Speaker
Description
For X-ray spectra with low counts, traditional statistics such as 𝜒2 statistic and Poisson (C) statistic cannot well enough evaluate the goodness-of-fit during model fitting, because the binning process is usually arbitrary and would result in the loss of channel information. The maximum-likelihood estimation method is an ideal alternative method that does not require any binning procedures. We establish the maximum-likelihood estimation method and test it with fake spectra with low counts. We also test other methods such as XSPEC fitting, MCMC method and BXA method for comparison, and we found that the results obtained by the maximum-likelihood estimation method are better than those obtained by XSPEC fitting. Both the maximum-likelihood estimation method and MCMC method can obtain pretty good results, however, the running time of MCMC can be much longer than that of the maximum-likelihood estimator. Besides, we also look into the effect that the parameter values have on the results. We further apply the method to the LEIA observational data and attain some satisfactory results. In a word, we have established the unbinned maximum-likelihood estimation method especially for low-count data and proved its practicability by applying to the real data.
