Roundtable
Doga
- Working on Fast Jet class for ACTS. This will be in the Examples and Plugins in ACTS
- The PR is already ready and can be reviewed
Andreas Stefl
- truth tracking for electron finding studies in ACTS
- inefficiencies in odd (unexpected)
- SP were sorted by time - it flips in pix the space point - removed now
- particles can slip through layers in odd without hitting any surface - but not to be touched now
- restriction on the pixel measurements and filtering out insufficient measurements
- changes in truth seeder and particle seeder
- odd layout not tuned, which is possible cause of the inefficiency
Carlo Vani
- Recent plots (regarding trigger?) requested by EFTracking are mostly addressed
- Definitions in IDPVM efficiency plots are ambiguous: understanding the denominator?
- Discussion on regional reconstruction with CPU VS. GPU
Alex
- ACTS spack installation, with help of Dimitra building a full chain of muons with Chisquare fitter for muons
- Can generate muon SP and convert them into measurements
- Importing geometry is the next step
- Some parts are missing from the full chain
- Incoherent input data - had to add blanks for testing purposes
Got almost through, setting up a prototype chain to reconstruct muon spce points with the gx2f.
Neza
Making CTD plots. Timing constraints CPU vs GPU.
- Debugging for the inefficiency gap
- Muons are looking fine
- Timing comparison for CPU VS. GPU - breakdown into pie charts with suggestion from Noemi
- Comparison between C-230 and G-230 , C-230, G-230 and G-050
Mikkel
- GNN on pixel only for seeding in ACTS/ATHENA
- Compiles but doesn't run - debugging
Started QP with gbts
Attila
- Report for DUNE experiment - they are writing a conceptual design report for their software - preliminary report is done
- Working also on Analysis I/O code - readability of ntuples - smaller files - faster writing
PF
Tune trackfinding.
- Tuning of track finding: legacy track finder uses strip width as error, acts strip pitch
- Stripcalibrator added (with Noemi) to do ascts-based track finding
- Efficiency checks with strip pitch VS. strip width:
- chi2 changed 50->100
- Technical efficiency: Improvement in barrel region by 1%
- Physics efficiency: almost matching C-000 in the barrel
- In technical efficiency pions are forced to traverse into strip but in the physics efficiency pions get absorbed
- Cost: higher fake rate
- No tuning at chi2 yet - fake rate VS. track pt
Luthien
- Starting with neural network for pixel splitting classification
- Converting Max (Hart)'s model into onnx and implementing it in Athena
Dimitra
- Seeding for new small wheel
- Monitoring plots that are added are too small, works on optimizing this
- Change of names of some algorithms for clarification for the plots
- In the geometry there are unmatched simhits on STGC layers
- Simhits are not right on the layer, but a step ahead (non-zero z value is observed )
- Possible reason is with post processing with Geant4
Working with Alex on drift circle fit with the gx2f.
