Investigate usage of sparse vectors in GObservations::likelihood::eval method

The GObservations::likelihood::eval() method still shows up as consuming a non-negligible amount of computing time when analysis SPI data due to the usage of GMatrixSparse::operator+, GMatrixSparse::column() and GVector::non_zeros(). Similar to actions done under issue #9 (closed), using sparse vectors can speed thing up (see below).

added enhancement label

assigned to @jurgen

What it boiled down to was implementing a native code for the addition of sparse matrices. This is done now, which brought a further little speed increase, reducing the ctlike fitting run (see #9 (closed)) from 188 CPU seconds to 176 CPU seconds (-6%).

closed

Following the code change, GMatrixSparse::stack_push_column() that is called by GMatrixSparse::add_to_column() is now the method that consumes about half of the computation time. So I need to check whether things can be optimised at that level.

I checked the code and there is no obvious solution for a speed-up. Hence the issue is closed.

reopened