Reference

These pages list every option FastMACE takes, with its values, default and effect. The user guide shows how to use them for common jobs. The API reference lists the functions and classes with their signatures.

How options reach a run

Every option is a field of a frozen record. Nothing is set through module-level switches or environment variables, except the few process-wide settings on Environment and cache.

Where you set it	What it takes	Page
fm.load(label, head=..., ...)	the model, head, charge, spin, trust	Models, D3, PolarMACE and schedules
fm.ForceField(model, atoms, precision=..., rewrites=..., dispersion=..., polar=...)	force precision, a RewriteConfig, D3, a PolarConfig	Rewrites and precision, PolarMACE
fm.Simulation(..., integrator=...) or optimizer=...	an fm.integrators or fm.optimize record	Integrators and optimisers
fm.Simulation(..., engine=..., neighbours=...)	EngineConfig, NeighbourConfig	Engine and neighbour list
fm.Simulation(..., mode=..., check=..., workload=..., fallback=..., gate=..., runner=..., adaptation=..., calibration=...)	the planner's inputs	Planner, gate and fallback
fm.md(..., mc=...), Simulation.run(..., mc=...)	fm.mc move records	Monte Carlo
fm.Simulation(..., output=fm.Output(...))	output intervals and files	Output files
fastmace md, fastmace run run.toml, ...	flags and TOML keys	Command line and TOML

fm.md takes most of these as keyword arguments too and passes them on.

The configuration records are not in the fm namespace. Import them from fastmace.config:

import fastmace as fm
from fastmace.config import EngineConfig, NeighbourConfig, RewriteConfig

model = fm.load("mh1", head="matpes_r2scan")
ff = fm.ForceField(model, atoms, rewrites=RewriteConfig(radial_tables="on"))
sim = fm.Simulation(atoms, ff, integrator=fm.integrators.Langevin(temperature_K=300, friction_per_fs=0.002), seed=1,
                    engine=EngineConfig(pipelined="off"), neighbours=NeighbourConfig(skin=0.5))
print(sim.plan.overrides())   # the options you set, and what the planner did with them

"auto", "on" and "off"

Many fields are toggles with three values.

• "auto" (the default) lets the planner decide, with a general rule based on the model, backend, precision and execution tier. The plan records the rule and the reason.
• "on" asks for the option. If the model or plan cannot use it, the planner raises PlanRefused with the reason before the first step. Nothing is substituted silently.
• "off" turns the option off.

print(sim.plan) shows the decisions that matter, and sim.plan.describe(full=True) shows every decision with its rule. The same text is in the run report and in metadata.json.

Units

Units follow ASE: Å, eV, eV/Å, K and amu. Times are in fs at the API. Pressures are in GPa wherever a name ends in _GPa.