Use when you have a written implementation plan to execute in a separate session with review checkpoints
npx skills add TaoELi/MaxwellLink --skill "mxl-driver-psi4-rtehrenfest"
Install specific skill from multi-skill repository
# Description
This skill should be used when users want to run the Psi4 RT-Ehrenfest driver (rtehrenfest) with MaxwellLink and need parameters, charge conventions, and socket/embedded invocation patterns.
# SKILL.md
name: mxl-driver-psi4-rtehrenfest
description: This skill should be used when users want to run the Psi4 RT-Ehrenfest driver (rtehrenfest) with MaxwellLink and need parameters, charge conventions, and socket/embedded invocation patterns.
Psi4 RT-Ehrenfest driver (rtehrenfest)
Confirm prerequisites
- Ensure
psi4is installed and importable in the driver environment. - Provide an XYZ geometry file whose second line specifies charge and multiplicity (e.g.,
0 1).
Configure socket mode
- Run:
mxl_driver --model rtehrenfest --address <host> --port <port> --param "molecule_xyz=../HCN_benchmark/hcn.xyz, functional=b3lyp, basis=cc-pvdz, dt_rttddft_au=0.04788, dft_grid_name=SG1, memory=2GB, num_threads=1, force_type=ehrenfest, n_fock_per_nuc=6, n_elec_per_fock=6"
Configure embedded mode
- Instantiate:
Molecule(driver="rtehrenfest", driver_kwargs={...})
Notes
- Use
force_type=bowhen nuclear gradients from Born–Oppenheimer surfaces are required; useforce_type=ehrenfestfor mean-field forces. - When passing
partial_chargeson the command line, use a space-separated list in brackets (no commas), e.g.partial_charges=[1.0 -1.0 0.0]. - On HPC, match
num_threadsto--cpus-per-task(single driver) or setnum_threads=1when launching many drivers in the same SLURM job. - Read full parameter docs in
docs/source/drivers/rtehrenfest.rst.
# Supported AI Coding Agents
This skill is compatible with the SKILL.md standard and works with all major AI coding agents:
Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.