Hi,

I am running toys with high statistics and input parameters that are quite different from the default values in my Parameters.cpp file. (I use several different sets of inputs, so I cannot simply just update the default values to solve the problem below).

Due to the high stats, and resulting steep chi^2 function, the fits need to start quite close to the solution to converge properly.

For the ProbScans I can simply use --parfile and everything works fine.

However, the initial MethodAbsScan::doInitialFit() does not pick up the parameters in --parfile and when it does not converge, I get a later error message

MethodAbsScan::confirmSolutions() : WARNING : solution 0 rejected (too large chi2: DeltaChi2>25 - chi2minGlobal: -1.000000e+30

because chi2minGlobal has a non-sensical value.

Would there be a problem in having MethodAbsScan::doInitialFit() pick up the input start value? Or is there a different way to resolve this? Happy to implement it!