8a6ecfdb1759ad2f67492a983ec5f90b5c5dd3ae,brainiak/reprsimil/brsa.py,GBRSA,_set_SNR_grids,#GBRSA#,4087

Before Change

            SNR_weights[0] = SNR_weights[0] / 2.0
            SNR_weights[-1] = SNR_weights[-1] / 2.0
        elif self.SNR_prior == "lognorm":
            log_SNR_grids = ((np.arange(self.SNR_bins)
                              - (self.SNR_bins - 1) / 2)) \
                / self.SNR_bins * self.logS_range * 6
            SNR_grids = np.exp(log_SNR_grids)
            log_SNR_grids_upper = log_SNR_grids + self.logS_range * 3 \
                / self.SNR_bins
            SNR_weights = np.empty(self.SNR_bins)
            SNR_weights[1:-1] = np.diff(
                scipy.stats.norm.cdf(log_SNR_grids_upper[:-1],
                                     scale=self.logS_range))
            SNR_weights[0] = scipy.stats.norm.cdf(log_SNR_grids_upper[0],
                                                  scale=self.logS_range)
            SNR_weights[-1] = 1 - scipy.stats.norm.cdf(log_SNR_grids_upper[-2],
                                                       scale=self.logS_range)
            SNR_grids[0] = 0
        else:  // SNR_prior == "exp"
            SNR_grids = self._bin_exp(self.SNR_bins)
            SNR_weights = np.ones(self.SNR_bins) / self.SNR_bins

After Change

                    lb=bounds[i], ub=bounds[i + 1]) * self.SNR_bins
            // Center of mass of each segment between consecutive
            // bounds are set as the grids for SNR.
            SNR_weights = np.ones(self.SNR_bins) / self.SNR_bins
        else:  // SNR_prior == "exp"
            SNR_grids = self._bin_exp(self.SNR_bins)
            SNR_weights = np.ones(self.SNR_bins) / self.SNR_bins
        SNR_weights = SNR_weights / np.sum(SNR_weights)

Instances