Newclid

Newclid is an open-source, easy-to-use fast solver for plane geometry problems.

Installation

Using uv (recommended)

Follow uv installation instructions

uv add newclid[yuclid]

Using pip

pip install newclid[yuclid]

Building from source

Follow uv installation instructions

git clone https://github.com/Newclid/Newclid.git
cd Newclid
uv sync

If you run into issues to build yuclid, you might need to set an environment variable like CXX=/usr/bin/g++-14, try adding environment variable that are present in .env. If you still have issues, submit an issue.

Quickstart

To simply solve a problem using Newclid, use the command line.

For example with a JGEX problem:

newclid jgex --problem-id orthocenter_consequence_aux --file ./problems_datasets/examples.txt

Or with a ggb problem:

newclid ggb --file ./notebooks/ggb_exports/incenter.ggb --goals "eqangle C B B D B D B A"

See other command line interface options with:

newclid --help
newclid jgex --help
newclid ggb --help

For more complex applications, use the Python interface. Below is a minimal example to build a problem setup from a JGEX string, then solve it:

from newclid import GeometricSolverBuilder, GeometricSolver
import numpy as np

# Set the random generator
rng = np.random.default_rng()

# Build the problem setup from JGEX string
problem_setup = JGEXProblemBuilder(rng=rng).with_problem_from_txt(
  "a b c = triangle a b c; "
  "d = on_tline d b a c, on_tline d c a b; "
  "e = on_line e a c, on_line e b d "
  "? perp a d b c"
).build()

# We now build the solver on the problem
solver: GeometricSolver = GeometricSolverBuilder().build(problem_setup)

# And run the solver
success = solver.run()

if success:
    print("Successfuly solved the problem! Proof:")
    solver.write_proof_steps()
else:
    print("Failed to solve the problem...")

print(f"Run infos {solver.run_infos}")

In the notebooks folder you will find more tutorials, for example:

imo_2025.ipynb to run and solve a problem from the IMO 2025 P2 problem and build the animated proof.

geogebra_problem.ipynb to run and solve a problem from a ggb file.

jgex_problem.ipynb to run and solve a problem from a JGEX string.

multiple_JGEX_problems.ipynb to run and solve problems from a file with multiple JGEX problems, one at a time or in bulk.

heuristics_implementation.ipynb to run a collection of problems and try to solve them using human-made heuristics to add auxiliary points to a problem.

You can also check tests to see some more advanced examples of scripts using the Python interface.

Documentation

See the online documentation for more detailed information about Newclid.

Contributing

Clone the repository

git clone https://github.com/Newclid/Newclid.git
cd Newclid

Install uv

Follow installation instructions

Install as an editable package with dev requirements

uv sync

Install pre-commit and pre-push checks

pre-commit install -t pre-commit -t pre-push

Run tests

pytest tests

About Newclid

Newclid is a successor to AlphaGeometry, introduced in this early 2024 Nature paper: Solving Olympiad Geometry without Human Demonstrations. whose original codebase can be found here.

If you found Newclid useful, please cite us as:

@article{newclid2024sicca,
  author  = {Sicca, Vladmir and Xia, Tianxiang and F\'ed\'erico, Math\"is and Gorinski, Philip John and Frieder, Simon and Jui, Shangling},
  journal = {arXiv preprint},
  title   = {Newclid: A User-Friendly Replacement for AlphaGeometry with Agentic Support},
  year    = {2024}
}

The AlphaGeometry checkpoints and vocabulary are made available under the terms of the Creative Commons Attribution 4.0 International (CC BY 4.0) license. You can find details at: https://creativecommons.org/licenses/by/4.0/legalcode