Artificial intelligence and machine learning generated conjectures with TxGraffiti

Randy Davila

2024-07-03

Abstract:\emph{TxGraffiti} is a machine learning and heuristic based artificial intelligence designed to automate the task of conjecturing in mathematics. Since its inception, TxGraffiti has generated many surprising conjectures leading to publication in respectable mathematical journals. In this paper we outline the machine learning and heuristic techniques implemented by TxGraffiti. We also recall its contributions to the mathematical literature and announce a new online version of the program available for anyone curious to explore conjectures in graph theory.

Artificial Intelligence,Combinatorics

What problem does this paper attempt to address?

The paper attempts to address the problem of how to automatically generate mathematical conjectures using artificial intelligence and machine learning techniques. Specifically, the paper introduces a program called TxGraffiti, which automates the process of generating mathematical conjectures through machine learning and heuristic methods. TxGraffiti has already generated many surprising conjectures and has been published in some respected mathematical journals. The paper details the implementation of TxGraffiti, its contributions to mathematical literature, and the newly launched online version for those interested in exploring graph theory conjectures.

### Main Issues
1. **Automated Generation of Mathematical Conjectures**:
- How to design and implement a computer program that can automatically generate meaningful mathematical conjectures.
- How to ensure that the generated conjectures are both meaningful and of research value.

2. **Technical Implementation**:
- Using machine learning and data-driven methods to generate reasonable mathematical inequality conjectures.
- Finding optimal parameters through linear optimization, generating conjectures in the form \(\alpha(G) \leq m \cdot i(G) + b\), where \(\alpha(G)\) is the independence number and \(i(G)\) is another graph invariant.
- Implementing two heuristic methods (Theo and Dalmation-static) to filter and rank the generated conjectures, ensuring that the final presented conjectures are of high quality and novelty.

3. **Historical Background and Related Work**:
- Reviewing early automated conjecture generation systems, such as Fajtlowicz's GRAFFITI program and its subsequent versions Graffiti.pc.
- Discussing other related automated conjecture generation programs, such as Lenat's AM, Epstein's GT, Colton's HR, Hansen and Caporossi's AGX, and Mélot's Graphedron.

4. **Applications and Results**:
- Showcasing some specific conjectures generated by TxGraffiti and their applications in mathematical research, such as the relationship between the independence number and the matching number of regular graphs.
- Demonstrating the practical significance and potential value of these conjectures through specific examples and theorem proofs.

### Conclusion
The paper demonstrates the effectiveness and potential of TxGraffiti in generating mathematical conjectures, not only producing many meaningful conjectures but also promoting the development of mathematical research. Future work will further develop and refine TxGraffiti to play a greater role in computer-aided mathematical research.