OptiProfiler#

Benchmark your optimization solver

OptiProfiler standardizes solver benchmarking with curated problem libraries, shared interfaces, baseline comparisons, and reproducible performance profiles.

Get started View on GitHub

quick_start.py

from scipy.optimize import minimize
from optiprofiler import benchmark

def bfgs(fun, x0):
    return minimize(
        fun, x0, method="BFGS").x

def powell(fun, x0):
    return minimize(
        fun, x0, method="Powell").x

scores = benchmark([bfgs, powell])
print(scores)

Python + MATLAB Benchmark solvers through either interface with the same workflow

Problem libraries Use CUTEst, S2MPJ, custom problems, and diverse features

Reproducible outputs Export profiles, scores, logs, and result files for reporting

Two ways to use OptiProfiler

Run it locally, or in the cloud

Run locally

Use Python & MATLAB packages

Install the open-source packages, run larger experiments, plug in your own problem libraries, and keep full control over private problems and compute.

Read the installation guide →

Run in the cloud

Try the hosted platform

Upload a Python solver, choose benchmark settings, compare against baseline solvers, and get publication-ready profiles with downloadable results in your browser.

Open app.optprof.com →

Why OptiProfiler?

Simple usage for beginners

Easy installation and quick start with a few lines of code

Multiple degrees of freedom for experts

Multiple built-in features and customization options for test suites

Automatic generation of high-quality profiles

Publication-ready PDF visualizations with precise, clear, and aesthetically pleasing figures

Reliable methodology for benchmarking

Based on the widely accepted performance profile and data profile

Trackable experimental results for reproducibility

Seed-controlled and time-stamped experimental results for easy reproducibility

Multi-language support

Python

The Python interface is ready to use. Check out the installation guide to get started, explore examples for common workflows, or dive into the API reference for details.

MATLAB

The full MATLAB interface is ready to use. Check out the installation guide to get started, explore examples for common workflows, or dive into the API reference for details.