-
Notifications
You must be signed in to change notification settings - Fork 5
Description
InpManager now supports a typed FitConfig + InpDatasetMetadata interface, but the higher-level PleiadesConfig orchestration is still disconnected.
Current docs and call patterns show direct FitConfig construction, while production workflow is expected to start from PleiadesConfig (workspace, datasets, fit_routines, global nuclear isotope config).
I would like to add first-class integration so INP generation can be driven directly from information in PleiadesConfig, including consistent mapping of PleiadesConfig metadata (min/max energy, element/mass hints, density/thickness/atomic mass, temperature), routine fit_config overrides, and isotope population from global nuclear config (populate_fit_config_isotopes).
Proposed scope: add an adapter/constructor path in InpManager (or a nearby orchestration utility) that takes PleiadesConfig + routine_id and produces INP deterministically, update docs to include this end-to-end flow, and add unit/integration tests validating round-trip behavior for routine selection, metadata precedence rules, and missing/invalid config handling.