Skip to main content
Pre-release

This release is a pre-release and may not be stable for production use.

pyscf-vscf

pyscf-vscf: an H2O symmetric-stretch mode over a coupled two-coordinate potential-energy surface

pyscf-vscf is a public-alpha package for PySCF-backed construction and reduced-dimensional variational analysis of vibrational potential-energy and dipole-moment surfaces.

The package now contains an actual state-specific VSCF solver for n-mode Hamiltonians with one-mode terms and explicit two-mode coupling corrections. It also provides 1D/2D sinc-DVR solvers, overlap-based 2D state assignment, harmonic and scan workflows, exact constrained normal-coordinate relaxation, and provenance-checked PES/DMS caches.

VCI is not implemented. The present VSCF is reduced-dimensional and uses rectilinear uniform DVR coordinates with diagonal one-mode kinetic operators. The package does not claim full-dimensional or general curvilinear VSCF/VCI.

Install

Install from PyPI:

pip install pyscf-vscf

The source project uses uv for development:

uv sync --extra dev

PySCF is the supported electronic-structure backend and is included in the standard installation. Electronic methods are specified using PySCF's native method syntax. Pure numerical modules and the VSCF solver avoid importing PySCF until an electronic-structure workflow is requested. On Windows, install and run the package under WSL because PySCF does not support native Windows installations.

Runnable VSCF Example

The wheel includes a deterministic coupled two-mode example that performs no electronic-structure calculations:

uv run python -m pyscf_vscf.examples.vscf_two_mode

The main API is:

from pyscf_vscf import NModePotential, VSCFSettings, vscf_spectrum

NModePotential represents

H = sum_i [T_i + V_i(q_i)] + sum_{i<j} V_ij(q_i, q_j)

where each V_ij is a coupling correction rather than a complete two-mode potential. Coordinates supplied to NModePotential must be in Angstrom.

The current VSCF implementation consumes a one-mode/two-mode PES through the Python API. nmode_model_from_pair_surfaces assembles that representation from overlapping complete pair surfaces, rejects inconsistent shared one-mode cuts, and removes independent pair-energy offsets. PySCF does not yet generate an arbitrary molecular n-mode PES or DMS automatically. VSCF transition intensities and a VSCF command-line workflow are also not implemented.

Command Line

The geometry commands below are source-checkout examples; geom/ is repository validation material and is not installed into the wheel:

pyscf-vscf --help
pyscf-vscf --version
pyscf-vscf --mmol geom/HDO.mmol --task harmonic
pyscf-vscf --mmol geom/H2O.mmol --task 1d --bond 0-1 --npts 41

The legacy pyscf_pme_pipeline.py remains in the source repository for historical compatibility, but release support and API guarantees apply to the package code under src/pyscf_vscf.

XYZ and MMOL files do not reliably preserve molecular charge and spin. Supply --charge and --spin whenever a non-neutral or open-shell geometry is reused.

Scientific Conventions

  • Transition dipoles are reported in Debye, never arbitrary units.
  • Integrated cross sections are integral sigma(omega) d omega in m^2/s.
  • Axis-projected values are for polarized light; vector-norm values use the isotropic 1/3 orientation average.
  • 2D state labels use phase-canonical wavefunction overlaps and expose unique labels, mixed-state signatures, participation ratios, dominant-manifold weights, and leading signed components.
  • Schema-v2 caches fingerprint geometry, isotope masses, charge/spin, every electronic-structure setting, scan coordinates, runtime versions, and array checksums. Legacy caches fail closed in production loaders.

See the intensity conventions and validation guide.

Validation

uv run ruff check src tests scripts/validate_archived_grids.py \
  scripts/generate_nh3_three_mode.py scripts/expand_nh3_three_mode.py \
  scripts/validate_nh3_three_mode.py
uv run ruff format --check src tests scripts/validate_archived_grids.py \
  scripts/generate_nh3_three_mode.py scripts/expand_nh3_three_mode.py \
  scripts/validate_nh3_three_mode.py
uv run pytest -m "not pyscf" -q
uv run pytest -m pyscf -q
uv run python scripts/validate_archived_grids.py \
  --nmax 12 --output validation_data/convergence_report.json
uv run python scripts/validate_nh3_three_mode.py

The archived-grid analysis performs no electronic-structure recomputation. It reports state-matched spreads across grid density and coordinate-window variants for H2O, HDO, and D2O. It also compares the two assigned stretch fundamentals of each isotopologue against independent ORCA harmonic IR intensities. The six calculated intensities reproduce the independent scale and isotope/mode trend within 35.4% relative error. This is a computational cross-check, not an experimental or rovibrational validation.

On the same archived molecular surfaces and a matched separable kinetic model, state-specific VSCF is also compared with exact 2D DVR for both fundamentals and their combination state. This is a molecular solver benchmark, not an independent validation of the underlying PES or omitted kinetic couplings.

The separate NH3 archive exercises three local N-H coordinates and all three pair surfaces. State-specific VSCF is compared with exact 3D DVR on the identical 1MR/2MR Hamiltonian across a documented window-convergence sequence. The fundamental, binary-combination, and triple-combination manifolds pass the 25 cm^-1 centroid-spread criterion; the first-overtone manifold is retained as nonconverged. Across the final three variants, the maximum VSCF/exact centroid error is 4.43 cm^-1. Reanalysis uses only the checked-in caches and does not require PySCF.

Metadata

Release files for pyscf-vscf 0.1.0a5

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

Source distribution (sdist)

Source distribution for pyscf-vscf 0.1.0a5
File Size Uploaded
pyscf_vscf-0.1.0a5.tar.gz 1.1 MB Details

Built distribution (wheel)

Table of built distributions (wheels) for pyscf-vscf 0.1.0a5
File Interpreter ABI Platform
pyscf_vscf-0.1.0a5-py3-none-any.whl Python 3 none any Details

Total release size: 1.1 MB

Release files / pyscf_vscf-0.1.0a5.tar.gz

Download URL pyscf_vscf-0.1.0a5.tar.gz
Size 1.1 MB
Tags Source
SHA-256 checksum
How to use checksums
df7468b837d464b59a427b5f99b47e82d7551691bc46993991a31c9aac5c6067
BLAKE2b-256 checksum
How to use checksums
c91d1220de4c814e8bca3ee097ab40500d76d81bed149ec525bb118675329ad2
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

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.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 8, 2026.

Transparency log

Release files / pyscf_vscf-0.1.0a5-py3-none-any.whl

Download URL pyscf_vscf-0.1.0a5-py3-none-any.whl
Size 68.2 kB
Tags Python 3
SHA-256 checksum
How to use checksums
e4ba75d9049aa80399907ef738f1d759cd6d4d5a4c4edd5b88a1a90c1e830032
BLAKE2b-256 checksum
How to use checksums
64dfacd8303972d1ffadb3dbe72e4f4530dd83a34d851efbce0fb7aeca089610
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

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.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 8, 2026.

Transparency log
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