Skip to main content

SEAMM MBE Plug-in

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

A SEAMM plug-in for many-body expansion (MBE) corrections of periodic cells and clusters

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

2026.10.6.1 – The levels run at the same time
  • The levels that run as batch tasks now run at the same time, with the whole cell and the periodic fragments started first. Their long calculations overlap the molecular ones instead of following them, roughly halving the time a frame takes. Levels that use an MDI engine still run one at a time.

2026.10.6 – A low level of their own for the triples
  • The new ‘triples low level’ gives the triples’ increments their own molecular low level, e.g. r2SCAN-D4 at def2-TZVPPD with the molecular low level at def2-QZVPPD for the pairs. Each triple’s increment is built entirely at that level, so the triples’ monomers and pairs are computed at both levels. It is shown only when there are triples; by default it is the molecular low level. A triples’ low level that is the molecular low level under another name is refused.

  • The increments file records each increment’s low level.

  • The user guide explains when to use the periodic low level for the monomers only.

  • Requires seamm-mbe 2026.10.6.

2026.10.5 – A high level of their own for the triples; the increments in a file
  • The new ‘triples high level’ gives the triples’ increments their own high level, e.g. revDSD at def2-TZVPPD with the high level at def2-QZVPPD for the monomers and pairs. Each triple’s increment is built entirely at that level, so the triples’ monomers and pairs are computed at both levels. It is shown only when there are triples; by default it is the high level. A triples’ level that is the high level under another name is refused, since every monomer and pair would be computed twice.

  • Each configuration’s increments are written to increments_c<configuration id>.json in the step’s directory, for later analysis: each fragment’s molecules, images, pair distances, levels, energy and forces, with triples marked “chain” or “closed”.

  • Requires seamm-mbe 2026.10.5.

2026.10.4.1 – Pairwise counterpoise corrects the high level only by default
  • ‘pairwise’ counterpoise now corrects each pair at the high level only. The new choice ‘pairwise, both levels’ also corrects the molecular low level, as ‘pairwise’ did before. In tests on Li⁺/BF₄⁻/water/EC clusters the r2SCAN-D4 correction was tiny where it could be computed, and its ghost calculations sometimes ended at spurious energies, while the revDSD correction was 3–4 kJ/mol per molecule.

  • The user guide notes that ORCA’s ghost-atom gradients are unreliable for double hybrids such as revDSD, so counterpoise-corrected forces from them need caution.

2026.10.4 – Requires seamm-exec 2026.10.3.2 for running on a cluster
  • Requires seamm-exec 2026.10.3.2: each bundle of calculations sent to a cluster’s queue now runs on one node, a calculation runs inside the job itself only when it fits there, and bundles get a time limit from their calculations’ estimated cost. Before, a VASP or ORCA calculation could be spread over several nodes and fail, or be cut off by the queue’s default time limit.

  • Checked end to end on TinkerCliffs: the 64-water pilot frame reproduces the prototype’s labels (energy to 0.001 meV, forces to 0.2 meV/Å RMS, pressures to about 1 atm).

2026.10.3 – Initial release: many-body expansion (MBE) corrections
  • A step that estimates high-level energies, forces and, for periodic cells, the stress of a cell or a large cluster: a low-level calculation of the whole system plus [high - low] corrections computed on monomers, pairs and triples of molecules, using the seamm_mbe library.

  • The levels are model chemistries: the high level, a molecular low level, and optionally a periodic low level (e.g. VASP) for the fragments of periodic cells, run as batches of calculations through seamm_exec.

  • Pair and triple cutoffs, a rule for which triples to include, and a check that the cell is large enough for the chosen cutoffs.

  • For a periodic low level, the fragments are placed on the cell’s FFT grid so that the grid errors cancel between the cell and its fragments.

  • Optional pairwise counterpoise correction of the basis-set superposition error.

  • Optional energy offsets per molecule type, and the results stored as properties and written to an extended XYZ file for training machine-learned force fields.

Metadata

Release files for mbe-step 2026.10.6.1

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

Source distribution (sdist)

Source distribution for mbe-step 2026.10.6.1
File Size Uploaded
mbe_step-2026.10.6.1.tar.gz 917.3 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for mbe-step 2026.10.6.1
File Interpreter ABI Platform
mbe_step-2026.10.6.1-py2.py3-none-any.whl Python 2, Python 3 none any Details

Total release size: 953.7 kB

Release files / mbe_step-2026.10.6.1.tar.gz

Download URL mbe_step-2026.10.6.1.tar.gz
Size 917.3 kB
Tags Source
SHA-256 checksum
How to use checksums
9d029ae957b1957d8d7399002a6cb68cb68c33b8235fa8000f5665a56249a096
BLAKE2b-256 checksum
How to use checksums
278f1008bf8ed787598b324453c13d4eb50e1c2e1582370f100a75c5a0a6242d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14

Release files / mbe_step-2026.10.6.1-py2.py3-none-any.whl

Download URL mbe_step-2026.10.6.1-py2.py3-none-any.whl
Size 36.4 kB
Tags Python 2 Python 3
SHA-256 checksum
How to use checksums
461b32bfc61cce1c3ed25808217700860a852b8578cd0320636596d1469dceba
BLAKE2b-256 checksum
How to use checksums
e4e0f90a7f720431b5b4201be744f7d66d64b4ef83ff47a68242c0d561d5760e
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.14
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