VCSMD¶
Classical variable-cell molecular dynamics in Python.
VCSMD combines a pure NumPy numerical core with a unit-aware Python interface. Describe a periodic structure with Pint quantities, choose a fixed-cell or variable-cell formulation, and evolve immutable states or write a complete, documented run to disk.
The package implements eight dynamics and minimization modes with a global Lennard–Jones potential and hydrostatic external pressure. It supports JSON, YAML, and TOML configurations, CSV results, native checkpoints, and one-way conversion of historical VCSMD data.
Start here¶
What you want to do |
Where to start |
|---|---|
Install VCSMD and run an original workload |
|
Download the complete original inputs |
|
Understand units, state, and the integration step |
|
Read the equations and physical assumptions |
|
Look up Python functions and data models |
|
Assess the available numerical evidence |
Functional core, explicit boundaries¶
The Python workflow is load_config → prepare → initialize → simulate.
Configuration objects retain physical units. Normalized models supply the
numerical solver with arrays and scalars. Public results expose dimensional
fields as Pint quantities, and published arrays are read-only.
The execution layer owns file writing, checkpoints, progress, and provenance. Each application run records its inputs and outputs in a dedicated folder. See the execution reference for that boundary.
Note
The supplied full workloads exercise fixed-dyn, metric-dyn, and metric-min.
The other five modes are implemented but lack supplied full-trajectory cases.
Electronic-structure calculations, tensor external stress, tabulated potentials,
and crystal standardization are outside the implemented scope. The
validation report describes the evidence and its limits.