search

RickKessler / SNANA

Supernova Analysis package
38 stars 23 forks source link

wfit occasionally fails catastrophically #1069

Closed helenqu closed 1 year ago

helenqu commented 1 year ago

path to example of failed wfit: /scratch/midway2/rkessler/PIPPIN_OUTPUT/HELEN_MISMATCH_TO_WFIT_WITH_CC_LOWZ/6_BIASCOR/BCOR_WRONGHOST_SERSIC_SCALE/output/PIP_HELEN_MISMATCH_TO_WFIT_WITH_CC_LOWZ_DESSIM_WRONGHOST_SERSIC-0025

the other realizations of the same simulation were fit successfully, i.e. /scratch/midway2/rkessler/PIPPIN_OUTPUT/HELEN_MISMATCH_TO_WFIT_WITH_CC_LOWZ/6_BIASCOR/BCOR_WRONGHOST_SERSIC_SCALE/output/PIP_HELEN_MISMATCH_TO_WFIT_WITH_CC_LOWZ_DESSIM_WRONGHOST_SERSIC-0024

this also seems to be a problem only for these simulations generated with HOSTLIB_SCALE_SERSIC_SIZE: 0.8. i don't see any examples of this happening for the set of otherwise identical simulations that don't scale the Sersic parameters.

RickKessler commented 1 year ago

crazy fit-errors were identified in this case, and the fit was repeated. But a bug crept in that forced all the fit-par ranges to be zero. Fixed.