Using XCSE / splitting / Russian Roulette in egs_chamber

[erjft]

Hi, I have been trying to use a combination of XCSE, photon splitting and Russian roulette to speed up some photon simulations. However, I have found by putting

:start calculation geometry:
    geometry name  = full_simulation_geometry
    cavity regions = water_voxel_label 
    cavity mass    = 1
    enhance regions = water_voxel_label 
    enhancement     = 64
:stop calculation geometry:

and

:start variance reduction:
    radiative splitting = 100   # splitting number n_split
    photon splitting = 100 
    cs enhancement = 1 
:stop variance reduction:

My photon splitting part gets ignored in my output, but my radiative splitting, RR and XCSE are applied:

 => initScoring: splitting radiative events 100 times ...

Variance reduction
====================================================
Photon splitting = off
Range rejection = Russian Roullette (RR)
    rejection in cavity for E < 0.521
    else RR with survival probability 0.015625
    on a region-by-region basis


photon cross-section enhancement = On
 1) full_simulation_geometry
    regions    :  2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
    enhancement:  64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64 64


region by region ECUT = Off

=============================================

Is it not possible to have both at the same time? - I guess I also mean for radiative splitting too:

I see that in the egs_chamber.cpp:

// ******** photon splitting
        //
        EGS_Float tmp; int err = vr->getInput("photon splitting",tmp);
        if( !err && tmp > 1 &! do_cse &! do_TmpPhsp) {
            fsplit = tmp; fspliti = 1/tmp;
        }
        
// ******** radiative event splitting
        //
        if( !vr->getInput("radiative splitting", csplit) && csplit > 1) {
            egsInformation("\n => initScoring: splitting radiative events %d times ...\n", csplit);
           the_egsvr->nbr_split = csplit;
        }
        
  
        

[rtownson]

Indeed, it's not possible to have XCSE and photon splitting at the same time. I believe this is because the algorithm for keeping the results unbiased for the mixed VRTs would be too complex.

[erjft]

Thanks again :), I remember seeing from the paper used as a reference for XCSE, that a factor between 32-128 was worth trying with a multiple of this for RR (I used the same value as the XCSE factor). I benchmarked and found XCSE=RR=64 with splitting=25 was around 3x faster than any other setting

[rtownson]

Sure, that makes sense.