Thank you Barry and Jed,
I think I'll go with MatDiagonalScale.
Michael.
On 06/20/2018 02:06 PM, Jed Brown wrote:
Michael Povolotskyi <[email protected]> writes:
Dear Petsc team,
what would be the most efficient way to compute the following:
V += A*B*C,
where V is a dense square matrix,
B is a square diagonal matrix, A and C are dense rectangular matrices?
I would like to benefit from the fact that B is diagonal.
MatDiagonalScale. No matter what you do, the cost will be dominated by the
dense product.
