Output files

Every calphy calculation writes its results into a dedicated simulation folder created in the working directory. The folder is named

<mode>-<lattice>-<phase>-<temperature>-<pressure>

for example fe-fcc-solid-500-0 (an mode fe run on FCC, solid reference_phase, 500 K, 0 bar). A custom prefix can be prepended with the folder_prefix input key.

All plain-text data files (*.dat) carry a #-commented header line naming their columns and units, so they can be read directly with numpy.loadtxt(..., comments="#"), pandas, or any plotting tool. Unless noted otherwise the units follow LAMMPS metal units:

Quantity	Unit
energy	eV/atom
length	Å
volume	ų
pressure	bar
temperature	K
mean-squared displacement	Ų

---

report.yaml — the primary result

The machine-readable summary of the calculation. It has three blocks:

input — the thermodynamic state the run targeted:

Key	Meaning
temperature	Target temperature [K] (temperature)
pressure	Target pressure [bar] (pressure)
lattice	Input lattice / structure
element	Element(s) (element)
concentration	Composition of each element

average — quantities measured during equilibration:

Key	Meaning
vol_atom	Equilibrium volume per atom [ų]
spring_constant	Fitted Einstein spring constant(s) [eV/Ų] (solid reference only)
density	Number density [1/ų] (liquid reference only)

results — the free energy and its components (all in unit, eV/atom):

Key	Meaning
free_energy	The result — Helmholtz/Gibbs free energy per atom
error	Statistical uncertainty on free_energy: the standard error of the mean over the n_iterations independent switching runs
reference_system	Free energy of the analytic reference (Einstein crystal or Uhlenbeck–Ford model)
einstein_crystal	Einstein-crystal contribution (solid)
com_correction	Fixed-centre-of-mass finite-size correction (solid)
work	Reversible switching work between reference and system of interest
dissipation	Mean switching dissipation, 0.5·(W_forward + W_backward). A measure of the irreversibility of the switching — ideally close to 0; large values indicate the switching was too fast or a structural change occurred
ts_dissipation	(ts / tscale only) maximum energy dissipation along the reversible-scaling temperature sweep. Clean sweeps give ~1e-4 eV/atom; much larger values flag a hidden phase transition and a contaminated temperature_sweep.dat
pv	Pressure–volume contribution
unit	Units of the above (eV/atom)

---

Free-energy curves

Produced by the temperature/pressure-scaling modes. These are the files you plot to get a free-energy curve rather than a single point.

temperature_sweep.dat — mode ts / tscale

Free energy as a function of temperature from a reversible-scaling sweep.

Column	Meaning
1	temperature [K]
2	free_energy [eV/atom]
3	error [eV/atom] — standard error of the mean across iterations

The spacing of the temperature points is controlled by lambda_schedule (linear clusters samples at the low-temperature end; uniform_temperature spaces them evenly in temperature).

pressure_sweep.dat — mode pscale

Free energy as a function of pressure.

Column	Meaning
1	pressure [bar]
2	free_energy [eV/atom]
3	error [eV/atom]

---

Pre-flight temperature scan

Produced only when phase_transition_detection block is enabled (mode other than none) for a ts run. A fast real-thermostat ramp is used to detect the onset of a phase transition before the production sweep.

File	Content
prescan.forward.dat	The diagnostic ramp time series — columns dU [eV/atom], press [bar], vol [ų], temp [K]
prescan_signals.png	Plot of the fluctuation response signals with the detected onset/peak marked (best-effort; skipped if matplotlib is unavailable)

---

Raw switching data

The per-step data recorded during each non-equilibrium switching run. There is one forward/backward pair per iteration (n_iterations); calphy integrates these to obtain work and dissipation. They are mainly of interest for diagnostics (checking hysteresis, plotting the switching path).

forward_<i>.dat / backward_<i>.dat — mode fe

The Frenkel–Ladd / non-equilibrium switching between the system and its analytic reference for iteration i.

Solid reference

Column	Meaning
1	dU_sys — system potential energy [eV/atom]
2 … N+1	dU_ref — Einstein reference energy of each of the N elements [eV/atom]
last	lambda — switching parameter (0 → 1)

Liquid reference

Column	Meaning
1	dU_sys [eV/atom]
2	dU_ref — Uhlenbeck–Ford reference energy [eV/atom]
3	lambda

For the two-leg liquid path the files are split as forward_leg1_<i>.dat, forward_leg2_<i>.dat (and the backward_ equivalents).

forward_<i>.dat / backward_<i>.dat — mode alchemy

Column	Meaning
1	dU_1 — potential energy of system 1 [eV/atom]
2	dU_2 — potential energy of system 2 [eV/atom]
3	lambda

ts.forward_<i>.dat / ts.backward_<i>.dat — mode ts / tscale

The reversible-scaling sweep for iteration i.

Column	Meaning
1	dU — potential energy [eV/atom]
2	press [bar]
3	vol [ų]
4	lambda (mapped to temperature T₀/lambda)

ps.forward_<i>.dat / ps.backward_<i>.dat — mode pscale

Same four columns as ts.*, but lambda maps to pressure.

---

Averaging data

Time series written during the NPT/NVT equilibration and volume-convergence stage.

avg.dat

Column	Meaning
1	TimeStep
2–4	lx, ly, lz — box lengths [Å]
5	press [bar]
6	pe — potential energy [eV/atom]
7	etotal — total energy [eV/atom]
8	temp [K]

msd.dat

Column	Meaning
1	TimeStep
2 …	msd<k> — mean-squared displacement of element group k [Ų]

The MSD is used to fit the Einstein spring constant for the solid reference.

---

Trajectories and configurations

These are written in native LAMMPS formats and are only produced when the relevant input options are set (n_print_steps for trajectories):

File	Format	Content
traj.*.dat	LAMMPS dump	Atomic trajectory snapshots (equilibration, switching, melting checks)
conf.*.data	LAMMPS data	Equilibrated configurations (restartable structures)

---

Logs

File	Content
calphy.log	The human-readable run log — stage timing, convergence messages, warnings, and the citations to use
calphy.seg<k>.lmp	The generated LAMMPS input script for execution segment k
*.seglog	The LAMMPS log for each segment

The *.seg* files are produced by the executable runner (calphy drives the lmp binary segment by segment) and are primarily useful for debugging a failed run — the LAMMPS error message will be at the end of the relevant *.seglog.