BEGIN:VCALENDAR VERSION:2.0 PRODID:-//linuxsoftware.nz//NONSGML Joyous v1.4//EN BEGIN:VEVENT SUMMARY:Descriptive and Predictive Modeling of Chaos in Cold Atom Physics using Explainable Deep Learning DTSTART:20251204T110000Z DTEND:20251204T130000Z DTSTAMP:20260615T075139Z UID:1e497eec-6db2-4a3c-b719-757b3d68f3e4 SEQUENCE:1 CREATED:20251128T094606Z DESCRIPTION: Ultracold atom clouds display intricate dynamics that can app ear chaotic and lack a comprehensive explanatory framework. This dissertat ion investigates photon bubble turbulence in a 85Rb atom cloud confined in a magneto-optical trap\, using recent advances in artificial intelligence to move beyond the limitations of classical modeling. Reduced-order metho ds and sparse regression provide interpretability but fail to capture long -term dynamics or generalize across regimes. To overcome these issues\, de ep learning models were developed\, with a focus on convolutional autoenco ders for reconstructing and analyzing atom cloud images. Results showed th at these architectures consistently outperformed classical baselines\, eve n in lowdimensional latent spaces\, demonstrating the ability of convoluti onal representations to extract physically meaningful features. Latent spa ce analysis revealed that the apparently chaotic turbulence is underpinned by compact structures\, suggesting a simple order to chaos. Classificatio n experiments showed that detuning values\, determined by laser frequencie s in the trap\, can be reliably inferred from images\, exposing discrimina tive features tied to light–atom interactions. Explainability was introd uced through Grad-CAM\, linking classification decisions to density struct ures in the clouds. The findings confirmed two dynamical regimes\, stable and turbulent\, separated by a phase transition\, with further evidence of a transitional regime near resonance. Overall\, explainable deep learning emerges as a powerful framework for predictive and descriptive modeling o f ultracold atom systems\, while also deepening our understanding of turbu lence\, phase transitions\, and light–atom coupling. LAST-MODIFIED:20251128T094606Z LOCATION:Sala V1.06 Pavilhão de Civil URL:http://df.vps.tecnico.ulisboa.pt/en/events/descriptive-and-predictive- modeling-of-chaos-in-cold-atom-physics-using-explainable-deep-learning/ X-ALT-DESC;FMTTYPE=text/html:

Ultracold atom clo uds display intricate dynamics that can appear chaotic and lack a comprehe nsive explanatory framework. This dissertation investigates photon bubble turbulence in a 85Rb atom cloud confined in a magneto-optical trap\, using recent advances in artificial intelligence to move beyond the limitations of classical modeling. Reduced-order methods and sparse regression provid e interpretability but fail to capture long-term dynamics or generalize ac ross regimes.

To overcome these issues\, deep learning models we re developed\, with a focus on convolutional autoencoders for reconstructi ng and analyzing atom cloud images. Results showed that these architecture s consistently outperformed classical baselines\, even in lowdimensional l atent spaces\, demonstrating the ability of convolutional representations to extract physically meaningful features. Latent space analysis revealed that the apparently chaotic turbulence is underpinned by compact structure s\, suggesting a simple order to chaos.

Classification experimen ts showed that detuning values\, determined by laser frequencies in the tr ap\, can be reliably inferred from images\, exposing discriminative featur es tied to light–atom interactions. Explainability was introduced throug h Grad-CAM\, linking classification decisions to density structures in the clouds. The findings confirmed two dynamical regimes\, stable and turbule nt\, separated by a phase transition\, with further evidence of a transiti onal regime near resonance. Overall\, explainable deep learning emerges as a powerful framework for predictive and descriptive modeling of ultracold atom systems\, while also deepening our understanding of turbulence\, pha se transitions\, and light–atom coupling.

END:VEVENT END:VCALENDAR