search

sidmohite / kNe-inference

A Bayesian inference framework to constrain kilonovae models
MIT License
0 stars 0 forks source link

Generic (non-analytic) model support #3

Open dlakaplan opened 4 years ago

sidmohite commented 4 years ago

On thinking about this further as well, wouldn't any model for the absolute mag evolution depend on some parameter after all. I mean I get the aspect of a numerical model where there won't be any direct dependence on a parameter but it could indirectly depend on things like ejecta mass, opening angle etc. My point being that eventually there will always be a map F (param1,param2,.... ) = M(t). I guess the inference then in that case could be over these parameters then. Currently what I have is agnostic about the functional form: M_g = np.array([self.lc_model_g(*params, t_0=t0, t=t_j)])

dlakaplan commented 4 years ago

Yes, definitely. I just didn't know if your code could handle models say without analytic models/derivatives. Or perhaps discrete variables. But as long as you built it generically that's good.