Dakota  Version 7.0
Explore and Predict with Confidence
Modifying Dakota's Input Specification

Dakota's canonical input grammar is the Pydantic model tree under python/dakota/spec. The parser under packages/dakota_parser/dakota_parser consumes the generated JSON Schema and materializes validated input into the map-based intermediate representation used by ProblemDescDB.

See Pydantic Input Grammar: Complete Reference and Procedure for the complete procedure for adding a keyword, authoring DakotaField materialization metadata, assigning IR defaults, implementing validation, regenerating parser and IR sources, and testing the result. Legacy NIDR handlers and Data classes are no longer part of this workflow.