I couldn’t find any similar question on this topic. Is there any established measure of “convergence” when varying the voxel size in the USRBIN detector analogous to mesh-convergence checks in FEM?
How can we determine or justify that the chosen voxel size is adequate for accurately representing the deposited energy in the given scoring mesh, particularly in dosimetry applications during X-ray or gamma exposure?
compared to FEM the MC simulation evaluates values not per object, but per bin (pixel, voxel, energy range), depending on your MC scoring instrument. That is why FLUKA has separate scoring/detector tools, shich allows you to adjust parameters of scoring, and so achieve accuracy reasonable for your task.
In case of MC and Fluka particularly, we have a standard deviation of each scored value, which depends on the parameters of coring and accumulated statistics. In. general it is recomemned to have SDT bettter that 10%, 50% can be considered as a rubbish output with respect to the confidence interval values, see p.17 https://agenda.infn.it/event/20624/contributions/105895/attachments/68619/85014/AdvancedScoring2019.pdf
For USRBIN card output you can access error values by converting bnn binary file to ascii (if the file is not very huge), or by checking “error“ checkbox in the plotting tool tab.
To clarify what I meant: suppose we compare two USRBIN meshes (fine and coarse). In both cases we can achieve a statistical uncertainty below 10%, but the fine mesh requires substantially more primary histories to reach that level. In that situation, it seems reasonable to use the coarser mesh—provided that the scored quantity of interest (e.g., absorbed dose or deposited energy density) is consistent between the two mesh settings.
yes, exactly. It is usuallz usful to have overlapped mesh of different sizes and steps, to control accuracy of calculation for the reasonable number of primaries.
And for zour convinience, zou can use detectro output to monitor exact accuracy (std error) value.