search

lmoffatt / macro_dr

GNU General Public License v3.0
1 stars 1 forks source link

equilibrium likelihood of new version is lower than previous version #197

Closed lmoffatt closed 3 months ago

lmoffatt commented 6 months ago

the equilibrium likelihood of scheme1 that run last night in tupac with levenberg was lower than the one obtained before with thermo.

The difference persisted running thermo in the new version.

The difference persisted when just looking at the likelihood. It appeared by a greater variance.

lmoffatt commented 6 months ago

I found it caused by a sloppy change in the function all_Covariance_elements.

Once this was corrected the likelihood values were restored.