Out of range variables in MRF

[moritz]

I have been using the multiple run fitter recently. I am setting positive limits on many of my variables. I noticed that the fitter stops running after a number of trials and there are errors in the command window indicating that the value of various variables are out of range. I tried upgrading to NEURON 5.9, but still get the same behavior.

Why does the fitter show an "error" when the random variable is out of range rather than simply continue generating variables until one is found within the given range? Is there a way to circumvent this problem when setting positivle limits on optimized variables?

Thanks in advance,
Mickey

[hines]

Is there a way to circumvent this problem when setting positivle limits on optimized variables?

Edit the nrn/lib/hoc/mulfit/eparmlst.hoc file and search for

Code: Select all

func efun()

and comment out the line

Code: Select all

printf("invalid %s low=%g val=%g high=%g\n",...