Astronomical scale
Topological descriptors for galaxy morphology, asteroseismology, and gravitational-wave signals, using tools such as persistent homology, Mapper, and graph-based representations.
Physics · Topological Data Analysis · Machine Learning
Gabriel Wendell Celestino Rocha
Physicist and PhD student at UFRN (Brazil) working on Topological Data Analysis and Topological Deep Learning applied to physical systems across scales – from magnetic materials and dynamical systems to galaxies, gravitational waves, and asteroseismology.
Currently open to international collaborations, research visits, and projects in TDA, topological machine learning, and physics-informed ML.
Also find me on ORCID, ResearchGate, Lattes, and Google Scholar.
Research overview
My work focuses on how topological and geometric methods can uncover structure in complex physical data, and how these structures can inform modeling, prediction, and inversion problems in physics.
Mesoscopic scale
Topology in magnetohydrodynamics, fluid instabilities, and dynamical systems, exploring how homological features relate to transport, mixing, and phase-space structure.
Microscopic scale
Topological analysis of Barkhausen noise, disordered media, and quantum information, linking topological signatures with domain-wall dynamics and entanglement structure.
Featured work
A small selection of ongoing and recent projects. For full details, see the research, publications, talks, and software pages.
TDA-SSL Galaxy Morphology
Self-supervised representations for Galaxy Zoo images combining persistent homology and deep neural networks to obtain robust, interpretable morphology descriptors.
TDA for GW detection at low SNR
Exploring topological summaries of time-series data for detecting gravitational-wave-like signals under low signal-to-noise ratios, within the context of the 4th School on Data Science & Machine Learning.
Topological diagnostics for CNN predictors
Using TDA to analyse magnetohydrodynamic shock tube simulations and understand the failure modes of neural temporal predictors via topological signatures.
News & highlights
A few recent activities and projects. For more details, see the publications, talks, and software pages.
- [2025] Developing a PhD project on Topological Data Analysis and Topological Deep Learning Applied to Physical Systems Across Scales.
- [2025] Project on TDA-based detection of gravitational-wave-like signals at low SNR for the 4th School on Data Science & Machine Learning.
- [2025] TDA diagnostics of CNN predictors applied to shock tube dynamics, with results presented at the XXXIX EFNNE meeting.
Contact
The best way to reach me is by email. I am always open to discussions and collaborations involving TDA/TML, astrophysics, complex systems, and scientific computing.