PolyChron: FAIR data for post-excavation analysis
This project focused on improving PolyChron, an archaeological dating software using MCMC methods. The work involved fixing bugs, refining the implementation of the Metropolis-Hastings algorithm, and improving reproducibility through better seeding. As the software is still under development, these changes contribute to making it more robust and reliable for future use.
What are the key results of your project?
The key results of the project were demonstrating that correct implementation of seeding makes the MCMC process reproducible and identifying how improper seed handling can lead to inconsistent or misleading outputs. By restructuring the seeding approach, the software now produces consistent results when the same seed is used, while still reflecting natural variability when different seeds are applied. Additionally, users can now choose between allowing the software to generate a random seed or manually entering one themselves.
GitHub Repository: https://github.com/bryonymoody/PolyChron
How do you feel you have benefitted from completing this internship?
This internship has been really valuable in developing both my technical and personal skills. I deepened my understanding of programming, particularly working with larger codebases where functions are spread across multiple files and need to interact correctly. The challenges I faced, especially around implementing seeding correctly, were some of the most complex coding problems I have worked on, and having to debug and refine those solutions really improved my ability to think critically and approach problems methodically.
Alongside this, the internship helped me improve my time management skills, as I was balancing this project alongside my university work during term time.
In terms of future career paths, I have always been interested in pursuing a career involving mathematics, and this experience has definitely reinforced that. It showed me how mathematical and computational methods can be applied to meaningful real world problems, which is something I would like to continue exploring.