Cyclops Tensor Framework
parallel arithmetic on multidimensional arrays
block_sparse

All-pairs shortest-paths via path doubling and Tiskin's augmented path doubling. More...

Collaboration diagram for block_sparse:

Namespaces

 CTF
 

Functions

Matrix flatten_block_sparse_matrix (Matrix< Tensor<> > A, std::vector< int > ranges, World &dw)
 
int block_sparse (std::vector< int > ranges, World &dw)
 perform block sparse matrix-matrix product More...
 
char * getCmdOption (char **begin, char **end, const std::string &option)
 
int main (int argc, char **argv)
 

Detailed Description

All-pairs shortest-paths via path doubling and Tiskin's augmented path doubling.

Function Documentation

int block_sparse ( std::vector< int >  ranges,
World dw 
)

perform block sparse matrix-matrix product

Parameters
[in]rangesblock sizes along each dimension
[in]dwset of processors on which to exectue
Returns
true if test passes

Definition at line 64 of file block_sparse.cxx.

References CTF::Pair< dtype >::d, CTF::Tensor< dtype >::fill_random(), flatten_block_sparse_matrix(), CTF_int::tensor::is_sparse, CTF::Pair< dtype >::k, CTF_int::tensor::lens, CTF::Tensor< dtype >::norm2(), NS, CTF_int::tensor::order, CTF::World::rank, SP, CTF_int::tensor::sym, and CTF::Tensor< dtype >::write().

Referenced by main().

Matrix flatten_block_sparse_matrix ( Matrix< Tensor<> >  A,
std::vector< int >  ranges,
World dw 
)
char* getCmdOption ( char **  begin,
char **  end,
const std::string &  option 
)

Definition at line 168 of file block_sparse.cxx.

Referenced by main().

int main ( int  argc,
char **  argv 
)

Definition at line 179 of file block_sparse.cxx.

References block_sparse(), getCmdOption(), ctf.core::np(), and ctf.core::rank().