Maximum likelihood estimation of a Bayesian model

junpenglao · April 10, 2018, 3:01pm

Yes you can get the logpt from model, which is a theano tensor. To evaluate the tensor you can compile it as a function, eg see:

github.com

pymc-devs/pymc3/blob/master/pymc3/step_methods/metropolis.py#L136




    # Determine type of variables
    self.discrete = np.concatenate(
        [[v.dtype in pm.discrete_types] * (v.dsize or 1) for v in vars])
    self.any_discrete = self.discrete.any()
    self.all_discrete = self.discrete.all()


    self.mode = mode


    shared = pm.make_shared_replacements(vars, model)
    self.delta_logp = delta_logp(model.logpt, vars, shared)
    super(Metropolis, self).__init__(vars, shared)


def astep(self, q0):
    if not self.steps_until_tune and self.tune:
        # Tune scaling parameter
        self.scaling = tune(
            self.scaling, self.accepted / float(self.tune_interval))
        # Reset counter
        self.steps_until_tune = self.tune_interval
        self.accepted = 0

Topic		Replies	Views
Interpreting Bayesian Inference with multiple parameters for the MLE Questions	8	1952	December 5, 2018
Calculating marginal likelihood of GP Model Questions	6	1319	December 31, 2018
Unconsistent logp, Evaluating the results of find_MAP on Custom Likelihood Questions	3	610	August 28, 2018
Adjusting models with mcmc Questions	5	1471	June 26, 2020
Sample external model parameters with PyMC3 Questions	4	1001	October 17, 2017

Maximum likelihood estimation of a Bayesian model

Related topics