Adaptive Covariance MC¶
- class pints.AdaptiveCovarianceMC(x0, sigma0=None)[source]¶
Base class for single chain MCMC methods that globally adapt a proposal covariance matrix when running, in order to control the acceptance rate.
Each subclass should provide a method
_generate_proposal()
that will be called byask()
.Adaptation is implemented with three methods, which are called in sequence, at the end of every
tell()
:_adapt_mu()
,_adapt_sigma()
, and_adapt_internal()
. A basic implementation is provided for each, which extending methods can choose to override.Extends
SingleChainMCMC
.- eta()[source]¶
Returns
eta
which controls the rate of adaptation decayadaptations**(-eta)
, whereeta > 0
to ensure asymptotic ergodicity.
- name()¶
Returns this method’s full name.
- needs_sensitivities()¶
Returns
True
if this methods needs sensitivities to be passed in totell
along with the evaluated logpdf.
- set_eta(eta)[source]¶
Updates
eta
which controls the rate of adaptation decayadaptations**(-eta)
, whereeta > 0
to ensure asymptotic ergodicity.