...
- Cameron – ldmx-sim
- Some discussion on std-hep as an input format. Omar agrees that this is reasonable "It benefits ldmx-sim to have many input formats". With std-hep file reading capability, compatibility with existing files (and tools) will be OK. Many tools can be put in ldmx-sim though.
- Cameron - hpstr update
- See slides.
- Robert Johnson - Kalman Filter
- Robert found an issue with coordinate transformations. Now the results are the same as the hps-java.
- He will fill an interface to GBL. This is useful for Millipede, and also the GBL code more easily provides the full co-variant matrix.
- Tongtong - stdhep tools
- He found a memory leak in the stdhep tools. This was causing issues running on large files.
- Issue with fie open not returning the number of events properly.
- Use of random_sample.cc -
- Need to produce more events than N for random sampling.
- random_sample.cc reads in all the input into a vector. This takes a lot of memory.
- random_sample.cc picks the input events at random. This means that an estimate of 37% input events are not used, and 37% are used more than once.
- Solution1: random_sample_usingInputEventsInOrder.cc – Just takes the events in order, avoiding the memory issue and the usage issue.
- Solution2: Update of random_sample.cc – Use the std::shuffle to randomize the order of the events. No waste of electrons, and less memory use.
- Tontong - Trigger Diagnostics 2019
- Confluence page detailing development: "Trigger Diagnostics 2019 "
- Replaces a number of the drivers that were used for the 2015/2016 data sets, instead of modifying the existing files, creates new "2019" versions. Kyle suggests to use inheritance to make this a clearer more maintainable update.
{"serverDuration": 50, "requestCorrelationId": "2eaab37c747de6bc"}