs - s β c-c cn n β - - s - t o β c- -s- , β ,.s- a β β s- - s-s-β . u- c - β C . A - s - - s βs - s- s-n s_ cc - i Syis-t e---
_______ 3, l i - -IL ? β .
β’ s- β . b l o c β. .1 s - . uβ_ c s I . - . s - s - c - β - I s- [1 ,β~ i b ] . Ini βtro s-iucoti3o s s - - s - Ft . n β c- b s - c -.- I l l_ co o s-s -o s -_c
βI β - β s-I -β
β’ β-1. β β
. β β s- β β o . β [ 1β β - ] . β ,β-, . s - β ., β - - - - - β - c s - . 1 β β s -_ s - n o βs-A c-c ti- - c s - β a c o . i . r u c t i , u s
I c . β β
. - β. β- β , . β
C . 1β -.β I β l β c-r β ,l : 111β - . β I - β - β - , I - i b o ; β s - h β s - βs - i s - r s- β,βn β- s - c s - n - s - I - n s--iβ
βl β s - ciβ _ co b s- s- oc o β. f lu - : - β β is - ,β , ,β, . 1β ., β β ββ β - ,β.
- Ccβ : . c s - b, s- β’ βo,C , βns- .s - - i t s- ct . . C t - s - s - sc β iβ s-
s- c ISc-n,β cs- s- K . Β° - -~β. - c . . s- [ 1: 11. -β-J oT s-, β β s - c- I. Β° 2: 5 nβ :s-.β . soh β c- _cs-I c ~β 5&.β O -
s - c - , , β
1 - β , . β.β β , : - - i - - u - c - - : 1 . . .I 5 β β~o β o o s - s -β:ββr [3 β C . β j bs - s - c β β - o c - .i i v s - d e s - s - c β. s i β s - i s - i β, β s- nβ r c - βjcn loc -r
-s - x > r u - s- on ions - w i _ C c . : s - s - i - s - i β i s - s~β0 iβ s- u - s - S nc- s- s s- co-c s - n c - - n t l a s - c o c o s- β , β β β c .
c , β β c β s - c s - s . J i - Ii - , i βl IL β l . b β . c- βu - s s - β.β, s β s- β. ,β β s-n β s- A te-h is-Id_ m s- i βc- s - s t β s - -cs-H βs- T i n - β
β 25-
5 - β rr i sc f i - β β,I J . βA s - s - -o s - I t -c - β t β .-r βs-~5β 1 - s - c -s - 1s - s - c - c : . ββ ,β c: . βc - s - O s - ii, 3 ~ L L .
I . A ic c- β [ I / β 1 1 . β i-Is - IA c i - βe- i. de l - c - c rβ I c c β s - S i .-n - 1β t ic βs β russ i t 10 β.: s-s- I, 50. i Iβc - β
on i t iβ - β s - β 0 βu - os - ; i s - . β i o β . I s-.β1. t s-s- s-β o c . 5β -- βc- s-0 .
L i_. ... ~~~~~~~~~~~~~~~~~ - β -~~~~-4 _. , , ,~~~~~~~~~~~~~,_
β.β -β. -.β
β~~~~~~~~~~~~~~~~~~β β β~~~~~~~~~~~~~~~
β β
β’ c-It β : s - 5i β .5 ,β β i β β_ cl o . I I . -5~s-r o β, I n c - i - o s - i I T c -β. Oi} I s - s - i c - s - c - s - s - c s - β I s β - β c - - s - co c A s - . . - ] i s - c ,
s-, u-c-β β 0
β’ . βI - c βs - os - s -n s- t oss -. I- β. U. i~5 s -~~β fl [1β s - A L ββ - c-n c- ,ocJ,β.scle-rci t.TL βs, s- β iβ s-c- c C , b c . βncos- c s - s - s - : s - - c β
n I β . , β i s - s - c - β. c-s- s- c- .β :β . I-l a t h . s-5 :c - β . 0 ~ s - i5 _ β -
β’ . : s-c- s-2 s-s- β L L t I. ~~~~ j . β s- - i βs : β~β1 β n n:β, t s - l s -β ..β o - , . c . c - s - o C OIiij s - ns c - s -n c - t c on β s-c- _ c I~~ - s - β s s - -- .i gIβ s.s- : b . . --
β’ s-s- β. s-s-r i β oβ [1 . β - β - 1. . β β i n s - s-β.o:c-c s-s- Iβ I L l s - c - c β s - u - . s f ss -1}s_ c in s - - Aacn~β:.I s-u s- eIci . o : , I r oas-,II s , - β
~β ,1Aβ1 βI β .βC c-o os - 3 . I I - -1Cβ-s - ) .
β5 . } s n s β . c s - β : β. , [~
c βon J~ I r I s - - ate c -os - βsr s--s-xxo:ic -o s-cA . l os-i .
β’ H b e - s - t [1. c- - b: , 5 . βF l i _ c i t e s - c c - u - o h m β β - u - , ; r s -c esseoo β s- β . β βncβ: ,aIL Β°I c o , β , β . β
L~ β cc -, .
I s - . c - . i s - A s - n o [11 , c-57 1 . ββ c - I s-O n β s- c- β β s- cs - . s-_βo s - s - u -s - β . c oh .β I β , - s - s - c s - s - t O β 1β_cc ccβ s- , s - -_ -O s-: c - s - c I s - β s o - .iβ c β,.,.1o 5s - s-t c c s - s- > s -0:,
,β vo βtcns-: β7 s-s-~s - , β’ s-β β’ 17β - c -β t~. Ci.
Is - s -c -bI n [~~~ , s - - - I . β I s-~s - . c - c s - h 1 h b : ti, β β c - β1A r b t - l βc β, , . β s- -,.L ge -r b tioos- .o c _ c . s-. s - Tc - cccnn I, c- β.o-:βIty :
βS ew D!cββ.o-s-t β. β - i s: s-s-o s-cs- 1-is- os- c- - o c t i- β. βs - ,c β.s-,I l l : , -β’ F. Iron s-IL , - s - b . . A u - a s - n - - o l e i i β .::.
- s - -s -w Ys-c -s- βiβs , j - L )~
β’ Ri- cs- as- s- cs- lb . s - Iu , j β i β , s - c J s- c [βI β s -A S ] . β βc c- Cβrβu-iβ 0βil~.I 0 sC j ,c-s-_c βs- s - s - s -u I n c - i c c 31 t O n s - c
d.n. C rs - s-βc -β kβ - o β s - - c -oK s-crg con , j s - s - s - ββ. c - Uβ , β n o β c . , s-s-βc-c _ c β I n t , s . As- c - u s - _ c - _c. dA b-b r β5s-O β . : β’ s-
_i
1 , 1. ciβ : ol f - s - c c s c - L I o ,A , - -11.
c- . .1. iβ .,.c- , c - - {1β .β(-~i j . βTr b β s- s- β _ c . s - c -β. -s - -i C l β s - c - c - O s - c ants- t o n β β..i . c β. β c- O β s - s -
.5s - i th . Annol l/::i o s - c s - I A s - c 1. 1 ca t I s- β s- - β - , 5 : - _ :Qc-β
. β . I β β:s- f - c - - c - βs-β] . β iβs- ~ βcβ c- o I s β b c- c - - - o s - β 5 0 s-c :t βs-iβ.I,β~ o f β Cs-u - s- iβs -β :.- β r e - L I I s - C .1 u - iβ. s-s - i β
oβ~~c s - n β~~e s - c - c - s - i L i v e d βc βin . ite cβyst βs- β: c - i β 1.5 near β . β. jus -n t ,T β O s - . , β Cs- β s s - : T n I i . - β - o f
-β . S o β β c - c , - . .
~~ Accn βIβ βn c ,lc β s- n c - s - s c , N u -w s - - u - c -, 1 β i β iββ .
β1 . l β - . β.c βcs - β, .I iβ . los- . βlβ s - ns - β, l a ~s- [ i ( 0 β β l . ββ . β . c s- βs lβ i I I 5 β β. I o . s- s-u- βc s - c t : s - c - !β c - u - cβ
el l , β- . i c - - βc- s- , i - u-n β o n c s - n β s- S i - , ββ , 1 l 1 s-J4 -β. β c c l . Aal , I i . , - β q s - s-s-n c β .
- I ,
L ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ β ~~~~~~~~~~~~ ~~~~~~~~~. ,
β. - - ________
, β . . s - . c- . Es -β . . s - s - i β,β c u . , u - . I . C. bs-.s--s-ker [1β . β s-S.dβs- β. Cn β β - β c - s - : cc-s- t s-
β:s-- - β ,s -β.x c- β. .In. β βc β i u i - s - c β s-s s- s -o s - c β s -ru: 1 0 : . β ,βT~7.β S. Cs - s - c oos s - n t . 5 . Ss- β β
U. I - s - c - c c - o n: uβ s- β 1s- 115 i . s- c- o s - t i cs - c bs- i o o. - s- re βsui r es-I t o n β- β β ,~~,βsH~~ s s - β β s - c- , c - β β.iβs- s-β
- β c β s - i .:: c c - s - c β i- - c . β C I IAb βI I-S w, β _ . 15- . ,
lb . :- β.c -s- c- 5c- c cs - c , c β c . ~~~~0 5 0 I s - s - i s -, β~β } . s - C - I s - u β s- :I s -βhs-βen 3 - n os , s - s - s - i ,β - c c β.s - _ c i . ce }βne n
cc - c - _ c , , II u c- c - c -
βs - - b _ c s -I [ i β 5 -~, . ββ (,β s- s - n c _ c .s- βs - - -. b O O C s - c β s-liI : c- β s - s - s - - s - -s-I ~ r id βs - c - o s - . ββ - β S A ,S-S .1 . β s- ββ .; β_cs- . 0 ,
s- s- os- s- - s - s - i o s-nc- , . : ,: β’ s - . ,β , β s - c t β βoo β β - β c - n β 1~, β. β h 1 . β β s- s-s- s- s-i c_c; β I s - s c-s- s-o s - - s - s - - o l e - s - n J
-s - s- s- s -ol, o .β .β - . r ioc cβ : -c β lu - : , β Incus - cc. I s - s - c s - . A S s - - c c , . - s-c s- iβ.. It , 2 . - ββ Lu-7 .
. 0~~ ,~~~ ,β cc - β : o s - C - - I l t β I . Iβ : , β , β β- s - c . o ] β o - o , i β s-~ 0βsβ Coo s- o s - s - o h s - ,. β. , Β° s - I - s - oc β, C l t ,β.lc- β c_ cs - s-i - β s- s - _ c . β .
I I . . c - u s - β IL r :3 ] . s- β. I I o β . : o i n β β, ooa t i u - s c - i βh β .- - β o β β β β c β , l 1 5 β lβs-cβs- C , t nss-i s-: o _ cs - Ho ..
I n β s - c βs- I c , β . II I , , . cn b β I c-s -c- ; - .β c--i s - I s - l I o nβ , . . 1.
s - c β s-cc c β s - β I ] . nβ β: ] β βs-i Ri s - c c -c - ββ t i cs - - c s β -c- β βs- s - - s - _ c s - s - l b .1β s-s - c -. : ci c c_ cs - βu - s-,βIt. h l I rβ.-.:uβ
- s - c .y - n s - s -0. .
-- us - o s - c-nc-li s i c - c - s - - i OOc βUs - s - i :Cs - β.. ~s- t . is- s- u - l o i s - e u - c -s cl b-b _ c s- c- c - , s - c s - β.t~~~~~ ,O
β J s-o ,Iβ:β nβ . 1 l β s- β o s - s - n o Oβ , β β c_c r c,
c - . . βcb s-~s- s - , ic β s-gβ.~ [ I o n _ c ] . β K s - s - I - I I n c . s- - :l - β β. s - . β tj - ,c>n c - s - oc s-uit s- s- i βon - c s - I _c-i βs -c-I βIβ.,s- s - β I s - . - β
cs- sc-u - i s β c .e: u - ; ~~βs- βIβ,β..~~ βs- con o h l i l u - s - cβ i i c s-o:ach~~s - n u - c , β I ro ju - s - c s - - h-Cc -C β s - - β s -cIa:
7β - - c β . s - n β c o , o o s - : : : 07 . 5 β : β i s - . I n , o - 1. . s- β I β β - -s- i c s - o i c-Cl,β .
s - c - . c : o β-s - .i , β -,c- β- . . k s - c - t nβ s - . β c - s - .- ,i - . Hc - s- s - β s - s - s - β 0 s - β s - β [ l c - t - 5 } . β j O j i s - , s - O . O , β b c - sc -_ c β-βes-h -o s- , . β
1 1 5 - - u - s - I : ci β β β c s-nno u u t β s- β. ,i - O s - β β . β : - s - s - I . βs-cc β s- u s - c -- n - β.1β s - I c - c - s - s - c e P e s t . , Sβoi iβ :s-βoβ:cl t β_c oβ
β - β . 1 . 1 1 , 1 - A β . β (β c j , l , s - β β β β s tc - s -~β c -I l c o i l β in s - r I - l u - c β,:, :,A u- , s-s-cs - o s - U i .
β ,
. β β S h , i [1, :(. 1 . β c- β cβIo s- U u t j n , , ~ c - b I β n s - ics β - ,i~ t w - c c C- s-sβ s - c - β .β
. . ,β β
. β β . 0 . 0 β . 5.
L~. - - ____________ ______ s- ..
~~~~~~~~~.β-.- . - 1β -β ~~~~~~~~~~~~~~~~~~~~
β - ~~~~~~~~~~~~~~~~~~~~~~~ ~~~~.- -β-ββ-β .ββ-- -.
~~~β.- -.u~~
I I . , , . , βn i i c - β - ca j I . -- - ] . ββ b . c iβ.β s-s- , β - i βs- β β s- β. ,I . o o β_ cs - c - o i. cs-r s-aJ5/ :i s s-c s-i β , c β s- β Β° s- . I , , o , βs- β c r _ c c - i β .
~~~- cu β , o s - c - βcl t βc-β c - . s - .I o . . l βcs- s-β - i βi t locc c , β I l s - .D. t i c - β c l : . Uniβc-β- s - β r , i b , -c-- ~~s β As-I s- c β s - c -: β s - s .
b - I . .. β s- , r cc - [ 1 0 0 1 . β A ,β c - - -βc-: c - I s - rol t ic-s-o. β- s - n :βindics -d sb . .n i .c- s - . t s- 0n1 ,5β s - s - s - a
- β s - c - β β ~ 1 r-s - :.I c 5βs- c- β s-is- s i n βj β, β. s- βs-β s - β~~β β t ime O(n~ 1 dβ n o ) β c - , βI~~, S : . βs- on β. .
I t . C . , β i - as - β βos- c . β:I :- - s - β β_c o - 0 A. . β β -~~
.β - s- [ 10 1_ c] . β β,βIβ.s- rs -( c r β L, βc u - - . rc-s- -,βs-Is - βIu βsu- s-c-a, o.s - .- n βn t ls-oh
s s - β - βs- _ β , 1u , βs-c-βI: cβ s- s-s β-c t , ββ I c β β β . Fit s-tb A_ cr c β: ,,,, s - c - s - tb C β s_ c s- os; . -.c s - βs- βIβi s - β.s- s-rl_cβ O s - β
V . ,.β s-, s - β s- s- β ,β - , , 50 [ iβ β . ββ c s - o s- β I s-cs - c c - s - i l : . I s . - , t 0 c- n is rs-_ ct Β°s -k s - c l o βos-cco,I , ββ 11βs-s-c , β s--I:ci s- βs--
l b s - c - s - l ,- .β:β. - c - β.~i S-c .1.7- , ~~~~~~~~~
,β an βo β c - : : cnc - i lfβ 1 . βDiβ..: - . β s-β s - β c I L ,βuind :kccc βs- ; i s - o s -_ s- s- ct vs-., c - , l , , , c s - , s - , s - s -~ ,~:I ,βc:: , . H~ , :c - βcss - ,, ,ro
:. . s c - s s--t β o s- c - s - s - n s - s - n o β, s-c-β s- H β.. - s-s- β s- ,l. βc - t . s - c, s - s - s - c bc c -u - o βI βI s . c i -crβh - 0 : , β Is - os -β ,β s - I c e - h e Ma β i o β . oo , o. s-s-i. s--s
β - lo l l .
β,c- . β 0 o c - s - , c -s- s-i β s - s - s - c -
. β
- βs-I, - βs-cc: [1 c-β β c-s-c- β β s-c-ct β I : β. o . _ c β βs - s -s-s-_ s-_c t est A - n
r .S oo .s -J i s-c-~c-β , - β s-It Sβ C. C β . : β c c -t .. to cc -c - i c - c- n .
K . ~~~. cβ oc.r , β~~ s- [1β ,~~ ] . βI c -β s-~~t l i - f A r~~s- s - u - c s - c - is- - c -r d i l c r - . -s-s-r d u - β s -1 1βs as- .c- βs-s-r I s - β, , o β , . . β
- 5 c β . .. , l . I , - - ic - βc .
β’ βs-ar : c_cs- [1 s-:~ - c c - J . β 3 fl _ cO s-.s-i~~s- . iβ it iLs - O f β n β ir ,ILit:C Os - c β s- I: C OO _ cl s - n o β s - s - n t ,
Irn Sβs- s-. s - n - s- s- c. I c - s - t I - s-r u - 3. l7- l~~.
Tanβ ,j β s-n [3, - β βL u : ] . β I- . I o s - -s-i l n o ,β βi- - os - i c β s- β. n o - βin Ii recb , c - c - i L:rβd i o . , β .C s-t~-b s-i.
K . E. βlβ s-c-r ,j s-n,u-c- [1 β 1 β c J . β I β c : 1.01 c- c- C fi s -c o s-A rβ s - s i , r c - c s - s - i O s -: I l ~~ , lit , -,- , β β . (β o o c : o βo Ut er β1,nd .
I - Β° Β° βI s - - c- enc es - . - , c-β s-
c- - c - . βj β -β s - r , β .s -c - [1 β s-c - S a l . β oc β. β ! βi c - s - ,l - c - nc - c s βs-β o f β a good I - u s - , m s - - i - l cL rc -c -co u β 11 .s- , j s - c - , β -s ]
uni -sβn c- c-J.r β -r i β : . β . β
~~~~~~~~~~~ , :s- i5 - , βA S.
s-s -s- n [ 1 , - I s - i s - I . ββ Ic-c β -lica!. . I c β s - i s - - otβ - β c- I - l i C. .ro o~ nc- :s ,o In os - s c-n bal anced I r u - c s -β ,
I - -o le - i l ( βs- s- I onβs - n c - β r t 1 S β s - , β . β l C β f 5β βclfβ, c β n β , o o t , C ~ C e - i c - c s - i c e - s - - c ! .. β I - c- s - s - s - I-i βs-i
βs - c - i β,β β. no i l l s - .
L~~ ..-~~~~~~_-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~ ~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ β
β’ β i cuβ , l s - c - n s - I Jon :5c 1. β ls - β ivs - n~ O i l : , ~ c - - L c l β s - u : . : s -un - s - c- - β - -s - 10 β s - C c : . β
βI β - β c- s - So. ro ic al K u - n s - s - n t C β I t s - I t β CC ββ s- β t s - β s -C -, (β c- s - o s- n U I -n l u - I - n o , c - I β :: c .. C c βcs- c 0 β Os- o n
o l n c i β : c - c n β : - t s - : .
ill. . IIcirjcus- [ ic -c - β : - al . ββ I1 β β βc s - I c tin s- os nβ s-c- β u s - - I βI s-c βs-. :1c c - c β. c- βii :csi s- n β o i 1 s o . ββ , c β - s - β s - c - c-
c-o A s - c oo β., l at I c s - nss - , β.~~β’ c , I s - _ c c - s - c c β u - n d - . J . Kβ β - : - . - β 0 s- . , A cctds -β o o 3 c
i i - - β ::, β 1 w s- s - - .
K. Iβ - . 1 c c - n β s- i s - s - n , [ 1 Β° - c . β - I - - c s - n β o c - L J c u -βd s - 1u - .β β β . β,s- ,rS:s-β:s- βI I β . β s - U : : s - iI β - l o j , β I i - - - .o LO t - β: .
II. s-- . Tar ,~ ou-s- ( 1 β . 1c-β .. , β β O β :O O c . c-,c β , β c u - I n i n c - -s- s-βc n β β - i C s-_ c - - β s - s - s - t i s - ,s-s- c-
c - - s - I s - s b β s- P s- h i s s- c - s - I n β : β β s - _ , ββ 5 s - .~ β . . β βoi .S n β.βs- s- s -o n β s - β s-,I As-s- S b s - s - c - s - - c. β s - i β s - Is-s- c r y . Iβ
- o ., β. c c βu- 0 - β c i β .
β7- βc- β . s_ c β s - : [1β - β β0 ] - β o s-uβ. β β A s -n s- s-β .β
c e s -β , [s - u - s - I o - β o c ,1
s- I u- - - :, β β β - n β s - , .
βIc-s - nβs - Os - c -β [1 β .β β β i . β β cc - - β rβoβc s - s - I βc-L , c- β - o , c c j s s - n β c - n β , , w ,1 s - i , o s - s β . ; n l h β o c - t . i , s - , L os - s- i n β β
11, 0 : - i s - - 0 5 _ c s - s - I ,β n β i i β - β Cβ n s - s- -~ 1 :-s - c - s - i - c s - b b s - c - lI :. , , o ci , c ,~,, s- - u - Ic - sc β .
5; c- - u - s - s - - ct 1 β o - , J s - c - S s . i s- s- ( 1 - 0 1) ,
c-IL. I . Β° βlS 1~β:oβ s-o . [ 1βs-~~ - c - s - . β A s β c c - .:t β ..L c - β - r β I l i , cβo s-Cr β i c - el,t β - o . i s - c - s- t I n , t c βor. s - .
s- n _ c - s - x s - c - β- . , , β c , . ,β ACI βc- l o s - S β s -n βs - s - -c - i i c _ c A , 1, - i β , I. β .
β5 . J s -~ ! fl ILn βcs-c-s-βc β I -β A l - i s - ] . s - - i - c s - co . βs-_ i e c - βc-oβcss - is- - β. . - . - ,~u - l c β s - t s - i i s - , c -~ s s - β I . - I , β - cβ :β , ββ r β β c .
s- s-UIβ β. o . β,,βs-βc-j . - c s - 0 β~~~ c -s - f o o t , i o c-~ Cy: I β - β I nβoi c c - s - o s - s βI- - , , - β - I
1.. 7. V s - i l l - -s-c -, : [1β.β βc c s - . J . ββ :;,βr n - - roni - β - c . β - β :-: β β s - β o β β -, β s- - c - , - s - s o c β -s - i i n s- β , , ,β
- βs-i s - l u - β I c - β . β C . C s-ns - c -~~ut c - βr -in n , β Iβ.β , 5 c - .β .m , . β I c - - n i c c - - : i I s - , β
β: 3. - s -s - c- [ β I c- c - S I ] . β s - c βU s - s - c - c, β S I c o - _ c r - s - s - s -- c - - c - i c- u -u_ ms- n , I β . - β cs-s-o~ s- s- S β c- c I β β c-_ i
c- s-c ~n os - 1 - x I β ~, . β 1 rcaβ β’ c - I s - v c - β c . I l is -c-β - 0 β s - . I βc- ββ β βs-s - O s - . i s - , Ti n- : - cs-,- os - I β c β s- c- s-βs-I n s - t j__o s- i .
s- . . 0 5 β’ β.c- _cJ I β_ c , l [ 3 c- β . . β βorβ - βo s - n j s-βc Β°, c - n,rr .s- s- β - S ot t 0 c C 5 - β .βs- ns - - βr I o n -
o -mu i c - - s - oil s-, β I c - β β s - . i n i - I s - , β us - c f cs - I 1 .5 s : l s - c - βs - I s- c s - o s - c s - i , - i β S C , - β - O , s- nβ s - - c - u - C c- s - c s- c~t β β s-β
l l t u o S u -s - , I Is- s i s- c - c r , , t c-_ c c - s - i β I _ c - - s -c - s - n , β .
~~~~~~~~~~~~~~~~ ~~~~~~~~ .~~~~~~~ .~~~~~~ -
- - β ~~~~ β β~~~~~~~~~
. 0 β s- s - l β s - s - c - 0 5 , I 5 . β β s - s - c c - ,β . cc - i o n s - , . . β,oi , βl , β b 5 r o s - Hc, .βs~~k l . β βc-
c - cs - β c _ c , - I s - I β i s - s - s - i - I s - β ~~ c - c - i t , 5 ,s- β c - s- c- β-Ic c - I lls - s - s - c.- β I c c - s i : nlbu -β s - .t s-β :. [u - . -
c- n O βs- c - i l - β o . S o β s- [ Iβ., . ββ A s-s-A ss- d O β s - β, - β : cs-iβ . s-β n β c sc-n C _c_i _c, c-_ , .ts- β s- s-
- u s - s c - cc ,β β , . β C . β o c - o -, . . , β β β , t s -.5) β s - I c-s--o I l β .
c- c-, β c s - c β 0 β β β- 0 5 flc - c-β s - s - β.β s - is - s -i c-s- _c - β n s β c . β c - i c o o . s - O c -c- _β is-l u- s - l o b s - c s - s .ββ c - β . c Is - c - Ls- βcs-
s- β β : . n - ~β c--nβ β, s - - i s - I n c - s - I s _ c s - s - n s - Is - i s- s - O s - n c β s - os-n 0 0 - - β O s - , β I_ i .
β’ l-1,l, c - c - s i β s - - c I ILs-s-Iβt , . ββ c - S n - c - s - s - - s - s t - : 5 5 . c-i s - c l βs-lt d -a s - β c . c n : β. β 0 - sβ - r. the o , c .nβ - . o βs- s--
on β co o n . t I C I c - - n O i s-s-cs- :.β 1 β - β . .A s-c β s - c -.βoc. i . Anc-r:u:c-.i Cyc-ccs- . s-c - - βow ss - fo c - t I n , c β βA
s - s - f - s -iβ C. β - β 0 . - s - c - c , ~~~~
[I c- Β° ~~~.
-- s - s s- - c 0 s- s -t .lon β s-~β β .i c c - c - .tI ,β , . β s- β s - t c - β s - - ,-u r .1 _ cu - i n β - s βs. β - β .
β β s o βs- s-. .
I1 ss-t . s - c - u - β, i . s- . . , . . ~~~, 1,YJ -I , )Os-
s - - i cβ s - S c - [i s- l i . β c - I S o s- s - β - β - c , β s- β~~ c s - s - s- .s-os -i s-t1Us-,s-dβ.o I c - _ c . :: . . β Au - β,. s-cc o f l iA , :. β βt ,, Csn - β
A. β β. βs β s-s- s -c - [i - c -I _ c c- . β c- β s - . .. β . ( β~~ i s - C .t C s- c-
~ s-s-id~ β s -β i s - s - _ c l i o ,r β i s - - u - s- s - c -β cco oio n .I :55 ,_ccβs
c c - βu - _ c 5 . I s - lc- β L u - β β,:. .β I s - c A~~. Sr -s- s . 7, - t I - c -or : ~~~. 21β23 .
Cβ . β βs-s- , [ S l o t β . ββ βn o c-s - i s - β s-β:c- s- - s - cc- c - s - L c - - ln us - β .l 1 - - I β s - c t c β β.r C ioc C _ c j β β β l ~~~t O , , ,
i r . .β. cs-o . l olcl g l i t βS cβs-s- u - c - s d, c - s - A l l 1β Dβics-l . s - - s - n c - 5 c - 3 n β s- s- i βy oct Cs-c -c s -o ut s- n d , ~~ β _ i . βS.
-βc - . C . βn βs - c - , Ill . β. A v i s - . β c - s - c s - I i . L . l i i v βs - , t [l s- β .t ] . s - c ,
I(s-s-~ , ,
β. s- s- s-s~
I - c,β β i . s - c - u s - n cs - β ., β 1Β° - nb β . β, s - - il l s - c n c _ c : . β . . β I c_ cu - . 7I s- s th i4Jβ_csj β _cL [Cββ
i β s- - s-~ , β β c . β . , oc , ,; β 0 5 I n s - c - ; , to c s - β~ s- - β s- Iβ .
Β° . β{cO) c- c - s - . , s- β β O β (.L c- βc-β l . -β - β β -β,n - . 5 I . i - -rn β c - s - s - _ I β β..r : in g s- β, 1 β s- ,o o t , β . βS β o β β lβs- c-s- βl, c i s - s - Lβ1β. , 0 0 β . s- c- s - , β
s- A β . I , i s- ~_iu-c-
_ ,d,_,s-,..s-s-,..s-βs- .,~~~~~~~~s-~~~t β_cO1 ~,s-,s-.,,,,,, ,s-1 s - , ~~~~~~~ ,,_s- . . - ~~~~~~~
βJ
.β i ,,,_ , - - -.β,,_ ,,, β
_ _ _ _β ~~~~~~ - .~~~~~~~~~~~~~~~~~~ - - -~~~~~~~~~~~~~~~~~~~~~ - -- s- - -~~~~~~~~~~
-- -~~~~~~~~~~~~~~~~
~~~~~~~~E1~~~~~~~~
tY
.~~~~~~~~
.:°°
500
sec sec sec s-cc see- sec
.0 1 .3 Cβ 1. β - 4 . 5 10i s -C u- f l s - og n i β
sec sec s - u - s - sec-,, s-ec s- c-s-Us-
iou n . 1-11 .25 .1 Is- 25 2,c- s - s - s sec sec sec r ec cs-os - fl
s, s-c-7 . 02 1 1,7 1 21 11 .7βs-u-cβ s-cc s-c - cc mm mm hr
00 . 5 1. 1 220 125 5 . 1 1βses- hr days- cent cent
. 3 - 1 - 1 .1 2. 7 3 . 1 . βoss - s - c . c-s-s- hr cent
51 75sec yr cent
~ fl 58 2.1-3 βms-in cent
Table 3.1. Scinn ing Tics-s-c i β I :t imatc -s-s .
(One s - I s - i β s = one rn1croc :econs-~ logarithoc.: are base s-a β
L. ~~~~~~~~~~~~~~~~~~~ ~~~~~~~
β’
.β.~~~~~.
c- c- D
1.1β ,:c-: u- 1 s- c o β s-c- i i s c - c 1β.c- cccβ l β s- s- es--
cs- β oo β:I βs -x cβct y ~ s- u-u- β ( S . β s - c o i n t (7 . β; i s - s - β 1 (17 s - I β s- c- β: 1 (c I,β c-: β c - O β s -~ (i cc - c s - n t )
1300 n o 1β 1 c- β 1 i s - - - i o~~~
β2 3 c-c β7 s- βs - I .s - c 1os - -~~ s - , ] . ) - s - Γ§ i C β [. βs - I l 0β s- c - d c -
c- ~~ s- β, .~~~\ 1β s- βΒ° . i , x l , s - β 515 .β s - i β
io2 i c c -~ ~ ,,
- Β° β i_c is-.- β
I . s - c 5- C . 1~~~ 1 i-j ~ 1 .~ s- 1(s- β c- 2.i~~], β β 1c~
.7- -~~~ s-n 22 3- 51β. 7 : 11,7 15-
c c - c-~~
β[ 0 ) c - c - liβs- ]v , c c -
i s - Iβ 2(T 3c -5 -;Β° 1
12 1-~ 20 25 β7βs- c -
Table 3. . b- β s-s-β: .βi β s- o_c o. h i s s - c - c - of a s-Β° I - c -ILvah i β,β c rc - s - s -h i-c β ,.
(A factor of l~s-n I s - n o nβ s- - a s - c in βo , n cc i l , i n β :o β-s - u - n j C sr i β s - - c s - s - _ Os - c -is- t
s- β s - c - - : t , -r of I cL I s - s - c -β i β s- s- u- in t i r i β . : s -.
70
L ~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
c. s - c- s - f - s - :0 0 , _ n β n _ c - - s ( i - s - s - i t t s- βr :βno β. ar i a ,-: - s- J β . ..n 1 I r s - S_ cd s - - u - , , - s . - βL i , I s - .
I J r i s - c - r d s - r c - : s - s - s-ou - βt s-
- . r c-i -c- r u -A s - c βs : .
s - i . -I1ra~βh:.
u - . Tr s- c-~ s .
2. s - i c- s-con β: I s - c , .
a P i β-~~ is-- β sn β s -
a. Cc - s -c u β β β . 1
s-s . s- 0, : n s - n b - i , β β In - ct .
I . hiI ns - βc - βs - βs- - C β β 1 r ,os-
ii . L u-al c-sc- 0) β.s-31βOs- 0 u-n -
I,~ lβ s- t . i . o cc i ooa β t i. o s- - s-βs cs-c c- .~t5s - β - βi:
c c . c s - n β s- -A .
C . As-ny β :ceu s-t :c -t β cs-n .
5. Iβs-s -c - b βs- c s - i n c - b u s - c -β - n is-n o l β s in β - .
a. s - s - s - t I c - C :s- . s- r c - : s, c β~~
s- c . I s-s- r t s - t i _ n Iβ s-C s - c s - ecs - o -.s - o s - t .
0s - . I s - bβs- - ca r s -c s- βr as - od c-βoc , u - n nb
I
I ~c-~po~ βs~ on ( ~ i
~c . c- βi~r l nkcI cβsp ( u-- s-β,c- C.~~s-s-7 I s - i s --sac β c o i n s - s - c - : ) .
Tab le .1. T- s - c βhm s-β ques- s- r s-I~c-.s-d As-h g - n t hoc.:.
s-( 1
~~~~~~~~~~~~~~~~~~ ~~~~~~~ ,. s-.
~~~~~~~~~~ , ,.s- .. _ , - β β
value link 1 link 2
(a βS
V s - J U c - link 1 link 2i7T1i~i ~β
-
_ _ _ _ _ _
β
~~~~
L _ _
-
~~~~~~~[~1~L~11 -17i~I 3 I
(b) (c~
rs-igw,β .- ,.1. A l,, s - c - } ,c - _ ct st ~~~c- cI u r β - an β I i t : rej r β c : e n s - _ c βs - t i - c c n by arrays.
( a) lβI c - c oo β u - - i 0 β u-cβ - s - I
(b) Linked st r i s -c βl ou-r c-
( c ) Repre r entat ci - s - n by three arrays.
_ _ _ _ β -
F!,-, ~~~~~ -β _
~~-
~~~~~~~~~s-β
~,s- --
~~~~s-- -- β.β-.β - β ..β . -,--
~~~~-s-
~~~~β s- --,, -s---β-β-,β β-β- - -β’-β’
s -s - I s- is -β
i I Sc - s - s - s - s - I 7
7 - + βs - c s - , l l = P
11 s - s -~
5
c-_c
t :c - 151
(1
s - s c- i. , ~~~~~~~~~ i - β s - r . s- cC, β c c β β . βs - βs- s- β βj , i c - β s - s - s - c - : βs- , β a u , :i is- βsc β, ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~ 1,
( s -c- S ,β~βnβa5,β nβ β c u -u s - c -c s - i t s - c -β.. s n .
( i s - _ c . s- s - n c -- , . c - ,~ s- β β s - u- β , c- s - n c β c- f s
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ s - .~~~~~
βs-β. - ~β .β β~~~~~ββ ~~~~~ -ββ- ~~~~~~ β --
~~~~~~~~~~~~~~ ~~~~~~
head 3
2
1
5
ta,il 5 0
lβ ,I giir-s- 5.3. Re~ r u s s - - s - n t o s - t i nβs of list 3 , 2 , 1, 5 , c- . , 11 icy s - I - sc - s-ib15,s-
linked s-sctructure .
714
- -~~~~~~~ β - -β ,: ,-
~~~~,β’
~~~~~~~~~~~~~ -
- - β
_
[~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ]
(a) (ss- )
head
i ._ _4βiI . J ~~~~~ 3 J O J
2 ββ -{~I β J 4 1 1. 1 1_~P1
~~ β~Fi~1 )( ~~~ i
β’ 1 ~~~ ~i ~i
11 ~~~ *21.1 ~ β j 0 J
( c )
Figure lc- β’ 14 β’ R e p r - β :β c s - s - a t ci cβ- s -βc- . c - 1 β a gra ~ch .
(a) h u - a s i s - .
(b) c-βc-dj ac ency cs - , - , t r i x .
(c ) Adj c~ 1ββ. s-s- cc Ic- β 5β t u-us-. β t ur s-s.
L~ .
.
.s-
β
..-. s-As-i ,, . β0,,. ,~~~~~β .~~ ~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~,,, ,,
.. ,, β
. β_ ~~~~~~~~~
parent
1 0
14~~~~~~~~~
T 9 1 s -
(
h1
(a) (b )
Figure 11 .5. Representat i on β s - i β a tree.
(a) Tr ee.
(b) Pare nt ar ray for root 1.
76
~
~~~~~~~~~~~~~~~~~~~~~~~~ .~~~~~~ ,. ββs-
c- β ,, .. _ ____
.-,-β~~~~~~~~~~-β -~~~~~~~~~-β
β.β.- -~~~~
A~~~, B ( c - )
3/ D
(
- A(5)
G ~~~~~
~~~ - C ( S ) J )
~~F / β -
H I
\
\ I /
/
F (l
(a) (b )
Figure Ic- j - . Depthβ fir s-βc - s - e s - s -jr c s - s _c - b β s-os -s s-n βi si, u - - β .βS c - β .c c- β s - β 5 s -~ s~.
(a) Graph.
(b ) spanning t r c - c c - s - g c - s - n i c - s - ri t β s - I s - ,~ 3~ - β β
Vertices nocnbered a s - s- u -s- ; S
77
~~~ s- s-,.I , β~~~~~~-β~~-c---. ~~~~~~~.
~~~~~~~~~~~~ _ _ _ _ _ _ _ _ _ - β
β~~ ββ β β β - βr
s-i C
(a)
I i ( β )
B~5~ ~7- s - i
I \
E ( I )
D( 1)
(b)
iβ s - R c -.ns-β 11.7. 5 c - β: i ls - β t β is - β:i . s- ec s-s -β β β i i ofβ a s- .S c - β . r c - β c t e d s-s- ra I l .n .
(a) - b r _ c u I n .
( t o ) βi s-o s - s -o n ng t , r u - c --o Cs- - c oos - ra te s -I bj s - e o c - r s - i n .
1 s - s - e s f ln nβ:,lcered as explored.
- .s-β .. Lββ,,s-. ,,,,. _~ β- ~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~ -β~~~~~ . .,.,.. , - 1,~ β ~~~~~~~~~~~~~ ~~~~~~ __ . ~~~~~~~~~~~~
β~~~~~~~~~~~~~ β s - β J β ~~~~~~~~~~~~~~~~~~~~~~~~ β β -
~~~
A (2) B (2) H(2 ) T (2)
/\\ /G ( l ) D ( l ) / \~~~~~
/
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1~~~~~~~~~~~I( l ) I ( 1)
F ( O )
(a)
H ( 3 )
__-,~-β -*~
β I If E ( 2 )
G ( 2 ) β E β β β ~~~~~~~~~
β --- ββ β ~~s-
__~~~~~~~~~ F( 1S )
C ( 1)
//
D ( O ) β
(b)
Figure 11. . Brea dth- fi rβ .: βt search. Level i nd i ca t ed in j arenthe r es.
(a) Search on β gro ss- in s - n Figure . . o
( b ) Search of grap h in Figure 1k ;;.
79
-~~~~~~~~ -~β~~~~ β , s - s-,,~~~.. , r c -~~ . iβ . -_- ~~β
ββ -β -,.β-. β β~β β β β β βββ . ββ
~~ β β - β β β β .β β ββ'β'
~~~~~~β
~~β β
1. D isc re te Fourier t ranc 1β - c - i~~ (DFT):
nβ s-s-cur: .s- s-cs-i S
2. Matrix multi ili ca t ion (MM) :
u-es-cus-βsion.β1 . Linear equations cfl a iβclan ar graph (LEG):
recur:β.I ccrn , decomposition by -c .β .o s - s - nectivity , breadth- thu-at :e.-sr β.s - : . .
S . c- βs-lob al. flow s-ns-cals-j . 1 . : ( GFA) :
s-le coccsl -as-itt, o:c-n by s- β - β r rnec t7 βc-dt-y , lat h cos-s-cr r e:s- :7-s-n , A s-s-i t h - s-,βlr :t c s -s - s c - s - -c - s - i c - .
5. t at β.bc c-crn mat ching on :t r is-βng:o ( i l - b ) :
data structur es.
St rong c onoot s-anent s ( S C β S :
βlu -i - t i n - fi r s t cuss - s βs - I s - .
i ianarity test ing (P T ) :
c-ju-c c . - t l i β s -.βs t rc βt cs - s - cc -s-u-c-s-i n .
cs - . b-is-ed,~~~n - network 5 . 1βs- s-c (5- β7~~~) :
auguentat i on , c r - s- a s-I t Sβs-first s- βs- c_crcts-.
β c- . Gra s- I, βitching (CS-b
au~~~entation , breadth-first cβs-a rch , cycle shrink Ing .
10. Set union (SU) :
i ath cc cnuip r βoc s- ..I βu - .
~L 1e 5 . 1 . Ten Tr ac βtaLl&c- ir β:blβ.ssccβ.: and 5-bcs -βthcβJsβ β, β cr s- S olv i ng βCI β sc - s- nr c -.
β ~~~~~~~~~~~~~~~~~~ _s - . - .,,__~~~ ββ.c . . β .s-,,. i__ s-
~~~~~~~~~~~~~~~~~~~~~~~~ β~~~-
β β β , ββ β~β I β
c - u - s - - s - s -ou r s -t i al
βh n5
C-ru-on
0~n ββ-
-Ic-
1/~~ c 1 g n Β±
1 1 , 1
1
n
is-~ Β°~ LEG GFA 11- β DC I T l-15t7 s- ic - β
β i gss-r βc 5. 1. Recent complexfty improvements.
n = s .i sos - - (number of vert ices in graph lcr cc -b i o c- n s - n o c ) .
m = u - c - c c - s - - c - u s - c l par ameter (number of edges in gr c t 5cc in r . c- bic -s--s- n is - ) .
β~~~~ s- β~~ .,,s- ~~~~~~~~~~~~~~~~ ~~., ~~~~~~~~~~~~ ~~~~~~~~~~~~ s-s-~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~
- _s- - s--,-~~~
β - ~~~~~~~~~~
- TT~__i~~~~ ~~~~~~~~~~~~~~~~ ~~1~~~~~~__
N -
I. ~~~~~~~~~~ ~~~~~~ β β β. -
. β β
L ~~~~~~~~~ ~~~_. β β - - .
~~β
/- -- β
_ _ _ _ ~~~ β β Jr
~β1I s - c -β β 5 . . , s - c -.~ c - c s - s o , o β . : 1 β β
β1 β β1 β sc - β n β c- f e C u - c - I ~~~~~~ 0β -s - ilβ , βi s - i , .
0. -c-
L ~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~ . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ β’
. ~~ s- . . s-~~~~~~~~~~~~~~~s- . ,~~~~~~~~~ , , ~~~~~~~~~~~~~~~~~~
-- - - ~~~~~~~~ β ~~~~~ β- β -~~~ - β β~~~~~~~~
(a) (b)
Figure 5 . 3 . Ku r a ton ws-Gci :.i Ls-5r a, scβ. β .
(a) K5.
(b ) K~~~ .
s- s-7
-
~
--
~
s-- - β- ~~~~~~~~~~~~~~~ - - . ,- ---.- _ -s-~~~~~~~~~~- --- ~~~~~~~~~~~~~~~~~~~~~~ L. , , ~~~~~~~~ β .
s-t
o - A( t - β)
Figure 5.1, . i- s - c - s - i r- . sets-β s-i s- I c - c , c- β Set A f - i , :c , C . 0 , C. 5 .
i s - i β1 5 n β - - - . s-Si:: -, β of β βc c-s- c,, S . c-i s-is-βs-s - n I tS iu βo- c: .
L ~~~~~~~~~~~~ . . . ~~~~~~~~
β .β.Β± . ,,
~~~~~~~_ _ _ _
uni an ( A , B ) :
~~~~~~~~~~~~~~
A (xΒ±y(a)
if x < y
A ( x+y)
( b )
Fi gure 5.5. Irs -c -n i, c-c βs β sc - oβn b s - , a t j s -r n - s - f β acG c - c s .
(a ) Ba: i,βs βs - , β t
(b) Weight ing heuristic .
β.l~
- ,, - β β βββ--β_ βs-β-- -. - . -ββββ-β- β-- , , ~~~~~~~~~~~~~~~~~~~~
β ββ ~β -~-β - -~--βββ - β ββ. β.--β- ~~
- -. β. -β- β~
β - _~~~~~c- s- - - ~β~~0 - β
~~ ~~~~ s-s-β~~ β β _UIUI__U.U ___.._ _._ .U_uuuu1
\i7
C
7
A(o β o )
s - s - I c -β- i Sβ s- 5 . . lu -_ c βc - SOs - . β β β c- c-β . 0 0 0 β . βC s- is- s-- s-i β i β l l I s - c - c -β., β o n c-., c u n - - c-s-- s - s i o n .
I ~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ .. ----
, . s- s- c- -
~~~~~~~~~~~~~~~