def __init__(self, likelihood_module, prior_type, prior_means, prior_sigmas, width_scale, sigma_scale): """ :param likelihood_module: likelihood_module like in likelihood.py (should be callable) :param prior_type: 'uniform' of 'gaussian', for converting the unit hypercube to param cube :param prior_means: if prior_type is 'gaussian', mean for each param :param prior_sigmas: if prior_type is 'gaussian', std dev for each param :param width_scale: scale the widths of the parameters space by this factor :param sigma_scale: if prior_type is 'gaussian', scale the gaussian sigma by this factor """ self._ll = likelihood_module self.n_dims, self.param_names = self._ll.param.num_param() lowers, uppers = self._ll.param_limits if width_scale < 1: self.lowers, self.uppers = utils.scale_limits( lowers, uppers, width_scale) else: self.lowers, self.uppers = lowers, uppers if prior_type == 'gaussian': if prior_means is None or prior_sigmas is None: raise ValueError( "For gaussian prior type, means and sigmas are required") self.means, self.sigmas = prior_means, prior_sigmas * sigma_scale self.lowers, self.uppers = lowers, uppers elif prior_type != 'uniform': raise ValueError( "Sampling type {} not supported".format(prior_type)) self.prior_type = prior_type self._has_warned = False
def test_scale_limits(): lowers_list, uppers_list = [0, -1, 5], [10, 9, 15] lowers, uppers = np.array(lowers_list), np.array(uppers_list) widths = uppers - lowers scale_factor = 0.5 lowers_s, uppers_s = sampling_util.scale_limits(lowers_list, uppers_list, scale_factor) npt.assert_equal(lowers_s, np.array([2.5, 1.5, 7.5])) npt.assert_equal(uppers_s, np.array([7.5, 6.5, 12.5])) npt.assert_equal(widths*scale_factor, (uppers_s - lowers_s))