stfc / alc-dls-exafs

50%

LAST BUILD ON BRANCH main

branch: SELECT

CHANGE BRANCH

x

Sync Branches

• No branch selected
• add_readme
• average_over_sites
• cff
• ci
• clarify-plot-include-help
• containers_2
• debye-waller
• dependabot/uv/cryptography-46.0.5
• dependabot/uv/cryptography-46.0.6
• dependabot/uv/cryptography-46.0.7
• dependabot/uv/filelock-3.20.3
• dependabot/uv/fonttools-4.60.2
• dependabot/uv/idna-3.15
• dependabot/uv/lxml-6.1.0
• dependabot/uv/marimo-0.23.0
• dependabot/uv/orjson-3.11.6
• dependabot/uv/pillow-12.1.1
• dependabot/uv/pillow-12.2.0
• dependabot/uv/pip-26.0
• dependabot/uv/pip-26.1
• dependabot/uv/pygments-2.20.0
• dependabot/uv/pymdown-extensions-10.21.3
• dependabot/uv/pynacl-1.6.2
• dependabot/uv/pytest-9.0.3
• dependabot/uv/python-dotenv-1.2.2
• dependabot/uv/requests-2.33.0
• dependabot/uv/starlette-1.0.1
• dependabot/uv/urllib3-2.6.3
• dependabot/uv/urllib3-2.7.0
• dependabot/uv/virtualenv-20.36.1
• extract_paths
• feat-arbitrary-custom-tags
• fix_cli_generate_typo
• fix_pipeline_all-frames-cli-opption
• main
• minmax
• move_marimo_utils
• notebook-advanced-tab
• notebooks
• reuse_potentials
• update-default-exchange-in-feff-configs
• validation_fix
• write_data

Build # 27427257496

Build Type

push

github

Committed by web-flow

Commit Message

Extract paths (#70)

* WIP Refactor to use HDF5 files to store the FEFF otuput including path information

Storing all the paths for all the frames/sites generates a very large number of files. This approach organises and compresses that output into an easy-to-transport single archive file. HDF5 was chosen as it's flexible saves the FeffConfig metadata also) and is already a dependency from larch.

Also added the option to filter out hydrogens as they typically don't contribute to EXAFS and can cause some issue with FEFF.

* feat: add curved-wave amplitude ratio (cw_ratio) path filtering

Parse FEFF's files.dat to extract the curved-wave chi amplitude ratio
for each scattering path and thread it through the full pipeline:

- feff_utils: add parse_files_dat() to read cw_ratio from files.dat
- hdf5_store: store and read cw_ratio as an HDF5 attribute per path
- exafs_data: add cw_ratio field to PathContribution; average across
  samples in PathAggregator.finalize(); add filter_path_contributions()
  utility (filters by min_cw_ratio threshold and/or top_n cap, preserving
  r_eff ordering among survivors)
- cli: add --min-cw-ratio option to both `analyze` and `pipeline` commands
- tests: add TestParseFilesDat and TestFilterPathContributions
- scripts: add pack_output_to_hdf5.py to import an existing
  pipeline_output/ tree into HDF5 without re-running FEFF
- README: document HDF5 output mode and the pack script

* change all return sites and unpack sites of (chi, k) → (k, chi) to match the documented convention.

* feat: store raw FEFF parameters and MD-averaged path contributions in HDF5

- `PathContribution` gains `amp`, `pha`, `lam`, `rep`, `k_param` (raw FEFF
  params for on-the-fly chi recomputation), `source_frames`, `source_sites`
  (MD provenance), and `contribution_pct`; `PathAggregator` averages and
  populates all new fields
- `iter_path_contributions` yields `(path_key, info, frame_index, site_index)`
  4-tuples; all call sites updated to unpac... (continued)

Coverage Stats

1883 of 3737 relevant lines covered (50.39%)

2.02 hits per line