Skip to content

Per-file Analyses

This section documents analyses that operate on a single ReaxFF output file. Each page describes what can be computed using only one handler without requiring data from other files.

These analyses are the building blocks of ReaxKit; composed analyses build on top of them.

What “per-file” means in ReaxKit

A per-file analysis:

  • Uses exactly one ReaxFF file type
  • Depends on one handler (*_handler.py)
  • Produces derived quantities, plots, or summaries from that file alone
  • Is typically exposed via a CLI command of the form:
reaxkit <file-kind> <task> ...

Examples: - reaxkit xmolout get - reaxkit fort74 get - reaxkit energylog plot

Available per-file analysis groups

Each subpage below corresponds to a ReaxFF file type and its supported analyses. To better understand the following categories, we suggest looking at ReaxFF documentation.

Trajectory / structure files

  • xmolout
    Atomic coordinates, frames, trajectories, MSDs, basic geometry analysis.

  • vels
    Atomic velocities and velocity-derived statistics.

Energetics / thermodynamics

  • energylog / fort.73 / fort.58
    Energies vs iteration/time (total, bonded, non-bonded, electrostatic, etc.).

  • fort.76
    Stress, pressure, and related quantities.

  • fort.74 Scalar properties per iteration (e.g. volume, density, enthalpy).

  • fort.99
    Training / optimization targets and error metrics.

Connectivity / bonding

  • fort.7
    Bond orders, connectivity tables, coordination numbers.

  • fort.57 Bond statistics and related summaries.

  • ffield
    Force-field parameters (atoms, bonds, angles, torsions, h-bonds).

  • params
    Optimization parameters and bounds.

What each per-file page contains

Each file-specific page typically documents:

  1. Supported analyses (what can be computed)
  2. Required handler
  3. Expected DataFrame schema
  4. Python API examples
  5. CLI usage examples
  6. Aliases and column naming rules
  7. Notes on units, iterations, and frames

When to look at composed analyses instead

If you need to:

  • Combine coordinates + connectivity
  • Compute local or cluster-level properties
  • Derive quantities that depend on multiple files

→ See the composed analyses section instead.