I am running medical linear accelerator simulation and using phase space as a particle generator. I have been studying some optimization techniques which not only can reduce run time but also provide better particle statistics about interaction with matter per voxels.
One of the techniques that I used is geometrical biasing. I collimated the particles in certain positions. This provides me a reduction of time but gives me worse statistics because of that less particle in the voxels. I am going to sum more phase spaces in order to get enough particles for the collimated region. But I wonder how many particles are required per voxel in order to get better statistics on distribution?
on which version of medical_linac / geant4 you work ?
First of all, thanks for the response. I am using version of 10.5. But i am simulating my own medical linac model. What I know that I need to store approximately 10000 particle into a voxel in order to obtain dose accuracy of %1 by using the equation which is 1/sqrt(Particle_In_The_Voxel)
So it’s hard to get that kind of a number of particles within the voxel because of the constraints of total particle numbers. So if this is the approximately particle that needed then I must run the simulation lots of times with different phase-spaceses of Linac. I must say, I run my simulation 45 times using 2xE9 particles for each run. Still I have got approximately 800 particles in each voxel. (Voxel sizes 5x5x5 mm3 and colimated phase-space area was 10x10 cm2.) Due to Phase-space has certain number of particles and also their physical specs, Number of phase-spaces were summed. (10 of them)
sorry the delay of my answer,
so if I understand correctly, you have the problem of having a sufficient number of particles in each volume voxel simulates.
I also work on medical linac, I simulated cases for a voxel size 5 x 5 x1 mm3 and I wait about 10 days to get good results. (but I do not use phase space as a particle source).
how of time can you have 800 particles for each voxel ?
It took approximately 3 days in order to collect 800 particles per voxel of 5x5x5 mm3.
If you dont mind i want to ask how many particles did you score to have that result of yours per voxel of 5x5x1 mm3? And how many times did you rerun? As you know because of the obstacles of 32bit we are able to use (2e31)-1 particles for generator at most. So if it is okey for you please let me know the details of your generator and scoring parameters.
My best wishes
I use the score of 1E9 particles, but I have the account in grid computing.