Skip to content

A comparison between several reordering algorithms for minimal non-zero fill-ins in the Cholesky factorization when solving Ax=b.

License

Notifications You must be signed in to change notification settings

MehdiPaak/Sparse-Matrix-Reordering

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Sparse-Matrix-Reordering

A comparison between several reordering algorithms for minimal non-zero fill-ins in the Cholesky factorization when solving Ax=b.

The reordering algorithms implemented are:

  • Minimum Degree (MD)
  • multiple Minimum Degree (MMD)
  • Nested Dissection (ND)
  • Metis library Nested Dissection

Timing and the number of non-zeros in the factored matrix is reported.

Usage:

MD.exe -filemat=CoeffMatrix.dat -filerhs=Rhs.dat -reordering=<0/1/2/3/4> Example data files are provided in the sample data.

-filemat: name of the binary file containing coeeficient matrix in compressed row storage (CRS) format, the number of rows/equations is specified first then the pointer vector, column index vector and finally values

-filerhs: file containing the right hand side to the equation A x= b (not necessary for reordering)

-reordering: 0: no reordring 1: Minimum Degree (MD) 2: Multiple minimum degree (MMD) 3: Nested Disectipon (ND) 4: Nested Disection Metis library implementation

  • compare the number of nonzeros in the factorized matrix for comparing the efficiency.

Based on the reordering algorithms proposed by Alan George and JWH Liu. See https://epubs.siam.org/doi/abs/10.1137/1031001

By: Mehdi Paak

About

A comparison between several reordering algorithms for minimal non-zero fill-ins in the Cholesky factorization when solving Ax=b.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages