Scientific AI for Physics Workshop

Contribution List

7. Manuel Szewc (Universidad de San Martin): Machine learning hadronization

04/11/2025, 16:30

1. Luidhy Santana-Silva (CBPF): From Pixels to Classes: Deep Learning Classification of 30 Million Galaxies in the DECam Local Volume Exploration Survey

05/11/2025, 09:50

3. Gustavo Dalpian Universidade de São Paulo): Learning from machine learning: discovering new science for materials

05/11/2025, 11:05

4. Pedro da Costa Huot: Study of hypernuclei identification in ultra-relativistic heavy-ion collisions using a machine learning approach

05/11/2025, 11:55

8. Luis Itza Vazquez-Salazar: Molecular Simulations with/without ML

05/11/2025, 16:30

13. Phelipe Antonie Darc de Matos (CBPF): Symbolic Regression Is All You Need: From Simulations to Scaling Laws in Binary Neutron Star Mergers

06/11/2025, 09:50

15. André Sznajder (UERJ): Ultrafast Jet Classification for the HL-LHC

06/11/2025, 11:05

16. Joaquin Armijo (Universidade de São Paulo): DeepLensingFlow: scored and flow-based networks for Weak-lensing map statistics

06/11/2025, 11:55

18. Luis Itza Vazquez-Salazar: Graph Diffusion Models

06/11/2025, 16:40

19. Daniel López-Cano (universidade de São Paulo): Semi-Supervised Domain Adaptation for Sim-to-Obs Astrophysics: DESI→J-PAS

07/11/2025, 09:50

21. Raphael Cóbe (UNESP): From Research to Impact: AI2's Model for Digital Social Innovation

07/11/2025, 11:05

22. Rafael Bicudo Ribeiro (Universidade de São Paulo): Physically-oriented clustering of conformations in multi-molecule systems

07/11/2025, 11:55

27. DeepLensingFlow: scored and flow-based networks for Weak-lensing map statistics

Joaquin Armijo (IFUSP)

Wide-field astronomical surveys provide unprecedented data that allow us to reconstruct the gravitational lensing maps tracing the large-scale distribution of matter in the Universe. In the weak lensing regime, these maps serve as a powerful probe of the Lambda-CDM cosmological model. However, their high dimensionality (millions of correlated pixels) poses significant challenges for...

28. emi-Supervised Domain Adaptation for Sim-to-Obs Astrophysics: DESI→J-PAS

Daniel López-Cano (Instituto de Física da Universidade de São Paulo (IFUSP))

Modern ML models trained on simulations often degrade on real data because of domain shift. I will present a semi-supervised domain adaptation (SSDA) pipeline that transfers a four-class pseudo-spectral classifier (high-z QSOs, low-z QSOs, galaxies, stars) from abundant DESI→J-PAS mocks (~1.5M) to real J-PAS observations using only a small labeled J-PAS subset. The method pretrains on mocks,...

24. From Pixels to Classes: Deep Learning Classification of 30 Million Galaxies in the DECam Local Volume Exploration Survey

Luidhy Santana-Silva (Centro Brasileiro de Pesquisas Fisicas)

Understanding the formation and evolution of galaxies over cosmic time requires a
comprehensive analysis of their morphologies, especially because morphological
features are strongly connected to other galaxy properties such as stellar
populations, environments, and kinematics. However, the growing size of modern
sky surveys has resulted in massive volumes of unclassified galaxies,...

29. Hadronization and Machine Learning

Manuel Szewc

Lectures 1/3

Hadronization, the transition from unobservable partons to measurable hadrons, is a key component of how the Standard Model of particle physics explains current data. However, due to its intrinsically non-perturbative nature, it remains challenging to model from first principles. In particle physics, where simulators are needed to relate theory and collider experiments, Monte Carlo event...

6. Lectures

30. Physically-oriented clustering of conformations in multi-molecule systems

Rafael Ribeiro (Universidade de São Paulo)

Unsupervised machine learning techniques are widely used to analyze the extensive data generated by molecular modeling. In particular, some tools have been developed to cluster configurations from classical simulations with a standard focus on individual units, ranging from small molecules to complex proteins. Since the standard approach computes the root-mean-square deviation (RMSD) of atomic...

26. Study of hypernuclei identification in ultra-relativistic heavy-ion collisions using a machine learning approach

Pedro Da Costa Huot (Universidade de Sao Paulo (USP) (BR))

We present a study on the application of different machine learning algorithms for the identification of hypernuclei produced in heavy-ion collisions, particularly those with mass numbers A = 3 to A = 5. The study focuses on three supervised learning algorithms - Boosted Decision Trees, Support Vector Machines, and Artificial Neural Networks - which were trained to distinguish true hypernuclei...

25. Symbolic Regression Is All You Need: From Simulations to Scaling Laws in Binary Neutron Star Mergers

Phelipe Antonie Darc De Matos (Centro Brasileiro De Pesquisas Físicas)

Gravitational wave sources with electromagnetic counterparts have highlighted the need for predictive, interpretable models linking the parameters of compact binary systems to post-merger remnants and mass outflows. In this work, we explore AI-driven symbolic regression (SR) frameworks to derive updated analytical relations for disk ejecta mass in binary neutron star mergers, trained on...

2. TBA

5. TBA

9. TBA

10. TBA

11. TBA

12. TBA

14. TBA

17. TBA

20. TBA

23. TBA