cranmer
commented
5 years ago Example 1:
- analysis use case Z peak see #6
- implementation with TBranch::getentry
- data set see #6
- some user experience reported at ACAT
- links: see #6
- see also CMS open data example "several hours" http://opendata.cern.ch/record/101
Example 2:
- CMS Higgs 4-lepton analysis with Open Data
- implementation was Kubernetes + containerized ROOT 6 analysis
- demo'd at KubeCon 2019
- metrics and computing system detailed in talk
Example 3:
- Template fitting analysis example connected to analysis systems
- either ttH or SUSY analysis
- both implemented with ATLAS toolkit including HistFitter and TRexFitter
- time to make ntuples, do fits, etc. from @Alexander-Held
Example 4:
- Recast and reinterpretation example based on ATLAS pub note.
- pyhf configuration on HEPData:
(associated with HEPData entry 
)
matthewfeickert
alexander-held
Assumptions
Existing systems refers to analysis tools that existed prior to IRIS-HEP and that are currently used in LHC experiments. For example, bare ROOT, ATLAS EventLoop, CMSSW.
It should be possible to re-implement the analysis examples that are chosen by the analysis systems group for benchmarking for later comparisons.
This uses analysis use cases defined in #3.
Acceptance Criteria
A benchmark should be described by the analysis use case, the implementation details, and relevant performance metrics such as I/O throughput, even processing rates, wall clock time, and ease of use.