2026-07-15 Normal Mode Sampling – initial development

Build the Normal Mode Sampling plug-in: given a molecule and its Hessian, draw an ensemble of displaced geometries by Wigner (quantum) normal-mode sampling and write them as configurations for downstream single-point labelling. The immediate driver is the internal-degrees-of-freedom (1-body) layer of a machine-learned-force-field (MLFF) training set.

This is the SEAMM implementation of decision #2 of the water → electrolyte MLFF training-set design. That campaign’s science home is the ~/Sites lab notebook – the design document mlff-training/2026-07-15_water-mlff-training-plan/ – not this repository. Keep science decisions there and implementation notes here, with light cross-links (see scope for the back-pointers).

Contents:

• Normal Mode Sampling – scope and design
  • Goal
  • Where this fits (cross-project)
  • The science: why Wigner, not classical
  • How it works
  • Parameters (draft)
  • GUI (prevent invalid combinations)
  • Completeness (not just the count)
  • Out of scope (v1)
  • Open questions
  • References