SPARSE 2::trsv

KokkosSparse::trsv()

Header File: KokkosSparse_trsv.hpp

Usage: KokkosSparse::trsv(uplo,trans,diag,A,b,x);

Solves for the left-hand side y of the upper- or lower-triangular linear system A*x=b

Interface

template <class AMatrix, class BMV, class XMV>
void
trsv (const char uplo[],
      const char trans[],
      const char diag[],
      const AMatrix& A,
      const BMV& b,
      const XMV& x);

Parameters:

• SolverMode: "U" (for upper triangular) or "L" (for lower triangular).
• InputMode: "N" for no transpose, "T" for transpose, or "C" for conjugate transpose.
• DiagonalMode: "U" (for implicit unit diagonal) or "N" (for not).
• InputMatrix: a KokkosSparse::CrsMatrix
• InputVector: a rank-1 or rank-2 Kokkos::View with non-const data type.
• Input/OutputVector: a rank-1 or rank-2 Kokkos::View with non-const data type.

Requirements:

• The input matrix A must be upper triangular or lower triangular.
• OutputVector::value_type == OutputVector::non_const_value_type
• A.rank == 2
• y.rank == b.rank
• y.rank == 1 or y.rank == 2
• y.extent(0) == b.extent(0)
• y.extent(1) == b.extent(1)

Example

#include<Kokkos_Core.hpp>
#include<KokkosSparse_trsv.hpp>

int main(int argc, char* argv[]) {
   Kokkos::initialize();

   int N = atoi(argv[1]);

   Kokkos::View<double**> A("A",N,N);
   Kokkos::View<double*> b("X",N);
   Kokkos::View<double*> y("Y",N);
   
   // Initialize A as a lower-triangular matrix
   // Initialize b as a vector solution in the range-space of A
   // ...

   KokkosSparse::trsv("L", "N", "N", A, b, x);
   KokkosSparse::trsv("N", alpha, input_mat, x, beta, y);

   Kokkos::finalize();
}