MH
- class sherpa.sim.mh.MH(fcn, sigma, mu, dof, *args, rng=None)[source] [edit on github]
Bases:
Sampler
The Metropolis Hastings Sampler
Changed in version 4.16.0: The rng parameter was added.
Random number generation is controlled by the
rng
argument. If set to None (the default) then the routines fromnumpy.random
are used, and so can be controlled by callingnumpy.random.seed
, otherwise it takes anumpy.random.Generator
object (or anumpy.random.RandomState
object which should only be used for testing or checking against old code).Methods Summary
accept
(current, current_stat, proposal, ...)Should the proposal be accepted (using the Cash statistic and the t distribution)?
accept_mh
(current, current_stat, proposal, ...)calc_fit_stat
(proposed_params)calc_stat
(proposed_params)dmvt
(x[, log, norm])draw
(current)Create a new set of parameter values using the t distribution.
init
([log, inv, defaultprior, priorshape, ...])mh
(current)MH jumping rule
reject
()update
(stat, mu[, init])include prior
Methods Documentation
- accept(current, current_stat, proposal, proposal_stat, **kwargs)[source] [edit on github]
Should the proposal be accepted (using the Cash statistic and the t distribution)?
- accept_mh(current, current_stat, proposal, proposal_stat)[source] [edit on github]
- calc_fit_stat(proposed_params)[source] [edit on github]
- calc_stat(proposed_params)[source] [edit on github]
- dmvt(x, log=True, norm=False)[source] [edit on github]
- draw(current)[source] [edit on github]
Create a new set of parameter values using the t distribution.
Given the best-guess (mu) and current (current) set of parameters, along with the covariance matrix (sigma), return a new set of parameters.
- init(log=False, inv=False, defaultprior=True, priorshape=False, priors=(), originalscale=True, scale=1, sigma_m=False)[source] [edit on github]
- mh(current)[source] [edit on github]
MH jumping rule
- reject()[source] [edit on github]
- tear_down()[source] [edit on github]
- update(stat, mu, init=True)[source] [edit on github]
include prior