There are two ways of evaluating prony coefficients. One can get individual prony constants for each temperature or one can use the shift coefficients along with one set of prony coefficients. Generally the procedure to evaluate prony coefficients is as follows. 1) Pick a temperature and evaluate prony coefficients at that temperature using the tref option. 2) Keep the prony coefficients constant(tbft,fix) and evaluate the shift coefficients. 3) And then fix the prony coefficients and solve for all the parameters if the fit from 2 is not good enough. Each data point gets equal weightage. We just minimize the error at all points when we evaluate shift and prony parameters. There is nothing special done to differentiate between prony parameters or shift parameters. All of them are used to fit the model unless the user decides to fix/hold a parameter constant. One can also solve for all prony and shift parameters simultaneously in one step. But it is harder to fit that way and one may not always get a result easily.