Skip to content

Fort99 Handler

ReaxFF training set error report (fort.99) handler.

This module provides a handler for parsing ReaxFF fort.99 files, which summarize force-field training errors by category and target during parameter optimization runs.

Typical use cases include:

  • analyzing training set contributions to total error
  • inspecting charge, geometry, and energy fitting quality
  • building diagnostics for force-field parameterization workflows

Fort99Handler

Bases: BaseHandler

Parser for ReaxFF training set error reports (fort.99).

This class parses fort.99 files and exposes individual training targets and their contributions to the total force-field error as a structured tabular dataset.

Parsed Data

Summary table One row per training target, returned by dataframe(), with columns: ["lineno", "section", "title", "ffield_value", "qm_value", "weight", "error", "total_ff_error"]

The ``section`` column categorizes each target as one of:
["CHARGE", "HEATFO", "GEOMETRY", "CELL PARAMETERS", "ENERGY", None].

Metadata Returned by metadata(), containing: ["n_records", "n_frames"]

Notes
  • The last five numeric values on each line are interpreted as (FF value, QM/reference value, weight, error, total error).
  • Section categories are inferred heuristically from the title text.
  • Unrecognized entries are retained with section=None.
  • This handler is not frame-based; n_frames() always returns 0.