Anisa Khatun (2026.07.01-10.30)

Abstract: Rare and exotic heavy-flavour hadrons constitute one of the key physics goals of the future ALICE 3 experiment at CERN. Among these, multi-charm baryons such as \(\Xi\)cc++, \(\Omega\)cc and \(\Omega\)ccc provides unique probes of Quantum Chromodynamics (QCD), charm-quark hadronisation, and the Quark–Gluon Plasma. However, their extremely small production cross sections make Monte Carlo (MC) simulation a major computational bottleneck, as very large simulated samples are required for detector-performance studies and machine-learning (ML)-based analyses.

The proposed project aims to investigate Generative Adversarial Networks (GANs) as a reconstructed-level data augmentation technique for rare heavy-flavour analyses. Building upon an existing GAN framework developed for heavy-flavour reconstruction in ALICE Run 3, this project will extend the methodology to simulated multi-charm baryons using publicly available ALICE 3 simulation datasets. The generated samples will be validated through multidimensional feature comparisons, correlation studies, and ML-based compatibility tests.

The expected outcome is an open, GPU-enabled AI workflow together with a methodological publication demonstrating the feasibility of reconstructed-level generative data augmentation for extremely rare heavy-flavour signals.