CSV Converter Improvements - #410
Conversation
Originally only an FSRoot style 4-vector indexing pattern was supported, where it was assumed particle components were stored like "Px1", "Py1", etc. This change allows for the FinalState style indexing pattern, where the particle components are stored in a single branch like "Px_FinalState[0]", etc. This commit also includes a bug fix where a background tree name would be searched when no background files existed.
Filtered amplitudes were not being properly used to build the coherent sum groups. Also, one of the Lme coherent sum groups was in reverse order.
|
Test status for this pull request: SUCCESS Summary: summary.txt Build log: make_csv_converter.log |
|
There is some discussion and implementation of weights on the generated MC in this code. We need to verify that the handling of this is consistent with what is done in AmpTools. I believe in AmpTools, all integrals are renormalized to the integer number of actual generated events. This means that weighting can only be used to redistribute events but not change the number of generated events. I don't think I understand how this code is being used enough to know how the generated MC weight enters the algorithm and whether the use is consistent with what is done in the NormIntInterface::forceCacheUpdate and AmpliutdeManager::calcIntegrals methods. (The changes to precision of output look fine, but there are other changes in addition to that!) |
3 participants
This pull request solves a handful of issues.
RootDataConverter Improvements
Px_FinalStatestyle 4-vector branches. Previously only thePx1FSRoot style branches were assumed. This should now cover most use cases.Other