Dear experts,
I was simulating a room. When I simulated for the first time, I set the number of particles to one million, and got the results of USRBIN as shown in the following figure. My statistical results are very problematic. The results and errors of some positions are all 0, and the statistical errors of most results are extremely large.
In my second simulation, I changed the number of particles to 10 million. The simulation results are as follows, and the statistical error is still very large.
Is it right for me to expand the number of particles?
Should I continue to expand the number of particles to reduce the statistical error?
Or does my grid accuracy need to be improved?
it is not necessary to get results or minimize the error in all bins of an USRBIN scorning, only in those locations where you are interested in the results.
yes, the easiest way to get better statistics is to simulate more primaries. This can be reached by increasing the number of primaries of a simulation, or running parallel, often on a HPC cluster.
The second is to increase the size of the bins (if it make sense).
However, sometimes this is not enough, and you need to “direct” where the simulation time is spent by using biasing. For more information see this lecture:
Thank you very much for your reply.
I will try to increase the number of my particles to see if it can solve the problem.
As for the second method, do you mean to increase the size of “bin” or increase the number of “bin”?
In my opinion, isn’t the more the number of “bin”, the finer the calculation result?
Maybe my understanding is wrong?
Dear @Junjie_Zhang,
What @horvathd means is to increase the size of the bin. It is true, as you say, that a larger number of bins corresponds to a finer calculation, nevertheless, if you don’t achieve the level of statistical significance you desire in each bin, then you can use larger bin. In this way you will be able to better estimate in each bin the quantity of interest, because you will have more contributions to it, i.e. more particles will go across that bin.
