Hi,
After last chat we talked about being careful to choose a minimum number of chains when running the MCMC method to fit the lines. The error of the MCMC method is inversely proportional to the sqrt(Number_chains) so for a given SN spectrum be careful that the number of chains is as big as the SN, preferentially higher so that the error is not dominated by the fitting methodology. A calculation with a minimum number of chains was proposed.