Skip to main content

Functionally pure definitions of optimisation problems extracted from Standard Input Format (SIF).

Project description

sif2jax

Functionally pure definitions of optimisation problems extracted from Standard Input Format (SIF), written in JAX.

This is for you if you write optimisation software in JAX (or Python) and want to stress-test it on the CUTEst set of benchmark problems. Features include

  • all JAX everything: no Fortran backends
  • full support for autodiff, batching, and JIT compilation
  • more JAX benefits: run on CPU/GPU/TPU
  • clear and human-readable problem definitions, no decoder required
  • lean API - no specific problem interface required

Installation

pip install sif2jax

Requires TODO fix

Getting started

We recommend running the benchmarks with pytest-benchmark - use the familiar testing infrastructure to run your benchmarks:

import sif2jax

benchmark_problems = sif2jax.bounded_minimisation_problems

@pytest.mark.benchmark
@pytest.mark.parametrize("problem", sif2jax.unconstrained_minimisation_problems)
def test_lbfgs(benchmark, problem):
    ...

Alternatively, you can run any arbitrary benchmark problem by passing an index, or directly import a problem by name

import sif2jax

problem = sif2jax.problems[42]
another_problem = sif2jax.cutest.get_problem("ROSENBR")

The problems all have the following methods:

  • objective - a callable with signature f(y, args), where y is the optimisation variable
  • y0 - returns the initial guess provided by the SIF file
  • args - returns any arguments (frequently none)

bounded problems also have a bounds method, and constrained problems additionally include a constraint method. You can find more information in our documentation TODO Fix.

cyipopt installation

I recommend using the conda install as it packages all necessary binaries. I can add this to the dockerimage later.

conda install -c conda-forge cyipopt

CUTEst problems in SIF format

To make the CUTEst problems in SIF format accessible to the LLM:

mkdir archive  # This name is in the .gitignore
cd archive
git clone https://bitbucket.org/optrove/sif ./mastsif

Suggestion: change the permissions to that folder to remove write access.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sif2jax-0.0.4.tar.gz (236.6 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

sif2jax-0.0.4-py3-none-any.whl (568.6 kB view details)

Uploaded Python 3

File details

Details for the file sif2jax-0.0.4.tar.gz.

File metadata

  • Download URL: sif2jax-0.0.4.tar.gz
  • Upload date:
  • Size: 236.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.3

File hashes

Hashes for sif2jax-0.0.4.tar.gz
Algorithm Hash digest
SHA256 888ab187b24d4da59e09b9823bd3e9ef3ba3043412340d736ac2ba306c0bc024
MD5 8b7320278ec2152b124dfdd14b15f0b1
BLAKE2b-256 260e692dd3dbb77df63cf468e9b873c5abe90b983dfb5e70034abbb87c5e3407

See more details on using hashes here.

File details

Details for the file sif2jax-0.0.4-py3-none-any.whl.

File metadata

  • Download URL: sif2jax-0.0.4-py3-none-any.whl
  • Upload date:
  • Size: 568.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.3

File hashes

Hashes for sif2jax-0.0.4-py3-none-any.whl
Algorithm Hash digest
SHA256 fe70575ceee642472bf5e70d2f36c0dabe87194e0566bf356cb0b1c1c46e5046
MD5 6aa047f077cbd269191a929fcfd45741
BLAKE2b-256 a04f5490c52b87aca927b5829384577057f84d5600a76bb2162ac8df39477fe6

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page