I am curious about something analysing .gas file data.
I have understood that ncoll in gas.GenerateGasTable(ncoll) defines the number of collision multiplied by 1e+7 while an electron moves through the gas. However, in my opinion, since Magboltz is based on the Monte-Carlo simulation, numerous times of simulations or independent events are required to form error bar or analyse it. Is my opinion wrong or not useful?
I think I do insufficiently understood the principle of Magboltz or gas file.
Hi,
thanks for your question! With a suitable choice of the initial conditions (in particular the initial kinetic energy), the time average approach (following a single electron over many collisions) and the ensemble average approach should be equivalent, i. e. give statistically compatible results for the transport parameters (drift velocity, diffusion coefficients, etc.).
And I have a question, how does vzerr in Magboltz have been calculated without general process evaluating standard deviations for each independent event? Since I have never seen this kind of Monte-Carlo, I am really curious about it!
I am not sure I understand your question. Note that “event” for Magboltz means an electron-atom scatter/collision, not the rigorous tracking of a single electron through the gas. vzerr is calculated based on the standard deviation of vz over many electron-atom collisions. If you like to know more, I would invite you to have a look at the Magboltz articles [1,2] or the very instructive PyBoltz article [3]. The last one is authored by scientists that have rewritten the orginal Fortran Code in Python together with Steve Biagi, and is very instructive to understand how the code works [3]:
