energy_bias_resolution_from_energy_dispersion still assumes wrong normalization of EDISP
Closed this issue · 1 comments
Tobychev commented
Plotting the bias+1 from energy_bias_resolution_from_energy_dispersion on top of the energy bias matrix gives the following result:
where the blue line and the red line are both estimates of the median in each column of the matrix. The red curve produced using the numpy percentile function using the 'median_unbiased' method (the recommended method in the paper numpy references but for some reason not the default value).