@hschindl
Hi hschindl, in Garfield examples, the diffusion coefficients can be obtained from the scripts “~/garfiledpp/Examples/Gasfile/generate.C”, however, I found small difference of the diffusion coefficients （~2%）each run. I‘m not sure whether it depends on the statistic of electrons or number of collisions?
Because the deviation of ~2% is higher than the impact of other factors (e.g. magnet) we preferred to compared, is there any ways to suppress it (input some other codes)?
Thank you very much for help.

Hi,
to reduce the statistical error of the diffusion coefficient (and other transport parameters) calculated by Magboltz, you need to increase the number of collisions.

Thanks for your information,
Which function should be invoked to modify the number of collisions? Is any example referred?

Modify the value of “ncoll” in the following codes?
“ const int ncoll = 10;
// Run Magboltz to generate the gas table.
gas.GenerateGasTable(ncoll);
”
But the description means generation of the gas table.

I check the source files “MediumMagboltz.cc”, the function “GenerateGasTable()” include number of collisions when invoke the function “RunMagboltz()”, so the number of Collisions is multiplication of “ncoll” and 10^(7).
I’m curious about what is the upper limit of the set value of “ncoll”？