Probabilities
This is a list of the implemented probabilities
- class triqs_maxent.probabilities.NormalLogProbability(log_measure=None, log_norm_S=None, log_prior_alpha=None)[source]
Bases:
GenericFunctioncalculate the \(\log\) of the probability of \(\alpha\)
- Parameters:
- log_norm_Sfunction
normalization of the entropy as a function of (\(\alpha\), \(N_{\omega}\)),
- log_measurefunction
measure for the integration as a function of \(A\),
The default is
\[\frac{1}{2} \det(-\frac{\partial^2 S}{\partial A_i \partial A_j}).\]Skilling, Classic Maximum Entropy, Maximum Entropy and Bayesian Methods, Kluwer 1989 gives this as a general expression for the measure.
- log_prior_alphafunction
prior of alpha as a function of alpha, default is \(-\log(\alpha)\), i.e. Jeffrey’s prior (a constant prior according to S.F. Gull, Developments in Maximum Entropy Data Analysis, in J. Skilling (ed.) Maximum Entropy and Baysian Methods, p. 57, Kluwer 1989 is possible by setting it to
lambda alpha : 0)
Methods
__call__(cost_function)Call self as a function.
parameter_change()Notify the function that parameters have changed
f