openseespy-solvers 0.1.3

scipy-style PythonSparse solvers for OpenSeesPy

openseespy-solvers

scipy-style sparse linear and eigen solvers for OpenSeesPy PythonSparse.

openseespy-solvers wraps familiar numerical backends such as scipy.sparse.linalg, cupyx.scipy.sparse.linalg, and NVIDIA nvmath.sparse as solver objects that OpenSeesPy can call directly. OpenSeesPy assembles the stiffness, mass, and right-hand-side arrays; the solver object performs the sparse solve and writes the result back to OpenSeesPy.

Documentation: openseespy-solvers.readthedocs.io

Installation

python -m pip install openseespy-solvers

The base install provides NumPy and scipy.sparse.linalg CPU solvers. It requires Python 3.12 or newer.

OpenSeesPy is optional so the package can be used and tested without forcing an OpenSees install. If OpenSeesPy is not already in your environment, install the extra:

python -m pip install "openseespy-solvers[opensees]"

Optional backends:

# CPU UMFPACK direct solver
python -m pip install "openseespy-solvers[umfpack]"

# NVIDIA GPU backends: choose the CUDA generation reported by nvidia-smi
python -m pip install "openseespy-solvers[cuda13]"   # or [cuda12]

On Windows, install UMFPACK with conda-forge instead of pip:

conda install -c conda-forge scikit-umfpack

See the installation guide for platform notes, CUDA 12.x alternatives, and verification steps.

Quick Example

import openseespy.opensees as ops
from openseespy_solvers.scipy import spsolve

solver = spsolve()

# after defining the OpenSeesPy model:
ops.system("PythonSparse", solver.to_openseespy())

For eigen analysis:

from openseespy_solvers.scipy import eigsh

eig_solver = eigsh()
eigenvalues = ops.eigen("PythonSparse", 5, eig_solver.to_openseespy())

Default Solvers

Analysis	CPU	NVIDIA GPU
Static or transient linear solve	scipy.spsolve; scipy.umfpack for larger CPU systems	nvmath.direct_solver
Generalized eigen solve	scipy.eigsh	cupy.eigsh

Tutorial and API docs cover wiring, alternatives, and optional backends: openseespy-solvers.readthedocs.io.

Modules

Module	Provides
openseespy_solvers.scipy	CPU solvers: spsolve, umfpack, cg, gmres, eigsh, lobpcg
openseespy_solvers.scipy.precond	CPU preconditioners: jacobi, ilu, direct
openseespy_solvers.cupy	GPU solvers: spsolve, cg, gmres, eigsh, lobpcg
openseespy_solvers.cupy.precond	GPU preconditioners: jacobi, ilu, direct
openseespy_solvers.nvmath	GPU direct sparse solver: direct_solver
openseespy_solvers.hybrid	Direct factorization reused as a GMRES preconditioner

Factory signatures follow the corresponding scipy/cupy functions where possible. The matrix and right-hand side are supplied by OpenSeesPy at solve time.

Examples and Development

The pip wheel contains the library. Examples, benchmarks, and tests live in the source repository:

git clone https://github.com/gaaraujo/openseespy-solvers.git
cd openseespy-solvers
python -m pip install -e ".[dev,opensees]"
pytest

Then run a smoke example:

cd examples
python solvers/scipy_spsolve.py
python solvers/scipy_eigsh.py

Project Links

• Documentation: openseespy-solvers.readthedocs.io
• Source: github.com/gaaraujo/openseespy-solvers
• Issues: github.com/gaaraujo/openseespy-solvers/issues

License

BSD 3-Clause. See LICENSE.