aiida_mcp.plugins.mlip module

MLIP calcjob tools: single point, geometry optimisation, MD and phonons.

aiida_mcp.plugins.mlip.is_available() → bool

Return whether the MLIP calcjob plugins are installed.

Returns:

True if the mlip.sp entry point is installed.

Return type:

bool

aiida_mcp.plugins.mlip.register(mcp: MCPServer) → None

Register the MLIP tools on the given MCP server.

Parameters:

mcp (MCPServer) – The server to register the MLIP tools on.

aiida_mcp.plugins.mlip.submit_md_job(code: str | int, structure: str | int, model: str | int | None = None, architecture: str | None = None, device: str | None = None, calc_kwargs: dict[str, Any] | None = None, ensemble: str | None = None, md_kwargs: dict[str, Any] | None = None, num_machines: int = 1, num_mpiprocs_per_machine: int = 1, label: str | None = None, description: str | None = None) → dict[str, Any]

Build the inputs for an MLIP molecular dynamics calcjob and submit it.

Parameters:
  • code (str or int) – Label (e.g. janus@localhost) or PK of the MLIP code.

  • structure (str or int) – Path to a structure file on the machine running the server, or the PK of an existing StructureData node.

  • model (str or int or None) – Path or URL of the MLIP model file, or the PK of an existing ModelData node. If omitted, the code’s own default model for architecture is used.

  • architecture (str or None) – MLIP architecture to use, e.g. mace_mp. Required unless model is the PK of an existing ModelData node, which already carries its own architecture.

  • device (str or None) – Device to run the calculation on: cpu, cuda or mps.

  • calc_kwargs (dict or None) – Extra keyword arguments to pass to the selected calculator.

  • ensemble (str or None) – Name of the thermodynamic ensemble.

  • md_kwargs (dict or None) – Keyword arguments for the molecular dynamics run.

  • num_machines (int) – Number of machines to run on.

  • num_mpiprocs_per_machine (int) – MPI processes per machine.

  • label (str or None) – Optional label to set on the calcjob node.

  • description (str or None) – Optional description to set on the calcjob node.

Returns:

The process PK, uuid, label, description, code, input node PKs, and daemon state.

Return type:

dict[str, Any]

Raises:

ToolError – If num_machines or num_mpiprocs_per_machine is below 1, code, structure or model does not match a known code or node, or an input fails validation.

aiida_mcp.plugins.mlip.submit_opt_job(code: str | int, structure: str | int, model: str | int | None = None, architecture: str | None = None, device: str | None = None, calc_kwargs: dict[str, Any] | None = None, opt_cell_fully: bool | None = None, opt_cell_lengths: bool | None = None, fmax: float | None = None, steps: int | None = None, minimize_kwargs: dict[str, Any] | None = None, pressure: float | None = None, num_machines: int = 1, num_mpiprocs_per_machine: int = 1, label: str | None = None, description: str | None = None) → dict[str, Any]

Build the inputs for an MLIP geometry optimisation calcjob and submit it.

Parameters:
  • code (str or int) – Label (e.g. janus@localhost) or PK of the MLIP code.

  • structure (str or int) – Path to a structure file on the machine running the server, or the PK of an existing StructureData node.

  • model (str or int or None) – Path or URL of the MLIP model file, or the PK of an existing ModelData node. If omitted, the code’s own default model for architecture is used.

  • architecture (str or None) – MLIP architecture to use, e.g. mace_mp. Required unless model is the PK of an existing ModelData node, which already carries its own architecture.

  • device (str or None) – Device to run the calculation on: cpu, cuda or mps.

  • calc_kwargs (dict or None) – Extra keyword arguments to pass to the selected calculator.

  • opt_cell_fully (bool or None) – Fully optimise the cell vectors, angles, and atomic positions.

  • opt_cell_lengths (bool or None) – Optimise cell vectors, as well as atomic positions.

  • fmax (float or None) – Maximum force for convergence.

  • steps (int or None) – Number of optimisation steps.

  • minimize_kwargs (dict or None) – All other keyword arguments to pass to the geometry optimizer.

  • pressure (float or None) – Scalar pressure when optimizing cell geometry, in GPa.

  • num_machines (int) – Number of machines to run on.

  • num_mpiprocs_per_machine (int) – MPI processes per machine.

  • label (str or None) – Optional label to set on the calcjob node.

  • description (str or None) – Optional description to set on the calcjob node.

Returns:

The process PK, uuid, label, description, code, input node PKs, and daemon state.

Return type:

dict[str, Any]

Raises:

ToolError – If num_machines or num_mpiprocs_per_machine is below 1, code, structure or model does not match a known code or node, or an input fails validation.

aiida_mcp.plugins.mlip.submit_ph_job(code: str | int, structure: str | int, model: str | int | None = None, architecture: str | None = None, device: str | None = None, calc_kwargs: dict[str, Any] | None = None, out: str | None = None, supercell: str | None = None, n_qpoints: int | None = None, displacement: float | None = None, no_hdf5: bool | None = None, dos: bool | None = None, pdos: bool | None = None, bands: bool | None = None, symmetrize: bool | None = None, qpoint_file: str | None = None, num_machines: int = 1, num_mpiprocs_per_machine: int = 1, label: str | None = None, description: str | None = None) → dict[str, Any]

Build the inputs for an MLIP phonon calcjob and submit it.

Parameters:
  • code (str or int) – Label (e.g. janus@localhost) or PK of the MLIP code.

  • structure (str or int) – Path to a structure file on the machine running the server, or the PK of an existing StructureData node.

  • model (str or int or None) – Path or URL of the MLIP model file, or the PK of an existing ModelData node. If omitted, the code’s own default model for architecture is used.

  • architecture (str or None) – MLIP architecture to use, e.g. mace_mp. Required unless model is the PK of an existing ModelData node, which already carries its own architecture.

  • device (str or None) – Device to run the calculation on: cpu, cuda or mps.

  • calc_kwargs (dict or None) – Extra keyword arguments to pass to the selected calculator.

  • out (str or None) – Name of the phonon output file.

  • supercell (str or None) – The size of supercells used in the phonon calculation.

  • n_qpoints (int or None) – Number of q points in the band path.

  • displacement (float or None) – Displacement for numerical derivatives.

  • no_hdf5 (bool or None) – Write force constants to the phonopy yaml, rather than separate HDF5.

  • dos (bool or None) – Calculate the density of states.

  • pdos (bool or None) – Calculate the partial density of states.

  • bands (bool or None) – Calculate the phonon band structure.

  • symmetrize (bool or None) – Symmetrize force constants.

  • qpoint_file (str or None) – The file for q-points in the phonon calculation.

  • num_machines (int) – Number of machines to run on.

  • num_mpiprocs_per_machine (int) – MPI processes per machine.

  • label (str or None) – Optional label to set on the calcjob node.

  • description (str or None) – Optional description to set on the calcjob node.

Returns:

The process PK, uuid, label, description, code, input node PKs, and daemon state.

Return type:

dict[str, Any]

Raises:

ToolError – If num_machines or num_mpiprocs_per_machine is below 1, code, structure or model does not match a known code or node, or an input fails validation.

aiida_mcp.plugins.mlip.submit_sp_job(code: str | int, structure: str | int, model: str | int | None = None, architecture: str | None = None, device: str | None = None, calc_kwargs: dict[str, Any] | None = None, out: str | None = None, properties: list[str] | None = None, num_machines: int = 1, num_mpiprocs_per_machine: int = 1, label: str | None = None, description: str | None = None) → dict[str, Any]

Build the inputs for an MLIP single point calcjob and submit it.

Parameters:
  • code (str or int) – Label (e.g. janus@localhost) or PK of the MLIP code.

  • structure (str or int) – Path to a structure file on the machine running the server, or the PK of an existing StructureData node.

  • model (str or int or None) – Path or URL of the MLIP model file, or the PK of an existing ModelData node. If omitted, the code’s own default model for architecture is used.

  • architecture (str or None) – MLIP architecture to use, e.g. mace_mp. Required unless model is the PK of an existing ModelData node, which already carries its own architecture.

  • device (str or None) – Device to run the calculation on: cpu, cuda or mps.

  • calc_kwargs (dict or None) – Extra keyword arguments to pass to the selected calculator.

  • out (str or None) – Name of the xyz output file.

  • properties (list[str] or None) – Properties to calculate, e.g. ["energy", "forces"].

  • num_machines (int) – Number of machines to run on.

  • num_mpiprocs_per_machine (int) – MPI processes per machine.

  • label (str or None) – Optional label to set on the calcjob node.

  • description (str or None) – Optional description to set on the calcjob node.

Returns:

The process PK, uuid, label, description, code, input node PKs, and daemon state.

Return type:

dict[str, Any]

Raises:

ToolError – If num_machines or num_mpiprocs_per_machine is below 1, code, structure or model does not match a known code or node, or an input fails validation.