def mh_wp_func(pkd_x):
(x, meta, beliefs, state, params) = pkd_x
if params['prior_wp_sm'] == "only_bij" and not fwd.bijection(state['sm'], x)[0]:
return float(-np.inf)
mll = meta_loglike_marg(meta, state['world'], state['sm'], x,
state['meta_noise'], params)
lpw = log_prob_world(state['world'], x)
if params['binary_beliefs']:
bll = beliefs_loglike_binary(beliefs, state['signals'], state['sm'], x,
params['own_noise_type'], state['own_noise'])
return mll + lpw + bll
else:
bll = beliefs_loglike_marg(beliefs, state['world'], state['sm'], x,
params['own_noise_type'], state['own_noise'])
return mll + lpw + bll
def mh_sm_func(pkd_x):
(x, meta, beliefs, state, params) = pkd_x
if (params['prior_wp_sm'] == "only_bij" and
not fwd.bijection(x, state['wp'])[0]):
return float(-np.inf)
#if experts:
# lps = log_prob_signals_expertise(state['signals'], state['world'],
# x, state['expertise'])
#else:
# lps = log_prob_signals(state['signals'], state['world'], x)
mll = meta_loglike_marg(meta, state['world'], x, state['wp'],
state['meta_noise'], params)
if params['binary_beliefs']:
bll = beliefs_loglike_binary(beliefs, state['signals'], x, state['wp'],
params['own_noise_type'], state['own_noise'])
else:
bll = beliefs_loglike_marg(beliefs, state['world'], x, state['wp'],
params['own_noise_type'], state['own_noise'])
return mll + bll
