Stochastic thermodynamics consists of a family of formalisms able to describe non-equilibrium processes in a general way and from a thermodynamic standpoint. It allows to obtain powerful results of universal nature, and has been employed in recent years to a wide variety of problems in different areas. I will introduce the basic concepts of the field and review recent developments. In particular, I will discuss a recent extension of the Second Law of thermodynamics to non-equilibrium steady states (Freitas and Esposito, Nat. Commun. 13, 5084 (2022)). I will also discuss the application of stochastic thermodynamics to non-linear electronic circuits (Freitas et. al. Phys. Rev. X 11, 031064 (2021)), and population dynamics models
Both the social and natural world are replete with complex structure that often has a probabilistic interpretation. In the former, we may seek to model, for example, the distribution of natural images or language, for which there are copious amounts of real world data. In the latter, we are given the probabilistic rule describing a physical process, but no procedure for generating samples under it necessary to perform simulation. In this talk, I will discuss a generative modeling paradigm based on maps between probability distributions that is applicable to both of these circumstances. I will describe a means for learning these maps in the context of problems in statistical physics, how to impose symmetries on them to facilitate learning, and how to use the resultant generative models in a statistically unbiased fashion. I will then describe a paradigm that unifies flow-based and diffusion based generative models by recasting generative modeling as a problem of regression. I will demonstrate the efficacy of doing this in computer vision problems and end with some future challenges and applications.
Despite eight years since the initial detection of gravitational waves, the astrophysical origin of these phenomena remains elusive. Recent years have witnessed a growing interest in a novel gravitational wave formation pathway: the active galactic nuclei (AGN) channel. I will begin by providing an overview of the current status of gravitational wave detections, the main astrophysical mechanisms driving the pairing and coalescence of black holes, and the observational signatures crucial for distinguishing between various formation scenarios. I will then describe the key features of the AGN channel, discuss our ongoing efforts in modeling compact objects within accretion disks in AGNs, and highlight the primary challenges associated with modeling black hole-gas disk interactions. Partially based on: arXiv:2403.00060, arXiv:2312.13281
This seminar presents ongoing research on the qualitative dynamics of reaction-diffusion processes, cornerstone complex systems in physics and biology. These equations, fundamental to modeling phenomena ranging from chemical reactions to ecological and cellular processes, encapsulate the intricate balance between transport mechanisms and local interactions. A central aim in the study of such systems is to decipher the macroscopic or qualitative behaviors that emerge from these complex interactions, seeking to understand how patterns, waves, and structures develop on larger scales. However, despite the valuable analytical insights offered by perturbation or renormalization techniques, these approaches can struggle in highly nonlinear or multi-scale regimes and do not easily generalize to new parameters. Addressing these challenges, this work leverages recent advancements in data-driven dynamical systems theory to uncover the low-dimensional dynamics governing macroscopic features of interest. By employing machine learning techniques to derive low-dimensional representations, this approach clarifies the emergence of qualitative structures which are often obscured in the high-dimensional data of the original systems. This method not only facilitates a deeper understanding of the system's dynamics but also opens new avenues for control and parameter identification. Preliminary results demonstrate the efficacy of this methodology in shedding light on the behavior of the Gray Scott model and Min protein dynamics, both examples of reaction-diffusion systems with significant theoretical and biological implications. The seminar will detail the theoretical underpinnings, methodological developments, and early outcomes of this research, highlighting its potential to advance our understanding of reaction-diffusion systems as they arise in biology