Skip to content

Optimize memory usage in `dofisher`

Giacomo Galloni requested to merge optimize_memory into optimize_trace

This PR improves memory usage of dofisher by avoiding allocating the Sig matrix during Fisher computation.

This build, which comprehends the improved trace computation (!7 (merged)), has been tested assuming nside=16, a mask with fsky~30% and using dofisher_QU.

In particular, the output Fisher matrix from the master branch and this one are identical.

Edited by Giacomo Galloni

Merge request reports