Amesos Interfaces to sparse direct solvers

1
Amesos Interfaces to sparse direct solvers

• October 15, 2003
• 830 930 a.m.
• Ken Stanley

2
Overview

• Background
• Simplifying AX B
• Getting started Amesos_Umfpack
• Amesos options
• Other Amesos interfaces
• Status
• Future Work and Discussion
• Summary

3
Background

• Solvers
• Sandia Users
• Challenges

4
Background - Solvers

• PSPACES Gupta f90 mpi SPD unsupported
• WSSMP Gupta binary mpi general
• WGSMP Gupta binary mpi SPD
• SPOOLES Ashcraftpublic domainc unsupported
• SuperLU Li,Demmelserialcgeneral
• UMFPACK Davisserialcgeneralrobust
• Unnamed Ngparallelf77SPD
• MUMPS Amestoyf90general/SPD
• Ksparse Kundert
• SNL_MPFRONT Bennerparallel
• OTHERS y12M, Boeing, MA57, MA43, DPSLDLT,
  DPSLDU, PARDISO, Shen, Dearholt, SPLU, Lusol,
  LUSOL, XYT

5
Background Sandia Users

• Tuminaro ML, etc.
• Shadid MP Salsa
• Williams Sierra
• Hoekstra Xyce
• Pierson
• Dohrman
• Schultz - GOMA
• Many Aztec
• Many more

6
Background - Challenges

• No single solver dominates
• Different interfaces
• Unique data distribution requirements
• Academic quality
• Sporadic development, bugs, untested features,
  changing interfaces
• Sandia modifications not consistently adopted
• Imperfect heuristics
• Interface changes between revisions

7
Simplifying AX B

• Example
• Clean, consistent interface
• Single point of contact
• Evaluation

8
Simplifying AX BExample

• Epetra_LinearProblem Problem( A, x, b)
• Amesos_Umfpack A_direct( Problem, ParamList )
  
• A_direct.SymbolicFactorization( )
• A_direct.NumericFactorization( )
• A_direct.Solve( )

9
Simplifying AX B Clean, consistent interface

• Amesos_BaseSolver class
• Seperation from underlying solver
• Identical interface for serial and parallel
  solvers
• Nearly identical interface for each solver
• Amesos redistributes the data
• Easy upgrade

10
Simplifying AX B Single point of contact

• Isolate report bugs follow-up
• Prompt bug fixes without maintaining third party
  code
• Encourage developers to adopt Trilinos practices
• Autotools CVS regression testing bug tracking
  etc.
• Regression testing
• Check for memory leaks

11
Simplifying AX B Evaluation

• Solvers
• Platforms
• Matrices
• Features
• Our goal is to find the bugs before you do

12
Getting started Building Umfpack

• Amesos User Guide http//software.sandia.gov/tri
  linos/packages/amesos
• Build UMFPACK
• Download from UMFPACK website
• Modify make.inc (see Trilinos3PL/config/UMFPACKv4
  .1)
• make Check output of example run
• Future Work
• autotool UMFPACK
• Trilinos/packages/UMFPACK ?
• -lumfpack?

13
Getting started Building Amesos_Umfpack

• Amesos User Guide
• http//software.sandia.gov/trilinos/packages/ameso
  s
• Build Amesos_Umfpack
• ./configure enable-amesos enable-amesos-umfpack
  with-trilinos3PLdir/home/userid/Trilinos3PL
• make
• cd amesos/test source AmesosUmfpackShort.exe

14
Getting started Calling Amesos_Umfpack

• Epetra_CrsMatrix A
• Epetra_Vector x, b
• AMESOSParameterList ParamList
• Epetra_LinearProblem Problem( A, x, b)
• Amesos_Umfpack A_direct( Problem, ParamList )
  A_direct.SymbolicFactorization( )
• A_direct.NumericFactorization( )
• A_direct.Solve( )
• Future Work Factory Method, TrilinosParameter
  List

15
Amesos options

• Same structure different data
• Multiple solves
• Blocked right hand sides
• Parameter control
• Different solver
• Matrix format

16
Amesos options Same Structure Different Data

• Epetra_LinearProblem Problem( A, x, b)
• Amesos_Umfpack A_direct( Problem, ParamList )
  A_direct.SymbolicFactorization( )
• for ( )
• // Modifications to the non-zero values of A
• A_direct.NumericFactorization( )
• A_direct.Solve( )

17
Amesos options Multiple Solves

• Epetra_LinearProblem Problem( A)
• Amesos_Umfpack A_direct( Problem, ParamList )
  A_direct.SymbolicFactorization( )
• A_direct.NumericFactorization( )
• for ( )
• Problem.SetLHS(x)
• Problem.SetRHS(b)
• A_direct.Solve( )

18
Amesos Options Blocked right hand sides

• Epetra_CrsMatrix A
• Epetra_MultiVector x, b
• AMESOSParameterList ParamList
• Epetra_LinearProblem Problem( A, x, b)
• Amesos_Umfpack A_direct( Problem, ParamList )
  A_direct.SymbolicFactorization( )
• A_direct.NumericFactorization( )
• A_direct.Solve( )

19
Amesos Options Parameter control

• Epetra_LinearProblem Problem( A, x, b)
• Amesos_Umfpack A_direct( Problem, ParamList )
  A_direct.SymbolicFactorization( )
• ParamList.SetParam( UmfpackOrder, 2 )
• for ( )
• // Modifications to the non-zero values of A
• A_direct.NumericFactorization( )
• A_direct.Solve( )

20
Amesos Options Matrix format

• Interface accepts any Epetra_RowMatrix
• Current implementation accepts any
  Epetra_CrsMatrix

21
Other Amesos Interfaces

• Klu Serial, unblocked, for circuit matrices
• SuperLUdist Distributed limited Amesos
  release
• MUMPS Distributed F90 limited Amesos release
• DSCPACK Distributed Symmetric
• Use distributed code for problems
• that are too large to solve on one processor
  -or-
• take more than one minute to factor
• Others WSMP, Benner, Ksparse, Ng

22
Status

• Nightly testing on four platforms
• Atlantis/IRIX,
• Stratus/OSF,
• Herouxsmp/Linux,
• Paunchy/SunOS
• Full support for three packages Amesos_Umfpack,
  Amesos_Klu and Amesos_Dscpack
• Other packages Epetra_SLU, SuperludistOO,
  Amesos_Mumps, Amesos_Superludist
• One example code example/AmesosDscpack

23
Future Work

• Capabilities
• Ease of use
• Single point of contact

24
Future Work - Capabilities

• Amesos_Superludist
• Amesos_Mumps, Amesos_SuperLU, Amesos_Ksparse,
  Amesos_Benner, Amesos_Ng, Amesos_Wsmp
• Expose capabilities to users
• Support transpose for all solvers
• Support row matrix interface
• Support Epetra_VbrMatrix in DSCPACK
• Extra precision iterative refinement

25
Future Work Ease of use

• Examples (TSF extended TSF core etc.)
• Pre-built libraries? (-lamesos, -lumfpack,
  -lsuperludist)
• Autotools use by third party developers
• Factory Method

26
Future Work Single point of contact

• Encourage developers to adopt Trilinos practices
• Autotools CVS regression testing bug tracking
  nightly testing etc.
• Trilinos package?
• Isolate, report bugs, follow-up
• Migration to latest revisions
• Add testing based on the Amesos_BaseSolver class
• More features
• More matrices
• Parameter control
• Test error conditions

27
Discussion

• Your needs?
• Platforms
• Packages
• Features
• Interface

28
Amesos Value Added

• Simple, consistent interface
• Short learning curve
• Easy to switch solvers
• Interface to third party developers
• Evaluation
• Bug workarounds

29
Summary

• Amesos is driven by your needs
• Tell me what you need
• Give me your matrices
• Simple, consistent interface
• Start with Amesos_Umfpack
• Switch to parallel codes for large matrices
• Matrices that dont fit in memory
• Factorizations that take over a minute
• Experiment with different solvers