Workflow overview

A typical pyNDUS calculation consists of four stages:

1. Generate sensitivity profiles with Serpent or ERANOS.

2. Obtain evaluated nuclear-data files in ENDF-6 format.

3. Process and parse multi-group covariance matrices with NJOY and SANDY.

4. Combine sensitivities and covariances using uncertainty, similarity, or representativity calculations.

The three public classes correspond to these stages:

Sensitivity

Reads, stores, extracts, collapses, and optionally merges sensitivity profiles.

Covariance

Processes or reads ERRORR covariance data and exposes covariance blocks by MF and MT identifiers.

Sandwich

Combines sensitivity profiles and covariance matrices and stores the requested numerical result.

The energy-group structure and ordering must be consistent between sensitivity profiles and covariance matrices. Multi-file sensitivity merging explicitly checks the group boundaries before combining the input data.