the number of 2-protected nodes in tries and patricia...
TRANSCRIPT
![Page 1: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/1.jpg)
The Number of 2-Protected Nodes in Triesand PATRICIA Tries
Guan-Ru Yu
Department of Applied MathematicsNational Chiao Tung University
August 2nd, 2014
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 1 / 34
![Page 2: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/2.jpg)
Three main classes of digital trees
Digital search trees (DSTs)
Tries
PATRICIA tries
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 2 / 34
![Page 3: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/3.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 4: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/4.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 5: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/5.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 6: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/6.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 7: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/7.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 8: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/8.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 9: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/9.jpg)
Digital search trees (DSTs)
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1
R2
0
R3
1
R4
1
R5
0
R6
1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 3 / 34
![Page 10: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/10.jpg)
Tries
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1 R2
R4 R5
R3 R6
0 1
01 0
1
0 1
0 1
0
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 4 / 34
![Page 11: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/11.jpg)
PATRICIA tries
R1 = 000001 · · ·R2 = 000110 · · ·R3 = 110111 · · ·R4 = 011011 · · ·R5 = 100001 · · ·R6 = 111110 · · ·
R1 R2
R4 R5
R3 R6
0 1
01 0
1
0 10 1
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 5 / 34
![Page 12: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/12.jpg)
Random model
(In probability theory) A sequence of RVs is independent andidentically distributed (iid): same probability distribution and all aremutually independent.
X1, X2, . . . is a random string: X1, X2, . . . is an iid sequence of RVswith P (Xn = 0) = p and P (Xn = 1) = q := 1− p.
A digital tree is a random digital tree of size n: constructed from ninfinite {0, 1}-random strings tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 6 / 34
![Page 13: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/13.jpg)
Size
Size of digital trees: number of internal nodes.
1
2 3
4 5 6
7
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 7 / 34
![Page 14: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/14.jpg)
Size
Size of digital trees: number of internal nodes.
1
2 3
4 5 6
7
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 7 / 34
![Page 15: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/15.jpg)
Size of tries
Mean: Knuth (1973).
Variance: Two groups independently.
1. P. Kirschenhofer and H. Prodinger: symmetric case (p = q = 1/2)with explicit expressions for involved constants and periodic functions(1991).
2. P. Jacquet and M. Regnier: general case (symmetric andasymmetric case) but without explicit expressions for involvedconstants and periodic functions (1988)
partial results on explicit expressions of involved constants andperiodic functions (1989).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 8 / 34
![Page 16: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/16.jpg)
Size of tries
Mean: Knuth (1973).
Variance: Two groups independently.
1. P. Kirschenhofer and H. Prodinger: symmetric case (p = q = 1/2)with explicit expressions for involved constants and periodic functions(1991).
2. P. Jacquet and M. Regnier: general case (symmetric andasymmetric case) but without explicit expressions for involvedconstants and periodic functions (1988)
partial results on explicit expressions of involved constants andperiodic functions (1989).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 8 / 34
![Page 17: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/17.jpg)
2-protected nodes
2-protected nodes: nodes that have distance at least 2 from leafs.
1
2
3
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 9 / 34
![Page 18: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/18.jpg)
2-protected nodes
2-protected nodes: nodes that have distance at least 2 from leafs.
1
2
3
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 9 / 34
![Page 19: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/19.jpg)
2-protected nodes in tries
Mean: J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012).
Variance: J. Gaither and M. D. Ward (2013).
Some mistakes:
1. wrong error term.2. forgot to pull out the average value.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 10 / 34
![Page 20: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/20.jpg)
2-protected nodes in tries
Mean: J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012).
Variance: J. Gaither and M. D. Ward (2013).
Some mistakes:
1. wrong error term.2. forgot to pull out the average value.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 10 / 34
![Page 21: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/21.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 22: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/22.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 23: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/23.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 24: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/24.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 25: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/25.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 26: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/26.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 27: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/27.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 28: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/28.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 29: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/29.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 11 / 34
![Page 30: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/30.jpg)
Main goal of thesis
Re-derive (and correcte) the results of J. Gaither, Y. Homma, M.Sellke and M. D. Ward by using the method of M. Fuchs, H.-K.Hwang, and V. Zacharovas.
Prove the conjectured central limit theorem.
Derive a bivariate central limit theorem of the number of internalnodes and the number of 2-protected nodes in random tries. Thisresult contains the central limit theorem for PATRICIA tries.
Derive asymptotic expansions for the number of 2-protected nodes inPATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 12 / 34
![Page 31: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/31.jpg)
Main goal of thesis
Re-derive (and correcte) the results of J. Gaither, Y. Homma, M.Sellke and M. D. Ward by using the method of M. Fuchs, H.-K.Hwang, and V. Zacharovas.
Prove the conjectured central limit theorem.
Derive a bivariate central limit theorem of the number of internalnodes and the number of 2-protected nodes in random tries. Thisresult contains the central limit theorem for PATRICIA tries.
Derive asymptotic expansions for the number of 2-protected nodes inPATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 12 / 34
![Page 32: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/32.jpg)
Main goal of thesis
Re-derive (and correcte) the results of J. Gaither, Y. Homma, M.Sellke and M. D. Ward by using the method of M. Fuchs, H.-K.Hwang, and V. Zacharovas.
Prove the conjectured central limit theorem.
Derive a bivariate central limit theorem of the number of internalnodes and the number of 2-protected nodes in random tries. Thisresult contains the central limit theorem for PATRICIA tries.
Derive asymptotic expansions for the number of 2-protected nodes inPATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 12 / 34
![Page 33: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/33.jpg)
Main goal of thesis
Re-derive (and correcte) the results of J. Gaither, Y. Homma, M.Sellke and M. D. Ward by using the method of M. Fuchs, H.-K.Hwang, and V. Zacharovas.
Prove the conjectured central limit theorem.
Derive a bivariate central limit theorem of the number of internalnodes and the number of 2-protected nodes in random tries. Thisresult contains the central limit theorem for PATRICIA tries.
Derive asymptotic expansions for the number of 2-protected nodes inPATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 12 / 34
![Page 34: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/34.jpg)
Additive shape parameters
A general framework for obtaining asymptotic expansions of meanand variance of additive shape parameter in random tries with explicitexpressions for periodic functions.
XIn X∗n−InXn
splitting
By splitting the trie,
Xnd= XIn +X∗n−In + Tn.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 13 / 34
![Page 35: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/35.jpg)
Additive shape parameters
A general framework for obtaining asymptotic expansions of meanand variance of additive shape parameter in random tries with explicitexpressions for periodic functions.
XIn X∗n−InXn
splitting
By splitting the trie,
Xnd= XIn +X∗n−In + Tn.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 13 / 34
![Page 36: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/36.jpg)
Additive shape parameters
A general framework for obtaining asymptotic expansions of meanand variance of additive shape parameter in random tries with explicitexpressions for periodic functions.
XIn X∗n−InXn
splitting
By splitting the trie,
Xnd= XIn +X∗n−In + Tn.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 13 / 34
![Page 37: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/37.jpg)
Main steps of the analysis
The first step is to take moments on both sides of the distributionalrecurrence for Xn. This gives
an =∑
0≤k≤nπn,k(ak + an−k) + bn
for all moments, where bn is a function of moments of lower order.
The second step is poissonization. Let
f(z) := e−z∑n
anzn
n!, g(z) := e−z
∑n
bnzn
n!.
We havef(z) = f(pz) + f(qz) + g(z).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 14 / 34
![Page 38: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/38.jpg)
Main steps of the analysis
The first step is to take moments on both sides of the distributionalrecurrence for Xn. This gives
an =∑
0≤k≤nπn,k(ak + an−k) + bn
for all moments, where bn is a function of moments of lower order.
The second step is poissonization. Let
f(z) := e−z∑n
anzn
n!, g(z) := e−z
∑n
bnzn
n!.
We havef(z) = f(pz) + f(qz) + g(z).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 14 / 34
![Page 39: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/39.jpg)
Main steps of the analysis
The first step is to take moments on both sides of the distributionalrecurrence for Xn. This gives
an =∑
0≤k≤nπn,k(ak + an−k) + bn
for all moments, where bn is a function of moments of lower order.
The second step is poissonization. Let
f(z) := e−z∑n
anzn
n!, g(z) := e−z
∑n
bnzn
n!.
We havef(z) = f(pz) + f(qz) + g(z).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 14 / 34
![Page 40: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/40.jpg)
Main steps of the analysis
The third step is doing Mellin transform on both sides. Therefore
M [f(z); s] :=
∫ ∞0
f(z)zs−1dz.
We have
M [f(z); s] =M [g(z); s]
1− p−s − q−s
The fourth step is using inverse Mellin transform
f(z) =1
2πi
∫↑
M [g(z); s]z−s
1− p−s − q−sds
and applying the converse mapping theorem.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 15 / 34
![Page 41: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/41.jpg)
Main steps of the analysis
The third step is doing Mellin transform on both sides. Therefore
M [f(z); s] :=
∫ ∞0
f(z)zs−1dz.
We have
M [f(z); s] =M [g(z); s]
1− p−s − q−s
The fourth step is using inverse Mellin transform
f(z) =1
2πi
∫↑
M [g(z); s]z−s
1− p−s − q−sds
and applying the converse mapping theorem.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 15 / 34
![Page 42: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/42.jpg)
Main steps of the analysis
The third step is doing Mellin transform on both sides. Therefore
M [f(z); s] :=
∫ ∞0
f(z)zs−1dz.
We have
M [f(z); s] =M [g(z); s]
1− p−s − q−s
The fourth step is using inverse Mellin transform
f(z) =1
2πi
∫↑
M [g(z); s]z−s
1− p−s − q−sds
and applying the converse mapping theorem.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 15 / 34
![Page 43: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/43.jpg)
Main steps of the analysis
The last step is depoissonization which is based on the Poissonheuristic
f(n) = E(aN ) ∼ an.
To make this step precise, we use Cauchy’s integral formula
an =n!
2πi
∮|z|=r
z−n−1ez f(z)dz
and the saddle-point method.
All the above steps can be merged via
an =n!
2πi
∫↑
M [g(z); s]
(1− p−s − q−s)Γ(n+ 1− s)ds.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 16 / 34
![Page 44: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/44.jpg)
Main steps of the analysis
The last step is depoissonization which is based on the Poissonheuristic
f(n) = E(aN ) ∼ an.
To make this step precise, we use Cauchy’s integral formula
an =n!
2πi
∮|z|=r
z−n−1ez f(z)dz
and the saddle-point method.
All the above steps can be merged via
an =n!
2πi
∫↑
M [g(z); s]
(1− p−s − q−s)Γ(n+ 1− s)ds.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 16 / 34
![Page 45: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/45.jpg)
Main steps of the analysis
The last step is depoissonization which is based on the Poissonheuristic
f(n) = E(aN ) ∼ an.
To make this step precise, we use Cauchy’s integral formula
an =n!
2πi
∮|z|=r
z−n−1ez f(z)dz
and the saddle-point method.
All the above steps can be merged via
an =n!
2πi
∫↑
M [g(z); s]
(1− p−s − q−s)Γ(n+ 1− s)ds.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 16 / 34
![Page 46: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/46.jpg)
Poisson-Mellin-Newton cycle
an
an =∑
0≤k≤n
πn,k(ak + an−k) + bn
f(z) = f(pz) + f(qz) + g(z)
M[f(z); s] =M[g(z); s]
1− p−s − q−s
f(z) =1
2πi
∫↑
M[g(z); s]z−s
1− p−s − q−sds
an =n!
2πi
∮|z|=r
z−n−1
ezf(z)dz
Recurrence relation
Poisson generating function
Mellin transform
Inverse Mellin transform
Analytic de-Poissonization
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 17 / 34
![Page 47: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/47.jpg)
Variance
The definition of the variance is
V(Xn) = E(X2n)− (E(Xn))2.
P. Jacquet and M. Regnier used the following
W (z) := f2(z)− f1(z)2
M. Fuchs, H.-K. Hwang and V. Zacharovas pointed out that a betterchoice is
V (z) := f2(z)− f1(z)2 − zf1(z)2.
We will make use of this in our analysis of 2-protected nodes in triesand PATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 18 / 34
![Page 48: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/48.jpg)
Variance
The definition of the variance is
V(Xn) = E(X2n)− (E(Xn))2.
P. Jacquet and M. Regnier used the following
W (z) := f2(z)− f1(z)2
M. Fuchs, H.-K. Hwang and V. Zacharovas pointed out that a betterchoice is
V (z) := f2(z)− f1(z)2 − zf1(z)2.
We will make use of this in our analysis of 2-protected nodes in triesand PATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 18 / 34
![Page 49: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/49.jpg)
Variance
The definition of the variance is
V(Xn) = E(X2n)− (E(Xn))2.
P. Jacquet and M. Regnier used the following
W (z) := f2(z)− f1(z)2
M. Fuchs, H.-K. Hwang and V. Zacharovas pointed out that a betterchoice is
V (z) := f2(z)− f1(z)2 − zf1(z)2.
We will make use of this in our analysis of 2-protected nodes in triesand PATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 18 / 34
![Page 50: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/50.jpg)
2-protected nodes in tries
Recall: Xn can be described by
Xnd= XIn +X∗n−In + Tn.
By splitting the trie,
X(T )n
d=
{X
(T )n−1, when In = 1 or In = n− 1;
X(T )In
+X(T )∗n−In + 1, otherwise,
with initial conditions X(T )0 = X
(T )1 = 0. So,
Tn =
{0, when In = 1 or In = n− 1;
1, otherwise.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 19 / 34
![Page 51: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/51.jpg)
2-protected nodes in tries
Recall: Xn can be described by
Xnd= XIn +X∗n−In + Tn.
By splitting the trie,
X(T )n
d=
{X
(T )n−1, when In = 1 or In = n− 1;
X(T )In
+X(T )∗n−In + 1, otherwise,
with initial conditions X(T )0 = X
(T )1 = 0. So,
Tn =
{0, when In = 1 or In = n− 1;
1, otherwise.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 19 / 34
![Page 52: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/52.jpg)
Mean of 2-protected nodes in tries
With χk = 2rkπilog p and log p
log q = rl and (r, l) = 1, let
F [G](x) :=
1
h
∑k∈Z\{0}
G(−1 + χk)e2kπix, if log p
log q ∈ Q;
0, if log plog q /∈ Q,
Theorem
We have
E(X(T )n )
n=pq + 1− h
h+ F [G
(T )1 ](r log1/p n) + o(1),
where G(T )1 (−1 + χk) = Γ(−1 + χk)χk(χkpq − pq − 1).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 20 / 34
![Page 53: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/53.jpg)
Mean of 2-protected nodes in tries
With χk = 2rkπilog p and log p
log q = rl and (r, l) = 1, let
F [G](x) :=
1
h
∑k∈Z\{0}
G(−1 + χk)e2kπix, if log p
log q ∈ Q;
0, if log plog q /∈ Q,
Theorem
We have
E(X(T )n )
n=pq + 1− h
h+ F [G
(T )1 ](r log1/p n) + o(1),
where G(T )1 (−1 + χk) = Γ(−1 + χk)χk(χkpq − pq − 1).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 20 / 34
![Page 54: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/54.jpg)
Variance of 2-protected nodes in tries
TheoremWe have
V(X(T )n )
n=G
(T )2 (−1)
h+ F [G
(T )2 ](r log1/p n) + o(1).
where G(T )2 (−1 + χk) is
Γ(3 + χk)(−2−3−χkp2q2
)+ Γ(χk + 2)
(2p2q(1 + p)−2−χk + 2pq2(1 + q)−2−χk
)+ Γ(χk + 1)
(−3pq + 2−χkpq − 2−1−χk
)+ Γ(χk)
(1− 2p(1 + p)−χk − 2q(1 + q)−χk
)+ Γ(χk − 1)
(1− 21−χk
)−
1
h
∑j∈Z
(χj − 1)G(T )1 (χj − 1)(−1 + χk−j)G
(T )1 (−1 + χk−j)
+ 2∑`≥1
(−1)`
`!
p1+` + q1+`
1− p1+` − q1+`K1(`+ χk − 1)K1(−1− `)Γ(`+ χk)
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 21 / 34
![Page 55: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/55.jpg)
Variance of 2-protected nodes in tries
Theorem (continued)
+ 2∑`≥2
(−1)`
`!
K1(−`)1− p` − q`
(pq(p` + q`)(`− 1)Γ(χk + `+ 1) +
(1− `(p` + q`)(2pq + p−`+2
+ q−`+2))
Γ(χk + `) + (p` + q`)(1− `+ p−`+2`+ q−`+2`)Γ(χk + `− 1)
)
and G(T )2 (−1) is
2p2q
(1 + p)2+
2pq2
(1 + q)2− 2pq −
p2q2
4+ 2p log(1 + p) + 2q log(1 + q) +
1
2+ h− 2 log 2
+ 2∑`≥2
(−1)`K1(−`)
1− p` − q`
((p` + q`)
(1− `+ p−`+2`+ q−`+2`
`(`− 1)+ pq`− 3pq − p−`+2
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 22 / 34
![Page 56: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/56.jpg)
Variance of 2-protected nodes in tries
Theorem (continued)
− q−`+2
)+
1
`
)+ 2
∑`≥1
(−1)`
`
p1+` + q1+`
1− p1+` − q1+`K1(`− 1)K1(−1− `)−
1
h(pq + 1− h)2
−
1
h log p
∑j≥1
4rjπ2
sinh(
2rjπ2
log p
) (p2q2 (2rjπ
log p
)2
+ (pq + 1)2
), if log p
log q∈ Q;
0, if log plog q
/∈ Q.
Here, we setK1(s) := −1 + pqs(s+ 1)− p−ss− q−ss
Our expression is different from J. Gaither and M. D. Ward.
We give an explanation why Abel summability gives the correct resultin the expression above.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 23 / 34
![Page 57: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/57.jpg)
Variance of 2-protected nodes in tries
Theorem (continued)
− q−`+2
)+
1
`
)+ 2
∑`≥1
(−1)`
`
p1+` + q1+`
1− p1+` − q1+`K1(`− 1)K1(−1− `)−
1
h(pq + 1− h)2
−
1
h log p
∑j≥1
4rjπ2
sinh(
2rjπ2
log p
) (p2q2 (2rjπ
log p
)2
+ (pq + 1)2
), if log p
log q∈ Q;
0, if log plog q
/∈ Q.
Here, we setK1(s) := −1 + pqs(s+ 1)− p−ss− q−ss
Our expression is different from J. Gaither and M. D. Ward.
We give an explanation why Abel summability gives the correct resultin the expression above.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 23 / 34
![Page 58: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/58.jpg)
Variance of 2-protected nodes in tries
Theorem (continued)
− q−`+2
)+
1
`
)+ 2
∑`≥1
(−1)`
`
p1+` + q1+`
1− p1+` − q1+`K1(`− 1)K1(−1− `)−
1
h(pq + 1− h)2
−
1
h log p
∑j≥1
4rjπ2
sinh(
2rjπ2
log p
) (p2q2 (2rjπ
log p
)2
+ (pq + 1)2
), if log p
log q∈ Q;
0, if log plog q
/∈ Q.
Here, we setK1(s) := −1 + pqs(s+ 1)− p−ss− q−ss
Our expression is different from J. Gaither and M. D. Ward.
We give an explanation why Abel summability gives the correct resultin the expression above.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 23 / 34
![Page 59: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/59.jpg)
V(X(T )n )/n
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 24 / 34
![Page 60: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/60.jpg)
Central limit theorem
Theorem (CLT)
We have,
X(T )n − E(X
(T )n )√
V(X(T )n )
d−→ N(0, 1),
whered−→ denotes convergence in distribution and N(0, 1) is the standard
normal distribution.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 25 / 34
![Page 61: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/61.jpg)
2-protected nodes in PATRICIA tries
By splitting the PATRICIA trie,
X(P )n
d=
X
(P )n , when In = 0 or n;
X(P )n−1, when In = 1 or n− 1;
X(P )In
+X(P )∗n−In + 1, otherwise,
with initial conditions X(P )0 = X
(P )1 = 0. So,
Tn =
{0, if In ∈ {0, 1, n− 1, n};1, otherwise.
By the same analysis as for tries, we get the results of PATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 26 / 34
![Page 62: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/62.jpg)
2-protected nodes in PATRICIA tries
By splitting the PATRICIA trie,
X(P )n
d=
X
(P )n , when In = 0 or n;
X(P )n−1, when In = 1 or n− 1;
X(P )In
+X(P )∗n−In + 1, otherwise,
with initial conditions X(P )0 = X
(P )1 = 0. So,
Tn =
{0, if In ∈ {0, 1, n− 1, n};1, otherwise.
By the same analysis as for tries, we get the results of PATRICIA tries.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 26 / 34
![Page 63: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/63.jpg)
Mean of 2-protected nodes in PATRICIA tries
Theorem
We haveE(X
(P )n )
n=pq
h+ F [G
(P )1 ](r log1/p n) + o(1),
where G(P )1 (−1 + χk) = pqΓ(χk + 1).
Here, we have the following relationship between 2-protected nodes intries and PATRICIA tries via the number of internal nodes:
X(P )n = X(T )
n −Nn + n− 1.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 27 / 34
![Page 64: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/64.jpg)
Mean of 2-protected nodes in PATRICIA tries
Theorem
We haveE(X
(P )n )
n=pq
h+ F [G
(P )1 ](r log1/p n) + o(1),
where G(P )1 (−1 + χk) = pqΓ(χk + 1).
Here, we have the following relationship between 2-protected nodes intries and PATRICIA tries via the number of internal nodes:
X(P )n = X(T )
n −Nn + n− 1.
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 27 / 34
![Page 65: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/65.jpg)
Variance of 2-protected nodes in PATRICIA tries
TheoremWe have
V(X(P )n )
n=G
(P )2 (−1)
h+ F [G
(P )2 ](r log1/p n) + o(1),
where G(P )2 (−1 + χk) is
Γ(3 + χk)(−2−3−χkp2q2
)+ Γ(χk + 2)
(2p2q(1 + p)−2−χk + 2pq2(1 + q)−2−χk − 2−1−χkpq
)+ Γ(χk + 1)
(− 3pq − 2−1−χk (1 + 2pq + p1−χk + q1−χk ) + 2p(1 + q)(1 + p)−1−χk
+ 2q(1 + p)(1 + q)−1−χk)
+ (Γ(χk) + Γ(χk − 1))(2(1 + p)1−χk + 2(1 + q)1−χk − 21−χk
+ (1− 21−χk )(p1−χk + q1−χk )− 3)
+ 2∑`≥2
(−1)`
`!
K1(−`)1− p` − q`
(pq(p` + q`)(`− 1)
Γ(χk + `+ 1) +(
(p` + q`)(`− 2pq`− p−`+2`− q−`+2`− 1) + 1)
Γ(χk + `)
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 28 / 34
![Page 66: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/66.jpg)
Variance of 2-protected nodes in PATRICIA tries
Theorem (continued)
+ (1− p` − q`)Γ(χk + `− 1)
)−
1
h
∑j∈Z
(χj − 1)G(P )1 (χj − 1)(−1 + χk−j)G
(P )1 (−1 + χk−j)
+ 2∑`≥1
(−1)`
`!
p1+` + q1+`
1− p1+` − q1+`K1(`+ χk − 1)K1(−1− `)Γ(`+ χk),
and G(P )2 (−1) is
−p2q2
4−
9pq
2− 1 +
2p(1 + q)
1 + p+
2q(1 + p)
1 + q+
2p2q
(1 + p)2+
2pq2
(1 + q)2
+ 2∑`≥2
(−1)`K1(−`)1− p` − q`
((p` + q`)
(1− p−`+2 − q−`+2 −
1
`− 1+ pq`− 3pq
)
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 29 / 34
![Page 67: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/67.jpg)
Variance of 2-protected nodes in PATRICIA tries
Theorem (continued)
+1
`− 1
)+ 2
∑`≥1
(−1)`
`
p1+` + q1+`
1− p1+` − q1+`K1(`− 1)K1(−1− `)−
p2q2
h
−
1
h log p
∑j≥1
4rjπ2
sinh(
2rjπ2
log p
) (p2q2 (2rjπ
log p
)2
+ p2q2
), if log p
log q∈ Q;
0, if log plog q
/∈ Q.
Here, we set K1(s) := (pqs+ 1− p−s − q−s)(s+ 1).
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 30 / 34
![Page 68: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/68.jpg)
Bivariate central limit theorem
Theorem
We have
Cov(Nn, X(T )n )
n=H2(−1)
h+ F [H2](r log1/p n) + o(1),
where
H2(x) =G
(T )2 (x) +G
(N)2 (x)−G(P )
2 (x)
2.
Theorem
We have,
Σ−1/2n
(Nn − E(Nn)
X(T )n − E(X
(T )n )
)d−→ N(0, I2),
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 31 / 34
![Page 69: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/69.jpg)
Bivariate central limit theorem
Theorem
We have
Cov(Nn, X(T )n )
n=H2(−1)
h+ F [H2](r log1/p n) + o(1),
where
H2(x) =G
(T )2 (x) +G
(N)2 (x)−G(P )
2 (x)
2.
Theorem
We have,
Σ−1/2n
(Nn − E(Nn)
X(T )n − E(X
(T )n )
)d−→ N(0, I2),
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 31 / 34
![Page 70: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/70.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
Unknown
conjectureCLT
C.-K. Lee (2014)
CLTvariance
Unknown
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 32 / 34
![Page 71: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/71.jpg)
Number of 2-protected nodes in digital tree
2-protected nodes
digital search trees
tries
PATRICIA tries
R. R.-X. Du and H. Prodinger (2012)mean
J. Gaither, Y. Homma, M. Sellke and M. D. Ward (2012)
mean
J. Gaither and M. D. Ward (2013)variance
some mistakes
G.-R. Yu (2014)
CLT
C.-K. Lee (2014)
CLTvariance
G.-R. Yu (2014)
mean
variance
CLT
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 33 / 34
![Page 72: The Number of 2-Protected Nodes in Tries and PATRICIA Tries2014comb.math.ntnu.edu.tw/slide/session1/GuanRuYu.pdf · Guan-Ru Yu Department of Applied Mathematics National Chiao Tung](https://reader035.vdocuments.mx/reader035/viewer/2022071001/5fbd8abb8fd7ac254a28f2ba/html5/thumbnails/72.jpg)
Thanks for your attention!
Guan-Ru Yu (NCTU) Number of 2-Protected Nodes August 2nd, 2014 34 / 34