Title: Parallel Implementation for SemiDefinite Programming with Positive Matrix Completion Method
1Parallel Implementation for SemiDefinite
Programming withPositive Matrix Completion Method
- Kanagawa University
- Dept. of Industrial Eng. and Management
- Makoto Yamashita
- Tokyo-Tech Masakazu Kojima Tokyo Denki Univ.
Katsuki Fujisawa Tokyo-Tech Kazuhide Nakata
2Motivation
- SDP (SemiDefinite Program) has many applications
- SDPs become extremely large
- Solve larger SDPs in shorter time
- Utilization of Parallel Computation
- SDPARA-C SDPARA (SemiDefinite Programming
Algorithm paRAllel version) with the completion
method
3Contents of this talk
- Definition of SDP
- Concepts of Parallel-ism
- The Completion Method and SDPARA-C
- Conclusion
4SDPs arisen from
- Control Theory
- Combinatorial Optimization
- Data Mining
- Quantum Chemistry
- Successive Relaxation Methods for Non-convex
optimization problem - and more direct/indirect applications
The Completion Method
5Standard form of SDP
6Primal-Dual Interior-Point Methods
7Computation for Search Direction
2.CHOLESKY
Schur complement matrix ? Cholesky Factorizaiton
Schur complement equation
3.PMATRIX
1.ELEMENTS
4.DENSE (dense matrix computation)
8Bottlenecks on Single Processor
Parallelize in SDPARA
SDPARA-C
Computation time in second
9Row-wise distribution for evaluation of the Schur
complement matrix
Symmetric Matrix4 Processors No
Communication Simple memory distribution
10Redistribution of Schur complement matrix for
parallel Cholesky factorization
11Numerical results ontheta 6 Lovászs theta
function(m 4375, n300)
15
5
12LiF Quantum Chemistry(m15313, n496)
13maxG51 Max-Cut Problem(m 2000, n2000)
14Drawbacks of SDPARA
- Ineffective on Max-Cut Problem
- PMATRIX,DENSE are bottlenecks
- Full-dense in primal variable matrix
- Inheritance of sparsity in dual
- m large, n medium size
15Sparsity in Dual
For example, Primal is
Dual is
16The Completion Method
- Introduce Sparsity in Primal Variable
- Nakata-Fujisawa-Fukuda-Kojima-Murota
- Positive Definite Matrix Completion
- Chordal Graph
- Sparse Cholesky Factorization
2003
17Aggregate Sparsity Pattern
All elements are required for PSD
Enough for objective function and equality
constraints
Aggregate Sparsity Pattern
18Positive Definite Matrix Completion
Positive Definite Matrix Completion Assign values
not in to be positive definite
19Chordal Graph
1 2 3 4 5 6 7
1 2 3 4 5 6 7
Chordal Graph
Length of minimal cycle ? 3
Aggregate Sparsity Pattern
Insufficient for Positive Definite Matrix
Completion
20Extended Sparsity Pattern
Chordal Graph
21Sparse Cholesky Factorization
- Instead of dense matrices
- Store such that
- The sparse structure of is determined by
Chordal Graph - Modify the Schur complement
22Performance of Completion Methods
PARALLEL
PARALLEL
PARALLEL
BETTER
WORSE
PARALLEL
Max Clique Problem m1891, n 1000, 64 processors
23SDPARA-C
- SDPARA Completion Method
- Nakata-Yamashita-Fujisawa-Kojima 2003
- Parallel Computation for
- ELEMENTS
- CHOLESKY (same as SDPARA)
- PMATRIX
24Based on Row-wise distributionfor ELEMENTS
In SDPARA-C, Evaluation of Schur complement
matrix is based on Row-wise distribution
Sometime, Load-balance becomes worse in the
combination with the completion methods
25Break-down of Row-wise Distribution
- Computation cost concentrates on inverse
multiplications - Strong dependency on the number of non-zero
column vectors - In Max Clique Problem
- has non-zero vectors
- Each have only one nonzero
vectors - Work of the Processor for 1st row is heavy
26Hashed row-wise distribution
- Divide non-zero vectors into subsets
- th processor computes
- Accumulate results with network communication
27Simple row-wise and Hashed row-wise
Communication
28Parallel Computation for PMATRIX
- Column-wise computation of
- Each column requires same computation cost
- Non-cyclic distribution
29Performance of SDPARA-C
PARALLEL
Completion
Max Clique Problem m1891, n 1000, 64 processors
30SDPARA,SDPARA-C,PDSDP for norm.10.990(m11,n1000,
?21)
PDSDP (Benson,2002)
31Memory Reduction in SDPARA-C
- Schur complement matrix on distributed memory
- Memory reduction for owing to the
completion method - SDPARA-C can handle extremely large SDPs
involving both many equality constraints and
large-scale matrix (m,n ? 40000 on Max-Cut)
32SDPs from SDPLIB and DIMACS
2017
1958
X
X
memoryover
memoryover
33Conclusion
- How to solve large SDPs in a short time
- The Completion Method
- Hashed Row-wise for ELEMENTS
- Parallel for PMATRIX
- SDPARA-C solves extremely large SDPs
- Satisfying scalability and load-balance
34Future Directions
- Combination of SDPARA, SDPARA-C
- SDPARA Dense or Medium
- SDPARA-C Sparse and Large
- Numerical Stability
- Application
- Quadratic Assignment Problem
- Network Location Problem
35Optimization in Quantum Chemistry
Nakata et al (2003), Zhao et al (2004)
- Compute the ground-state energy of some molecules
- N-representability and Reduced Density Matrix
(Kernel Matrix) - Hilbert Space
- Ray-Leigh ratio
Energy
Excited states
n4
n3
n2
n1
Stable state
ground state energy
36Quantum Chemistry
- Prepare basic wave functions
- Find a wave function which minimizes the ground
state energy
Hamiltonian
37SDP from Quantum Chemistry
- The ground state energy is an optimal value of
the SDP
Total electron numbers
Each basic wave function contains an
electronwith probability between 0,1
where
38Methods for the ground state energy
- Self Consistent Field
- Configuration Interaction
- Full CI
- Single-Double CI
- SDP relaxation
Energy
Single-Double CI
Full CI
The ground state energy
SDP relaxation
39Numerical Results of Accuracyfor SDPs arisen
from Quantum Chemistry
N of electron, 2K of orbits ?SDP
difference bet. SDP and Full-CI ?SDCI difference
bet. Singly-Double excited CI and
Full-CI Full-CI Accurate value with enormous
computation time