Here I collect and organise the code developed as part of my research.
Each subdirectory is associated one-to-one to a specific work and it includes the simulation scripts, postprocessing routines and plotting utilities for reproducibility.
In each subdirectory a README file explains the main content of the project and provides a description of each experiment in exp/.
If you are interested in one particular work you can follow one of the links listed below.
What follows is a list of repositories associated to peer-reviewed work that has been Published in a scientific journal.
- Quadrature of functions with endpoint singular and generalised polynomial behaviour in computational physics (2024, external repository).
- The neural network shifted-proper orthogonal decomposition: a machine learning approach for non-linear reduction of hyperbolic equations (2022, external repository).
Submitted projects are works whose manuscripts have been submitted to a journal and are under peer-review. Once accepted they are promoted" to Published repositories.
In progress gives a list of repositories whose work is already been typed in a manuscript and close to submission.
- Dynamics, bifurcations and extensions of models of unified growth (Late 2026, estimated).
- Optimal, skilfull and interpretable prediction of high-dimensional tipping events (Late 2026, estimated).
- Statistical and numerical analysis of an early-warning signal approximation method based on large deviaton theory (Early 2027, estimated).
In Preliminary investigation we find projects that are at the primordial stage of development where test cases and theoretical advancement are being researched. These works, while could be typed in a manuscript, are nowhere near the stage of completion to become publishable material and thus cannot be listed as In progress.
- Localised patterns of the Lugiato-Lefever equation in 2 dimensions.
- Rate-induced tipping in a feedback-controlled, game-theoretic model.
- Characterisation of rate-induced tipping on parameteric families of parameter shifts.
- Nonlinear neural reduced manifold reconstruction of parametrised hyperbolic PDEs.
- Just a bunch of uncategorised tests.
A fair question to ask.
I strongly believe that scientific progress should be as transparent as possible. This, in my opinion, not only comes in the form of open access publications but also in the data that we collect and in the software that we use.
Now more than ever before, the scientific community relies on computational resources, simulations and high-performance scientific computing, not only to validate their hypothesis, but to formulate their theories in an increasingly stronger fashion. It is astonishing to me that, despite this, many (many, many, many, many...) peer-reviewed papers that heavily use these tools do not provide nor share the code they used in the production of the manuscript (e.g. in figures).
Not only this is hugely disripectful to developers (and sometimes fellow academics) who invested a considerable amount of time and effort to build these amazing tools we all use (in one way or another) to carry out our scientific investigations, but it is also hindering the process of actual peer-review (the one that happens AFTER the paper gets published) while contributing to the ongoing reproducibility crisis in science.
I thus think I should not be alone in making these sort of repositories, especially in an age in which text-based large-language models trivialized a lot of technicalities away in setting up systems like this one.
So if you are reading this, you find this repository useful, you are a fellow member of the scientific community AND you haven't yet created your own, please consider doing so. You will be providing a great service to the rest of us.