snow_white GaiaID str fix #42
Conversation
| @@ -201,12 +202,12 @@ def snow_white( | |||
| if first_T <=13000.: | |||
| tmp_Tg,tmp_chi= grid_param[grid_param[:,0]>13000.], grid_chi[grid_param[:,0]>13000.] | |||
| second_T= tmp_Tg[tmp_chi==np.min(tmp_chi)][0][0] | |||
| fit_params['teff'] = lmfit.Parameter(name="teff",value=second_T,min=13000,max=80000) | |||
| fit_params['teff'] = lmfit.Parameter(name="teff",value=second_T,min=12000,max=80000) | |||
There was a problem hiding this comment.
I understand fixing the string comparison bug we discussed by email, but I’m confused by the logic of splitting up to 12,000 K in one optimisation and then 14,000 K in another. What am I missing here?
There was a problem hiding this comment.
Hi Andy,
Unless I messed up I did not change where the split happens. I still look for a secondary solution starting below 13000 if the first one was hotter and above 13000 if the first solution was colder.
What I did was make the maximum value in the fit less a of strict wall. So yes I am looking for a solution above 13000 so I start the fit from the best initial value above 13000, but the fit may still only converge below 13000 so I am now giving some margin to go below 13000 during the fit so we don't just get a bunch of exactly Teff=13000K for fits that would converge just below it.
Similarly I expanded the maximum limit to 14000 if I am looking for a second solution starting below 13000.
No description provided.