In evaluate_model(), when it gets a user-supplied time vector, supersampling parameters were never passed on to batman/catwoman, resulting in wrong lightcurves being generated. I saw this when plotting the lightcurve of a K2 planet with particularly long ingress/egress, which made it very visible .
In this pull request, I have taken the liberty of removing keyword arguments resampling, nresampling, and etresampling from evaluate_model(). Instead I make sure that it always gets the supersampling parameters that were used in the fit via self.dictionary, as it was already done some places.
In
evaluate_model()
, when it gets a user-supplied time vector, supersampling parameters were never passed on to batman/catwoman, resulting in wrong lightcurves being generated. I saw this when plotting the lightcurve of a K2 planet with particularly long ingress/egress, which made it very visible .In this pull request, I have taken the liberty of removing keyword arguments
resampling
,nresampling
, andetresampling
fromevaluate_model()
. Instead I make sure that it always gets the supersampling parameters that were used in the fit viaself.dictionary
, as it was already done some places.