search

drbenvincent / darc_toolbox

Run adaptive decision making experiments
MIT License
16 stars 2 forks source link

ensure fixed parameters are dealt with properly #50

Closed drbenvincent closed 5 years ago

drbenvincent commented 5 years ago

From working through the parameter sweep plots (#6) it looks as if we are not properly dealing with fixed parameter values. At the moment, if we assert the true parameter of alpha to be some value, this is set in model.θ_true, but it is not acted on. Ie, we don't:

In fact, delegation between fixed vs free parameters is simply hard wired at the moment. This problem has not even been solved. For example, in CumulativeNormalChoiceFunc we just use θ_fixed['ϵ'] so as it stands, there is no easy way to decide to do inference over ϵ

drbenvincent commented 5 years ago

Actually, no. I'm going to stick with a model which has various free and fixed parameters and it stays that way. I you want to change a parameter from free to fixed, or visa versa then just copy/paste/update the model.