How ammelsayed/hepdataset works
HEPDataset turns Delphes ROOT inputs and a samples card into selected, analysis-ready ROOT trees, with summaries and optional merging or parallel event processing. The main path is CLI dispatch → dataset orchestration → sample parsing and loop selection → event/object selection and configured feature building → ROOT output. Other documented capabilities include standalone loop modes, sample-card inspection/conversion, and output merging. Edges are limited to relationships shown in excerpts, indexed references, or README; several loop internals and integrations are unsampled.
Main entry points: make_dataset.py Dataset orchestration · adaptive_delphes.py Adaptive loop · object_selection.py Object selection · cli.py CLI dispatcher
Components 19
Entry and orchestration
- CLI dispatcher
src/cli.py - Dataset orchestration
src/make_dataset.py
Inputs and configuration
- Samples reader
src/samples_reader.py - Branch-card reader
src/core/branches_reader.py - Packaged defaults
src/defaults
Event analysis
- Delphes access
src/core/delphes_utilis.py - Object selection
src/core/object_selection.py - Channel selection
src/core/event_selection.py - Kinematics features
src/core/kinematics.py
Loop execution
- Adaptive loop
src/loops/adaptive_delphes.py - Basic loops
src/loops - Loop arguments
src/loops/loop_utilis.py - Parallel processing
src/loops/parallel_loop.py
Output management
- Selected ROOT trees
- Selection summaries
- Sample merger
src/merge_samples.py
Other components
- Analyst
- ROOT inputs
- Sample-card user
Connections 27
- Analyst → CLI dispatcher: invokes
README.md - CLI dispatcher → Dataset orchestration: dispatches to
- Dataset orchestration → Samples reader: reads card
src/make_dataset.py - Dataset orchestration → Adaptive loop: loads selected loop
src/make_dataset.py - Dataset orchestration → Branch-card reader: configures features
- Dataset orchestration → Packaged defaults: uses defaults
README.md - Dataset orchestration → Parallel processing: runs chunks
src/make_dataset.py - Dataset orchestration → Object selection: creates selector
src/make_dataset.py - Dataset orchestration → Channel selection: creates selector
src/make_dataset.py - Adaptive loop → Delphes access: builds tree chain
src/loops/adaptive_delphes.py - Adaptive loop → Branch-card reader: reads config
src/loops/adaptive_delphes.py - Adaptive loop → Object selection: selects objects
src/loops/adaptive_delphes.py - Adaptive loop → Channel selection: classifies events
src/loops/adaptive_delphes.py - Adaptive loop → Kinematics features: computes features
src/loops/adaptive_delphes.py - Adaptive loop → Selected ROOT trees: writes trees
src/loops/adaptive_delphes.py - Adaptive loop → Selection summaries: writes summaries
src/loops/adaptive_delphes.py - Object selection → Kinematics features: uses observables
src/core/object_selection.py - Basic loops → Object selection: uses selector
src/loops/basic2_delphes.py - Basic loops → Channel selection: uses selector
src/loops/basic2_delphes.py - Parallel processing → Selected ROOT trees: merges chunks
src/loops/parallel_loop.py - ROOT inputs → Samples reader: lists file paths
- ROOT inputs → Delphes access: provides events
- Sample-card user → Samples reader: inspects or converts
README.md - Sample-card user → Sample merger: requests merge
README.md - Sample merger → Selected ROOT trees: merges outputs
- CLI dispatcher → Basic loops: dispatches loop
README.md - Loop arguments → Parallel processing: configures execution
src/loops/loop_utilis.py