Todo List

Class PEXSI::DistSparseMatrix< F >

Add the parameter of numColLocal

Member PEXSI::PMatrix< T >::PreSelInv ()

Move documentation to a more proper place and update the information.

Member PEXSI::PMatrix< T >::SelInv ()

Move documentation to a more proper place and update the information.

Member PEXSI::PMatrixUnsym< T >::PreSelInv ()

Move documentation to a more proper place and update the information.

Member PEXSI::PMatrixUnsym< T >::SelInv ()

Move documentation to a more proper place and update the information.

Member PEXSI::PSelInvOptions::maxPipelineDepth

This option should not be here and should be moved into PMatrix.

Member PEXSI::SuperLUMatrix< Complex >::Distribute ()

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< Complex >::DistSparseMatrixToSuperMatrixNRloc (DistSparseMatrix< Complex > &sparseA, const SuperLUOptions &options)

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< Complex >::NumericalFactorize ()

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< Real >::Distribute ()

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< Real >::DistSparseMatrixToSuperMatrixNRloc (DistSparseMatrix< Real > &sparseA, const SuperLUOptions &opt)

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< Real >::NumericalFactorize ()

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< T >::Distribute ()

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< T >::DistSparseMatrixToSuperMatrixNRloc (DistSparseMatrix< T > &sparseA, const SuperLUOptions &options)

Better way to incorporate both real and complex arithmetic.

Member PEXSI::SuperLUMatrix< T >::NumericalFactorize ()

Better way to incorporate both real and complex arithmetic.

Member PPEXSISolveInterface (int nrows, int nnz, int nnzLocal, int numColLocal, int *colptrLocal, int *rowindLocal, double *HnzvalLocal, int isSIdentity, double *SnzvalLocal, double temperature, double numElectronExact, double mu0, double muMin0, double muMax0, double gap, double deltaE, int numPole, int maxIter, double numElectronTolerance, int ordering, int npPerPole, int npSymbFact, MPI_Comm comm, double *DMnzvalLocal, double *EDMnzvalLocal, double *FDMnzvalLocal, double *muPEXSI, double *numElectronPEXSI, double *muMinPEXSI, double *muMaxPEXSI, int *numIter, double *muList, double *numElectronList, double *numElectronDrvList, int *info)

The estimation of deltaE should be estimated in a more automatic way later, using a few steps of Lanczos.

Page TODO List

• Add support of 64-bit integer.
• Add tree-based parallelisation for the asymmetric PSelInv.
• Add expert user interface for PEXSI for k-point sampling and for spin related calculations.