python-mle icon indicating copy to clipboard operation
python-mle copied to clipboard

Published 20 hours ago • verified publisher icon ibab

Implement random number generation

Open ibab opened this issue 10 years ago • 2 comments

We should be able to easily sample data from the model.

ibab avatar Jan 30 '15 12:01 ibab

I've added a very crude implementation of this. The .sample method on every distribution calls numpy to sample from the right distribution. This could also be implemented with Theano. It would be nice to have parallel generation of samples. Also, there might be smarter ways to get the "true" values needed for sampling than just passing them in.

ibab avatar Feb 01 '15 09:02 ibab

With the increased flexibility of the new API, sampling has become a lot harder. There's probably no way around implementing MCMC if we want to sample arbitrary models.

ibab avatar Feb 08 '15 22:02 ibab