Advanced Diagnostics And Validation

Use this section after the user-facing OSSG/MSG/physics result is understood. Advanced fields explain how the route reached that result and how data move between settings; they are not additional material properties.

When Advanced Output Is Appropriate

  • an OSSG label changes under small tolerance variations;
  • an input, convention, database-standard, or ACC operation list disagrees;
  • a transformed cell cannot be paired with its SSG operations;
  • an input-cell operation export warns that the cell is not magnetic primitive;
  • SCIF/POSCAR round trips change the identification;
  • a batch regression differs from an accepted baseline.

For ordinary interpretation, stay with Interpret Your Result.

Settings At A Glance

Setting Main purpose Typical user product
input Preserve the user-supplied cell Interoperability with an external code
input_magnetic_primitive Remove magnetic-cell supercell redundancy Primitive reference for input-cell warnings
database_standard Match identify-index database conventions Reproduction and route diagnostics
convention Public OSSG/GSPG presentation Symbols and human-facing operations
acc_primitive ACC-aligned magnetic primitive cell Matched VASP POSCAR/KPOINTS
acc_conventional Conventional ACC presentation Comparison/classification workflows

A valid transformation chain must keep the cell and SSG paired. If a selected setting transform cannot carry both through the required chain, that failure is diagnostic evidence; silently substituting an unrelated basis would hide the problem.

Diagnostic Payloads

identify_index_details. Exact identification components, selected database records, and route evidence.

acc_primitive_resolution_audit. How the ACC primitive transform/setting was resolved.

transforms.audit in StructuredResult. Audits for standard-to-ACC and convention transform chains.

operation_views. Named complete/generator/nontrivial/MSG operation views. Each view carries a real-space and spin-frame setting.

tolerances. Numerical thresholds that belong to the result provenance.

Validation Order

When a result is surprising:

  1. verify input moments, occupancies, and coordinate conventions;
  2. reproduce quick analysis with defaults;
  3. vary one justified tolerance at a time;
  4. compare index, MSG, conf, and phase evidence;
  5. only then inspect identify-index, transforms, and operation settings;
  6. validate generated files by parsing/rerunning them through the intended public route;
  7. use the batch workflow for broad regression claims.

See Parameters And Reliability before interpreting a tolerance-dependent change.

Batch Regression Evidence

The repository batch workflow records baseline, comparison, and error artifacts. A clean single example is not evidence that a transform/output change is safe across SSG types. Changes to standard-cell selection, operation transforms, generated SCIF/POSCAR, or public schemas should use the relevant focused tests and the agreed batch suite.

Operational commands are maintained in the repository batch workflow guide. Accepted baselines should record code version and tolerance profile; do not silently overwrite an accepted successful case with a different result.

Scientific Boundary

Route diagnostics can establish reproducibility and internal consistency. They cannot establish energetic stability, response magnitude, experimental domain population, or transition kinetics. Those require the appropriate electronic- structure, thermodynamic, or experimental analysis.