Angela Pond
Angela Pond
when phase binning results in empty bins and consequently, no sigmas, all two-Gaussian models fail with a "Residuals are not finite in the initial point.". Should either ignore sigmas altogether...
A Sun-Jupiter system with rv_grav enabled returns orders of magnitude higher RVs for the Sun than expected (~600 m/s expected, ~700 km/s returned) A single Sun-like star has the correct...
Meshes generated from a conda installation of phoebe differ in the number of triangles and therefore horizon triangles, which leads to different integrated flux, prominent in meshes with a small...
When using atmosphere tables instead of blackbody, the observed integrated flux never seems to converge but keeps increasing with number of triangles.
Caught this when trying to bypass the phoebe-backend mesh being computed for a system for which it typically fails by setting 'pblum_method' to 'stefan-boltzmann'.
calls to phoebeBackend.lc and phoebeBackend.rv1/2 take >10 times longer when called from within phoebe2 compared to just using phoebeBackend on its own.
If ff=0 it returns a somewhat comprehensible error, but ff=1.0 fails with either "lobe doesn't seem to exist" or segmentation fault
Need to catch those cases and not compute their potential in the nekmin algorithm.