rslab (Python bindings)
NumPy/SciPy bindings for RSLAB, a
pure-Rust sparse direct solver: symmetric and complex-symmetric LDL^T
(Bunch-Kaufman), unsymmetric LU, and a KLU path for circuit-shaped matrices.
All numeric work happens in Rust. Supported dtypes: float64, float32,
complex128, complex64.
pip install rslab
Usage
import numpy as np
import scipy.sparse as sp
import rslab
x = rslab.spsolve(A, b) # one-shot, on the automatic path
f = rslab.ldlt(A) # symmetric or complex-symmetric
x = f.solve(b)
X = f.solve_many(B) # n x nrhs
print(f.factor_nnz, f.inertia, f.diagnostics()["summary"])
f = rslab.lu(A_general) # unsymmetric
y = f.solve_transpose(b) # A.T @ y = b, on every handle
Circuit-shaped matrices (MNA / SPICE class) take the KLU path, with a numeric-only refactor for fixed-pattern sweeps and its factors exported:
f = rslab.klu(A)
x = f.solve(b)
f.refactor(A_next.data) # same pattern, new values, no pivot search
y = f.solve_transpose(b) # A.T @ y = b (conjugate b and y for A^H)
L, U, F = f.L, f.U, f.F # (R A)[perm_r][:, perm_c] = L @ U + F
The automatic path picks among the three by the work the analysis predicts: KLU for sparse work, LDL^T for symmetric values or LU for dense work.
f = rslab.auto(A)
print(f.backend) # 'klu', 'ldlt' or 'lu'
f.refactor(A_next.data) # in place, on the path the values take
Symbolic reuse
The analysis depends on the pattern only; pay it once per sweep:
sym = rslab.analyze(A) # AutoSymbolic; path='ldlt', 'lu' or 'klu' for one path
plan = sym.memory_plan(A.dtype.name, threads=8) # heap before any numeric work
assert plan["peak_bytes"] < available_bytes
for omega in frequencies:
f = sym.factor(K + 1j * omega * C) # the matrix or its data array
x = f.solve(b)
Preconditioners and Krylov solvers
Static pivoting never fails; refinement or a Krylov method recovers the
accuracy. gmres, gmres_block, cocg and cocr take any factor handle as
preconditioner:
f = rslab.ldlt(A, preconditioner=1e-4)
x = f.solve(b, refine=20)
M = rslab.lu(A, factor_dtype="complex64") # complex128 A, half the factor memory
x, converged, iters, res, stop = rslab.gmres(A, b, M, tol=1e-10)
x = M.solve(b, refine=5) # mixed-precision refinement, full accuracy
Settings
Factor knobs are keyword arguments of auto, ldlt, lu, spsolve and analyze,
or a rslab.Settings object (rslab.KluSettings for klu). Every tuning
constant of the solver is one (nd_fm_passes, race_candidates,
solve_block, ...), with the tuned value as default;
rslab.Settings().to_dict() lists them all. Common overrides:
f = rslab.lu(A, ordering="metis", threads=8, pivot_threshold=0.5)
s = rslab.Settings(preconditioner=1e-6, drop_tol=1e-3)
Settings a path does not read are listed under
diagnostics()["warnings"]. Every keyword, handle and method is documented
in the generated reference docs/api.md (help(rslab.lu)
shows the same text).
Logging
rslab.set_log_level("info") # or the RLA_LOG environment variable
log = logging.getLogger("rslab")
rslab.set_log_sink(lambda level, msg: log.log(logging.getLevelName(level.upper()), msg))
License
MIT.
1 maintainer
Metadata
1 maintainer
Release files for rslab 1.5.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| rslab-1.5.0.tar.gz | 595.4 kB | Details |
Built distributions (wheels)
| File | Reset | |||
|---|---|---|---|---|
| rslab-1.5.0-cp39-abi3-win_amd64.whl | CPython 3.9 | abi3 | Windows x86-64 | Details |
| rslab-1.5.0-cp39-abi3-manylinux_2_28_x86_64.whl | CPython 3.9 | abi3 | Linux glibc 2.28+ x86-64 | Details |
| rslab-1.5.0-cp39-abi3-manylinux_2_28_aarch64.whl | CPython 3.9 | abi3 | Linux glibc 2.28+ ARM64 | Details |
| rslab-1.5.0-cp39-abi3-macosx_11_0_arm64.whl | CPython 3.9 | abi3 | macOS 11.0+ ARM64 | Details |
| rslab-1.5.0-cp39-abi3-macosx_10_12_x86_64.whl | CPython 3.9 | abi3 | macOS 10.12+ x86-64 | Details |
Total release size: 16.2 MB
Release files / rslab-1.5.0.tar.gz
| Download URL | rslab-1.5.0.tar.gz |
|---|---|
| Size | 595.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
bbbbd71b08bf6e4524c0e88de9d4f3834e393343988ff261381e408292e29a79
|
|
BLAKE2b-256 checksum How to use checksums |
ca3a1cf038bf1fa72ac18cb08210d8877039b0a9e9555c907c35919c38bfd334
|
| 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 Oct 16, 2026.
Transparency logRelease files / rslab-1.5.0-cp39-abi3-win_amd64.whl
| Download URL | rslab-1.5.0-cp39-abi3-win_amd64.whl |
|---|---|
| Size | 3.2 MB |
| Tags | CPython 3.9 Windows x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
d846798d2b972ba8680311b05e7ec0853d37953f371fe185783065245b258dd6
|
|
BLAKE2b-256 checksum How to use checksums |
10f277752fb60e0405aa2f1f01b49d4ff8a18a42c01b56d2bb2e48cfbc398438
|
| 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 Oct 16, 2026.
Transparency logRelease files / rslab-1.5.0-cp39-abi3-manylinux_2_28_x86_64.whl
| Download URL | rslab-1.5.0-cp39-abi3-manylinux_2_28_x86_64.whl |
|---|---|
| Size | 3.5 MB |
| Tags | CPython 3.9 Linux glibc 2.28+ x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
d7475cf1017c45de0da396a539e363b630282c909e4ea219bf41b1672a3ba7bf
|
|
BLAKE2b-256 checksum How to use checksums |
9fbcc47d8d8ef1af5c7897a614bcea863fd83d7341fb5f94d1529ce7ed4113f7
|
| 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 Oct 16, 2026.
Transparency logRelease files / rslab-1.5.0-cp39-abi3-manylinux_2_28_aarch64.whl
| Download URL | rslab-1.5.0-cp39-abi3-manylinux_2_28_aarch64.whl |
|---|---|
| Size | 3.1 MB |
| Tags | CPython 3.9 Linux glibc 2.28+ ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
28c06dff5bb05ddb2cc7aab08b1ed040f63156efd81cc3cef755ae56d37abde8
|
|
BLAKE2b-256 checksum How to use checksums |
abaef25647929949005fdfdee95ac461a793083cf77e81df4f60fc17a3f8d76b
|
| 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 Oct 16, 2026.
Transparency logRelease files / rslab-1.5.0-cp39-abi3-macosx_11_0_arm64.whl
| Download URL | rslab-1.5.0-cp39-abi3-macosx_11_0_arm64.whl |
|---|---|
| Size | 2.7 MB |
| Tags | CPython 3.9 abi3 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
a9d3b7ac7b5efda22bc109b4c602f40622e893dac3d2793a1745021ad04b266d
|
|
BLAKE2b-256 checksum How to use checksums |
e70a8e9da8952be3073f94ac52223ad8f7297b4d84c9e43ac5f2d2660c7e8756
|
| 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 Oct 16, 2026.
Transparency logRelease files / rslab-1.5.0-cp39-abi3-macosx_10_12_x86_64.whl
| Download URL | rslab-1.5.0-cp39-abi3-macosx_10_12_x86_64.whl |
|---|---|
| Size | 3.2 MB |
| Tags | CPython 3.9 abi3 macOS 10.12+ x86-64 |
|
SHA-256 checksum How to use checksums |
33a4768894f2b5936969da8fbf87ed24bbdd179e59b2aa7394eda308978b8e75
|
|
BLAKE2b-256 checksum How to use checksums |
7023235e04476459e2bb483d590faba6ad94c1fe428933c89287336390d542f8
|
| 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 Oct 16, 2026.
Transparency log
