conjugate gradient method

28
CONJUGATE GRADIENT METHOD Monica Garika Chandana Guduru

Upload: lieu

Post on 20-Feb-2016

175 views

Category:

Documents


1 download

DESCRIPTION

CONJUGATE GRADIENT METHOD. Monica Garika Chandana Guduru . METHODS TO SOLVE LINEAR SYSTEMS. Direct methods Gaussian elimination method LU method for factorization Simplex method of linear programming Iterative method Jacobi method Gauss-Seidel method - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: CONJUGATE GRADIENT METHOD

CONJUGATE GRADIENT METHOD

Monica GarikaChandana Guduru

Page 2: CONJUGATE GRADIENT METHOD

METHODS TO SOLVE LINEAR SYSTEMSDirect methods Gaussian elimination method LU method for factorization Simplex method of linear programmingIterative method Jacobi method Gauss-Seidel method Multi-grid method Conjugate gradient method

Page 3: CONJUGATE GRADIENT METHOD

Conjugate Gradient methodThe CG is an algorithm for the numerical solution of particular system of linear equations

Ax=b. Where A is symmetric i.e., A = AT and positive

definite i.e., xT * A * x > 0 for all nonzero vectors

If A is symmetric and positive definite then the function

Q(x) = Β½ x`Ax – x`b + c

Page 4: CONJUGATE GRADIENT METHOD

Conjugate gradient methodConjugate gradient method builds up a

solution x*€ Rn

in at most n steps in the absence of

round off errors.Considering round off errors more than n

steps may be needed to get a good approximation of exact solution x*

For sparse matrices a good approximation of exact solution can be achieved in less than n steps in also with round off errors.

Page 5: CONJUGATE GRADIENT METHOD

In oil reservoir simulation, The number of linear equations corresponds to the number of grids of a reservoirThe unknown vector x is the oil pressure of

reservoirEach element of the vector x is the oil

pressure of a specific grid of the reservoir

Practical Example

Page 6: CONJUGATE GRADIENT METHOD

Linear System

Square matrixUnknown vector

(what we want to find)Known vector

Page 7: CONJUGATE GRADIENT METHOD

Matrix Multiplication

Page 8: CONJUGATE GRADIENT METHOD

Positive Definite Matrix

> 0[ x1 x2 … xn ]

Page 9: CONJUGATE GRADIENT METHOD

ProcedureFinding the initial guess for solution π‘₯0Generates successive approximations to π‘₯0Generates residualsSearching directions

Page 10: CONJUGATE GRADIENT METHOD

x0 = 0, r0 = b, p0 = r0

for k = 1, 2, 3, . . . Ξ±k = (rT

k-1rk-1) / (pTk-1Apk-1) step length

xk = xk-1 + Ξ±k pk-1 approximate solution

rk = rk-1 – Ξ±k Apk-1 residual

Ξ²k = (rTk rk) / (rT

k-1rk-1) improvement

pk = rk + Ξ²k pk-1 search direction

Conjugate gradient iteration

Page 11: CONJUGATE GRADIENT METHOD

Iteration of conjugate gradient method is of the form

x(t) = x(t-1) + s(t)d(t) where, x(t) is function of old value of vector x s(t) is scalar step size x d(t) is direction vector Before first iteration, values of x(0), d(0) and g(0) must be set

Page 12: CONJUGATE GRADIENT METHOD

Steps to find conjugate gradient methodEvery iteration t calculates x(t) in four steps : Step 1: Compute the gradient g(t) = Ax(t-1) – b Step 2: Compute direction vector d(t) = - g(t) + [g(t)` g(t) / g(t-1)` g(t-1)] d(t-1) Step 3: Compute step size s(t) = [- d(t)` g(t)]/d(t)’ A d(t)] Step 4: Compute new approximation of x x(t) = x(t-1) + s(t)d(t)

Page 13: CONJUGATE GRADIENT METHOD

Sequential Algorithm1) π‘₯0 = 02) π‘Ÿ0 := 𝑏 βˆ’ 𝐴π‘₯03) 𝑝0 := π‘Ÿ04) π‘˜ := 05) πΎπ‘šπ‘Žπ‘₯ := maximum number of iterations to be done6) if π‘˜ < π‘˜π‘šπ‘Žπ‘₯ then perform 8 to 167) if π‘˜ = π‘˜π‘šπ‘Žπ‘₯ then exit8) calculate 𝑣 = 𝐴𝑝k

9) Ξ±k : = rkT rk/pT

k v10) π‘₯k+1:= π‘₯k + ak pk

11) π‘Ÿk+1 := π‘Ÿk βˆ’ ak v12) if π‘Ÿk+1 is sufficiently small then go to 16end if13) 𝛽k :=( r T

k+1 r k+1)/(rTk π‘Ÿk)

14) 𝑝k+1 := π‘Ÿk+1 +Ξ²k pk

15) π‘˜ := π‘˜ + 116) π‘Ÿπ‘’π‘ π‘’π‘™π‘‘ = π‘₯k+1

Page 14: CONJUGATE GRADIENT METHOD

Complexity analysisTo Identify Data DependenciesTo identify eventual communicationsRequires large number of operationsAs number of equations increases complexity

also increases .

Page 15: CONJUGATE GRADIENT METHOD

Why To Parallelize?Parallelizing conjugate gradient method is a

way to increase its performanceSaves memory because processors only store

the portions of the rows of a matrix A that contain non-zero elements.

It executes faster because of dividing a matrix into portions

Page 16: CONJUGATE GRADIENT METHOD

How to parallelize?For example , choose a row-wise block

striped decomposition of A and replicate all vectors

Multiplication of A and vector may be performed without any communication

But all-gather communication is needed to replicate the result vector

Overall time complexity of parallel algorithm is

Θ(n^2 * w / p + nlogp)

Page 17: CONJUGATE GRADIENT METHOD

Row wise Block Striped Decomposition of a Symmetrically Banded Matrix

( a ) (b )

Page 18: CONJUGATE GRADIENT METHOD

Dependency Graph in CG

Page 19: CONJUGATE GRADIENT METHOD

Algorithm of a Parallel CG on each Computing Worker (cw)1) Receive π‘π‘€π‘Ÿ0,𝑐𝑀 𝐴,𝑐𝑀 π‘₯02) 𝑐𝑀𝑝0 =𝑐𝑀 π‘Ÿ03) π‘˜ := 04) πΎπ‘šπ‘Žπ‘₯ := maximum number of iterations to be done5) if π‘˜ < π‘˜π‘šπ‘Žπ‘₯ then perform 8 to 166) if π‘˜ = π‘˜π‘šπ‘Žπ‘₯ then exit7) 𝑣 =𝑐𝑀 π΄π‘π‘€π‘π‘˜8) π‘π‘€π‘π›Όπ‘˜9) π‘π‘€π·π›Όπ‘˜10) Send π‘π›Όπ‘˜,π·π›Όπ‘˜11) Receive π›Όπ‘˜12) π‘₯π‘˜+1 = π‘₯π‘˜ + π›Όπ‘˜π‘π‘˜13) Compute Partial Result of π‘Ÿπ‘˜+1: π‘Ÿπ‘˜+1 =π‘Ÿπ‘˜ βˆ’ π›Όπ‘˜π‘£14) Send π‘₯π‘˜+1, π‘Ÿπ‘˜+115) Receive π‘ π‘–π‘”π‘›π‘Žπ‘™16) if π‘ π‘–π‘”π‘›π‘Žπ‘™ = π‘ π‘œπ‘™π‘’π‘‘π‘–π‘œπ‘›π‘Ÿπ‘’π‘Žπ‘β„Žπ‘’π‘‘ go to 2317) π‘π‘€π‘π›½π‘˜18) π‘π‘€π·π›½π‘˜19) Send π‘π‘€π‘π›½π‘˜ , π‘π‘€π·π›½π‘˜20) Receive π›½π‘˜21) π‘π‘€π‘π‘˜+1 =𝑐𝑀 π‘Ÿπ‘˜+1 + π›½π‘˜ π‘π‘€π‘π‘˜22) π‘˜ := π‘˜ + 123) Result reached

Page 20: CONJUGATE GRADIENT METHOD

Speedup of Parallel CG on Grid Versus Sequential CG on Intel

Page 21: CONJUGATE GRADIENT METHOD

Communication and Waiting Time of the Parallel CG on Grid

Page 22: CONJUGATE GRADIENT METHOD

We consider the difference between f at the solution x and any other vector p:

1) 21) 21 1) ( )2 2

1 1simplify) ( )2 21 1replace ) ( )2 2

1 12 2

t

t

t t

t t

t t t t

t t

i f

ii f

ii i f f

f f

f f

t

t

t t

t t

p p Ap b p c

x x Ax b x c

x p x Ax b x c p Ap b p c

x p x Ax b x p Ap b p

b x p x Ax x Ax p Ap x Ap

x Ax p A

12

positive-definitess) ( ) 0

t

t

f f

p x Ap

x p A x p

x p

Page 23: CONJUGATE GRADIENT METHOD

Parallel Computation Design– Iterations of the conjugate gradient method can be executed only in sequence, so the most advisable approach is to parallelize the computations, that are carried out at each iteration The most time-consuming computations are the multiplication of

matrix A by the vectors x and d– Additional operations, that have the lower computational complexity order, are different vector processing procedures (inner product, addition and subtraction, multiplying by a scalar).While implementing the parallel conjugate gradient method, it can be used parallel algorithms of matrix-vector multiplication,

Parallel Computation Design

Page 24: CONJUGATE GRADIENT METHOD

β€œPure” Conjugate Gradient Method (Quadratic Case)

0 - Starting at any x0 define d0 = -g0 = b - Q x0 , where gk is the column vector of gradients of the objective function at point f(xk)

1 - Using dk , calculate the new point xk+1 = xk + ak dk , where

2 - Calculate the new conjugate gradient direction dk+1 , according to: dk+1 = - gk+1 + bk dk where

T a k = - gk dk

dkTQdk

b k = gk+1TQd kdkTQd k

Page 25: CONJUGATE GRADIENT METHOD

ADVANTAGES Advantages:1) Gradient is always nonzero and linearly

independent of all previous direction vectors.2) Simple formula to determine the new

direction. Only slightly more complicated than steepest descent.

3) Process makes good progress because it is based on gradients.

Page 26: CONJUGATE GRADIENT METHOD

ADVANTAGESAttractive are the simple formulae for

updating the direction vector.Method is slightly more complicated than

steepest descent, but converges faster.Conjugate gradient method is an Indirect SolverUsed to solve large systemsRequires small amount of memoryIt doesn’t require numerical linear algebra

Page 27: CONJUGATE GRADIENT METHOD

ConclusionConjugate gradient method is a linear solver

tool which is used in a wide range of engineering and sciences applications.

However, conjugate gradient has a complexity drawback due to the high number of arithmetic operations involved in matrix vector and vector-vector multiplications.

Our implementation reveals that despite the communication cost involved in a parallel CG, a performance improvement compared to a sequential algorithm is still possible.

Page 28: CONJUGATE GRADIENT METHOD

ReferencesParallel and distributed computing systems by

Dimitri P. Bertsekas, John N.TsitsiklisParallel programming for multicore and cluster

systems by Thomas Rauber,Gudula RungerScientific computing . An introduction with parallel

computing by Gene Golub and James M.OrtegaParallel computing in C with Openmp and mpi by

Michael J.QuinnJonathon Richard Shewchuk, ”An Introduction to

the Conjugate Gradient Method Without the Agonizing Pain”, School of Computer Science, Carnegie Mellon University, Edition 1 1/4