Utilities¶

Several useful functions for use in various places in PyNFG

Part of: PyNFG - a Python package for modeling and solving Network Form Games

Created on Tue May 7 15:39:38 2013

GNU Affero General Public License

pynfg.utilities.utilities.convert_2_pureCPT(anarray)[source]¶

Convert an arbitrary matrix to a pure CPT w/ weight on maximum elements

Parameters:	anarray (np.array) – The numpy array to be converted to
Returns:	a normalized conditional probability distribution over actions given messages with all elements zero or one.

pynfg.utilities.utilities.mceu(Game, dn, N, tol=30, delta=1, verbose=False)[source]¶

Compute the move-conditioned expected utilities for all parent values

Parameters:	Game (SemiNFG or iterSemiNFG) – the SemiNFG of interest dn (str) – the name of the decision node where MCEUs are estimated N (int) – the max number of iterations for the estimation tol (int) – the minimum number of samples per parent value

pynfg.utilities.utilities.mh_decision(pnew, pold, qnew=1, qold=1)[source]¶

Decide to accept the new draw or keep the old one

Parameters:	pnew (float) – the unnormalized likelihood of the new draw pold – the unnormalized likelihood of the old draw qnew (float) – the probability of transitioning from the old draw to the new draw. qold (float) – the probability of transitioning from the new draw to the old draw.
Returns:	either True or False to determine whether the new draw is accepted.