Skip to content
/ AI-Hilbert Public

AI Hilbert is an algebraic geometric based discovery system (based on Putinar's Positivstellensatz), that enables the discovery of fundamental laws of nature (or non-physical systems) based on knowledge (articulated in formal logic terms) and experimental data.

License

Notifications You must be signed in to change notification settings

IBM/AI-Hilbert

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

51 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GitHub tag DOI License: MIT

AI-Hilbert

This repository contains the code and the data used for the experiments in the paper Evolving Scientific Discovery by Unifying Data and Background Knowledge with AI Hilbert.

AI Hilbert is an algebraic geometric based discovery system (based on Putinar's Positivstellensatz), that enables the discovery of fundamental laws of nature (or non-physical systems) based on knowledge (articulated in formal logic terms) and experimental data.

Visit our website for a general overview, references, and some introductory videos: → AI-Hilbert website

system overview

Code

The code is organized in 3 folders. One containing the problems studied in the main paper, one containing the problems studied in the paper appendix and one for data generation.

The folder main_problems contains:

The folder suppl_material_problems contains a notebook for each of the following problems:

The folder data_generation contains:

  • A python file for generating the data for the Kepler problem data_gen_kepler.py
  • A python file data_gen_mix.py for generating the data for the following problems:
    • Inelastic Relativistic Collision
    • Decay of Pion into Muon and Neutrino
    • Radiation Damping and Light Scattering
    • Escape Velocity
    • Hall Effect
    • Compton Scattering
    • Radiation Gravitational Wave Power
    • Hagen Poiseuille equation

How to cite

@misc{AI_Hilbert_2024,
      title={Evolving Scientific Discovery by Unifying Data and Background Knowledge with AI Hilbert}, 
      author={Ryan Cory-Wright and Cristina Cornelio and Sanjeeb Dash and Bachir El Khadir and Lior Horesh},
      year={2024},
      eprint={2308.09474},
      archivePrefix={arXiv},
      primaryClass={cs.AI}
}

About

AI Hilbert is an algebraic geometric based discovery system (based on Putinar's Positivstellensatz), that enables the discovery of fundamental laws of nature (or non-physical systems) based on knowledge (articulated in formal logic terms) and experimental data.

Resources

License

Stars

Watchers

Forks

Packages

No packages published