Skip to main content

SEAMM Normal Mode Sampling Plug-in

GitHub pull requests Build Status Code Coverage Code Quality Documentation Status PyPi VERSION

A SEAMM plug-in for Wigner/thermal normal-mode sampling of the Hessian to generate displaced structures (e.g. for MLFF training sets)

Features

  • Please edit this section!

Acknowledgements

This package was created with the molssi-seamm/cookiecutter-seamm-plugin tool, which is based on the excellent Cookiecutter.

Developed by the Molecular Sciences Software Institute (MolSSI), which receives funding from the National Science Foundation under award CHE-2136142.

History

2026.11.5 – The Hessian from the shared driver, with the engine’s environment and cores
  • The Hessian now comes from the code the Thermochemistry and Vibrational Analysis steps share (seamm_exec 2026.11.5), with the same routes as before.

  • The MDI engine now runs with the program’s own environment – the modules it is installed with, ORCA’s MPI runtime – and on the job’s cores, as the Thermochemistry step’s did. A module-installed ORCA could not start its engine here before.

2026.10.3.1 – The finite-difference Hessian as separate calculations, on this machine or a cluster
  • When the job’s calculations go to a cluster queue and the program can run separate calculations (ORCA, MOPAC), the Hessian is the finite difference of the gradients, run there as 6N calculations; nothing is started on the job’s own machine, where the program may not be installed.

  • Otherwise the analytic Hessian is used when the program’s MDI engine has one, and ORCA’s model chemistry now says so without ORCA being started to ask. Failing that, ORCA’s finite-difference calculations run several at a time on this machine; MOPAC, xTB and MLFFs use the warm MDI engine as before.

  • Rerunning the job reuses finished finite-difference calculations.

  • If the program is not installed on this machine, the separate calculations are used and the output says so; other failures to start the engine stop the step with the reason rather than quietly switching method.

  • Requires seamm-exec 2026.10.3 or later.

2026.10.3 – Bugfix: the Hessian used the program’s method name and default basis
  • The MDI engine for the Hessian was launched with the model chemistry’s method name alone, so ORCA ran def2-SVP whatever basis was chosen, and a functional whose name the Model Chemistry step had to alter was not recognized. It now gets the program’s own keyword and the chosen basis (with model_chemistry_step 2026.10.3).

  • The shared CI now runs on uv: devtools/conda-envs/test_env.yaml is removed, so requirements.txt is the one dependency list.

2026.7.15 – Initial release
  • Generates an ensemble of displaced structures by normal-mode sampling of a molecule’s Hessian, for building machine-learned-force-field training sets and similar uses.

  • Amplitudes follow the quantum (Wigner) distribution by default, so stiff modes such as O-H stretches get the real zero-point spread that classical 300 K sampling misses; classical-thermal and ground-state (0 K) distributions are also available.

  • The Hessian is obtained from the Model Chemistry defined earlier in the flowchart, over MDI: the analytic Hessian when the engine provides one, otherwise a finite-difference of the forces over the resident engine.

  • A temperature, per-mode selection, amplitude cap, and harmonic-energy outlier rejection are all controllable, and the random seed can be fixed for reproducible ensembles.

  • Off-minimum geometries are allowed with a warning, so a transition state is sampled along its real modes while its imaginary reaction coordinate is left alone.

  • Each generated structure records its predicted harmonic energy and the per-mode normal-coordinate displacement as properties, for downstream filtering and analysis.

Metadata

Release files for normal-mode-sampling-step 2026.11.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for normal-mode-sampling-step 2026.11.5
File Size Uploaded Provenance Actions
normal_mode_sampling_step-2026.11.5.tar.gz 324.9 kB Without provenance Details for normal_mode_sampling_step-2026.11.5.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for normal-mode-sampling-step 2026.11.5
File Interpreter ABI Platform Actions
normal_mode_sampling_step-2026.11.5-py2.py3-none-any.whl Python 2, Python 3 none any Details for normal_mode_sampling_step-2026.11.5-py2.py3-none-any.whl

Total release size: 349.4 kB

Release files / normal_mode_sampling_step-2026.11.5.tar.gz

Download URL normal_mode_sampling_step-2026.11.5.tar.gz
Size 324.9 kB
Tags Source
SHA-256 checksum
How to use checksums
1ecde9102db10500b3c64121ebd59781bf9e70c69fa47c8d900855e80870e71c
BLAKE2b-256 checksum
How to use checksums
f1e6acd3c7ef10e10ca1c42c59541ec5da868c12133af61e13fa183347edd866
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.16

Release files / normal_mode_sampling_step-2026.11.5-py2.py3-none-any.whl

Download URL normal_mode_sampling_step-2026.11.5-py2.py3-none-any.whl
Size 24.5 kB
Tags Python 2 Python 3
SHA-256 checksum
How to use checksums
27b4cc5f3798333b55d16b011c16722e1de0b7161728a90e6c419e1d3406c425
BLAKE2b-256 checksum
How to use checksums
2fae83c00adf5ecfb85780452d580018d6efce99785c74f4bd51c2c3f6e02ef7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.16

Security for normal-mode-sampling-step 2026.11.5

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page