csmmlab / kit-rt Goto Github PK

Title says it all. Determine which test cases are good for work and create documentation files

Hello, the simulation ran without issues with VOLUME_OUTPUT_FREQUENCY = 0, but it terminated with an error (Return code: -9) when VOLUME_OUTPUT_FREQUENCY was set as 1.
I tested it using WSL2 Ubuntu on a machine with Windows 11 Home (version: 23H2).
The attachment contains a log of the solver, the config file, and the hardware information of the machine.
Thank you!

doc.zip

Ideas to find the error:

• Compare Initial conditions
• Find why our dose is by 10³ higher. Moment normalization factor? Factor in dose computation/integration?
• Compare Stopping powers
• Find out details about StarMAP cross-section plot + its legedend.
• Hardcode StarMap IC s.t. they are comparable to ours
• Check PN Scattering influence

Title says it all

• small meshes work, even with valgrind
• big mesh crashes in runtime
• Could be a section, that compiles once correctly and once false

• Config files from KiTRTPaperData seem to work for 13 moments
• Config files with smaller mesh, density one but same number of moments also work
• when increasing moments to 20, solver breaks -> mass explodes, then becomes -nan, flux is inf then -nan
• Edit: the breaking point is 20 moments, below everything is fine, above problems ensue -> maybe the StarMap/physical data only goes up to 19?

• Compare 1st and 2nd order for standard PN,SN

• Implement 2nd order for CSDPN/SN

• See if the cross artifact remains

• Consider 2nd order DG since it is AP

• Consider adjustable parameters in unstructured triangular mesh need to be considered!

Presprint TODO:

• validate PN in different cases against starmap

Sprint TODO

• Adding MPI
• Collided-Uncollided Split
• Running CT Scans with Triangular meshes (with Code of Chinmay)
• Validate & Check second order (if working, add to all methods / new branches)
• Prepare steps to 3D code (to fuse with Pias Matlab code) [IC, BC, Fluxes] (not high prio)
• Adding a beam test case
• Change Cell based flux to Face based Flux / Streamline mesh class for this case