Monochromatic Arithmetic Progressions: A Machine Learning Approach
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of the Witwatersrand, Johannesburg
Abstract
Arithmetic progressions, fundamental in both number theory and combinatorics, have consistently fascinated mathematicians. This thesis explores the application of machine learning models to predict the fraction of monochromatic arithmetic progressions (APs) in randomly colored integer sequences, demonstrating the potential of Artificial Intelli- gence (AI) to contribute to the field of pure mathematics, particularly extremal combi- natorics. Specifically, we investigated the use of Linear Regression, Random Forest, and Multi-Layer Perceptron models to predict these fractions for varying AP lengths. We examined two distinct approaches: training individual models for specific AP lengths, and training combined models that learn across multiple lengths to enable generaliza- tion. Our results demonstrated the capability of machine learning to capture complex relationships within colored integer sequences. While models trained on individual AP lengths effectively predicted fractions for those specific lengths, the Multi-Layer Per- ceptron trained on combined lengths successfully generalized to unseen AP lengths, outperforming simpler models and showcasing its ability to capture underlying pat- terns. This work strengthened the intersection of AI methods with problems in pure mathematics, particularly extremal combinatorics, and laid the groundwork for future investigations.
Description
A research report submitted in fulfillment of the requirements for the Master of Science, in the Faculty of Science, School of Computer Science and Applied Mathematics, University of the Witwatersrand, Johannesburg, 2025
Citation
Malatsi, Tebogo. (2025). Monochromatic Arithmetic Progressions: A Machine Learning Approach [Master’s dissertation, University of the Witwatersrand, Johannesburg]. WIReDSpace. https://hdl.handle.net/10539/50273