Currently, the pulsar likelihoods are normalized as $P(\dot{P})$ rather than $P(\dot{P}/P)$. In effect, this means that each contribution from each pulsar to the total likelihood was weighted by the inverse of its period. This is an especially large effect for the orbital periods whose importance was reduced by several orders of magnitude compared to the spin periods.
While I made the above fixes, I also switched our Milky Way potential from BovyMWPotential2014 to MilkyWayPotential2022. This potential is fit to more recent data, but most importantly does not depend on GSL which was the cause of many installation issues. The differences between these potentials in terms of period derivatives are several orders of magnitude smaller than the effect of the cluster potential. Closes https://github.com/nmdickson/GCfit/issues/46.
Currently, the pulsar likelihoods are normalized as $P(\dot{P})$ rather than $P(\dot{P}/P)$. In effect, this means that each contribution from each pulsar to the total likelihood was weighted by the inverse of its period. This is an especially large effect for the orbital periods whose importance was reduced by several orders of magnitude compared to the spin periods.
While I made the above fixes, I also switched our Milky Way potential from
BovyMWPotential2014toMilkyWayPotential2022. This potential is fit to more recent data, but most importantly does not depend onGSLwhich was the cause of many installation issues. The differences between these potentials in terms of period derivatives are several orders of magnitude smaller than the effect of the cluster potential. Closes https://github.com/nmdickson/GCfit/issues/46.