[WIP] Include validation for LST-1 Crab analysis #151
Conversation
Maybe the ones from the validation paper? But that uses a lot more data, so I am not sure. You can maybe also add this to the benchmarks. I remember we had performance issues with the LST that were then resolved, so would be good to ensure that we dont reintroduce them |
|
Thanks, @AtreyeeS. I'll use the ones from the LST-1 performance paper as reference values and include the analysis in the benchmarks. Internal note: the comparison cannot be directly done between both results. IRFs used for the LST-1 performance paper (for each DL3 file we used an IRF produced from MC performed in the nearest pointing node simulated) are different from the ones included in this DL3 Crab released (IRFs are interpolated between singular IRFs produced in different pointing positions). |
There was a problem hiding this comment.
Thanks @morcuended . This looks nice.
This should be run in two steps? First to download and then to run the analysis? You could add an option on run-analysis to perform data download in all cases.
|
What is the status here @morcuended ? Can this become a regular PR? |
|
I started looking into the plotting part of the generated spectrum vs reference one. Doing a bit of cleanup as well. The thing is that with new Gammapy versions I have to change some stuff in the way I read the reference results from https://github.com/cta-observatory/lst-crab-performance-paper-2023 I also simplified the Zenodo download part but I did not push the changes yet. |
Added
make.pyscript to perform the 1D analysis of LST-1 Crab data from Zenodo. Since we use the energy-dependent angular cuts, we cannot use the HLI analysis interface. So here I follow the low-level API.To be done:
Question: Which should be the reference model and flux points to compare with?