pysr
Simple and efficient symbolic regression
Decision gist · record as of 2026-08-14
Yes, if you work with low-dimensional data where interpretability is a priority or you need to extract equations from neural networks. The active maintenance, permissive license, low install friction, and scikit-learn-compatible interface make it a solid choice. Not recommended if you need black-box performance on high-dimensional datasets or require guaranteed convergence to a global optimum.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Julia to be installed or will download it on first import; LD_LIBRARY_PATH conflicts with other Python packages loading libstdc++ may require manual configuration.
- Low friction install with a pure Python wheel; Julia dependencies install automatically on first import.
- Actively maintained with recent releases.
License · maintenance · safety
permissive license (permissive) — Apache 2.0 permissive license allows commercial and private use with minimal restrictions; you may modify and distribute derivative works provided you include license notices.
last release 2026-03-30 (137 days)
0 known vulnerabilities (OSV.dev, 2026-08-14) · 98,850 downloads/mo, #13,056 on PyPI
Alternatives
Verify before relying
pip install pysr
import numpy as np
from pysr import PySRRegressor
X = np.random.randn(100, 5)
y = 2.5 * np.cos(X[:, 3]) + X[:, 0] ** 2
model = PySRRegressor(maxsize=20, niterations=40, binary_operators=["+", "*"], unary_operators=["cos", "sin"])
model.fit(X, y)
print(model.predict(X))- Whether the package handles datasets larger than typical low-dimensional symbolic regression benchmarks without significant performance degradation.
- Specific performance characteristics or memory requirements for typical use cases.
- Whether custom loss functions and operators are fully documented and stable across versions.
What it is and what it does
PySR is a symbolic regression tool that discovers interpretable mathematical expressions fitting your data. Instead of returning a neural network or opaque model, it uses evolutionary search powered by Julia to explore the space of possible equations, combining operators like addition, multiplication, and functions like cosine or exponential. It follows scikit-learn's fit/predict interface, making it familiar to data scientists while offering something fundamentally different: equations you can read, understand, and use analytically.
The package is designed for low-dimensional datasets where interpretability matters more than squeezing out the last percentage point of accuracy. It also supports symbolic distillation—converting trained neural networks into analytic equations—which can help explain deep learning models. Configuration is extensive: you define which operators to search, custom loss functions, complexity penalties, and iteration counts, giving you fine control over the search process.
Use it for
- Discover interpretable formulas from experimental physics or chemistry data where understanding the equation matters as much as prediction.
- Convert a trained neural network into a symbolic equation for scientific interpretation or deployment in resource-constrained environments.
- Fit low-dimensional datasets where you suspect a simple closed-form relationship exists but don't know its structure.
- Generate candidate equations for hypothesis testing in scientific research where the model itself is the research output.
- Reduce model complexity and improve generalization by finding sparse, parsimonious expressions instead of high-degree polynomials.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you work with low-dimensional data where interpretability is a priority or you need to extract equations from neural networks.
The active maintenance, permissive license, low install friction, and scikit-learn-compatible interface make it a solid choice. Not recommended if you need black-box performance on high-dimensional datasets or require guaranteed convergence to a global optimum.
Install
pysr on PyPI
Before you install
Low friction install with a pure Python wheel; Julia dependencies install automatically on first import. Actively maintained with recent releases.
Requires Julia to be installed or will download it on first import; LD_LIBRARY_PATH conflicts with other Python packages loading libstdc++ may require manual configuration.
License in practice
Apache 2.0 permissive license allows commercial and private use with minimal restrictions; you may modify and distribute derivative works provided you include license notices.
Quickstart
pip install pysr
import numpy as np
from pysr import PySRRegressor
X = np.random.randn(100, 5)
y = 2.5 * np.cos(X[:, 3]) + X[:, 0] ** 2
model = PySRRegressor(maxsize=20, niterations=40, binary_operators=["+", "*"], unary_operators=["cos", "sin"])
model.fit(X, y)
print(model.predict(X))Verify before relying
- Whether the package handles datasets larger than typical low-dimensional symbolic regression benchmarks without significant performance degradation.
- Specific performance characteristics or memory requirements for typical use cases.
- Whether custom loss functions and operators are fully documented and stable across versions.
Package facts
| License | permissive license permissive |
| Python support | Supports the current Python release >=3.9 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 7 packagesclickjuliacallnumpypandasscikit-learnsympytyping-extensions |
| Maintenance | Actively maintained 137 days since the last release |
| First released | |
| Downloads | 98,850 / month, #13,056 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | License :: OSI Approved :: Apache Software LicenseOperating System :: OS IndependentProgramming Language :: Python :: 3 |
Evidence: pysr-1.5.10-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “symbolic regression”
- pysrPySR searches for symbolic expressions that fit data by combining…
- symenginePython wrappers for SymEngine, a fast C++ symbolic manipulation…
- sympySymPy is a Python library for symbolic mathematics, performing…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also optlang · sympy · py-expression-eval · math-verify · pyroots · latex2sympy2 · latex2sympy2-extended · lineax · optimistix · pytensor