Online (Email amgrad for the link)
Candidate
Adam Teixido-Bonfill | University of Waterloo
Title
Observational entropy as a framework to derive thermodynamics
Abstract
Thermodynamics is one of the most successful theories in physics, built originally from empirical observations. However, its laws are expected to emerge from the underlying rules of classical and quantum mechanics. A key concept to form this connection is entropy, which plays apparently different roles depending on the context: in thermodynamics, it signals the direction of spontaneous processes, and in information theory, it quantifies uncertainty. At the microscopic scale, these roles can be linked.
Observational entropy (arXiv:2008.04409) is a proposal that received renewed attention with the aim to make this link explicit. This proposal consists of defining entropy in terms of what an observer can actually measure, incorporating the idea of coarse-graining due to limited resolution. A recent generalization allows to account for known constraints on the state of the physical system (arXiv:2503.15612). This formulation unifies most previous entropy definitions under a common framework.
In this talk, I will introduce observational entropy and its generalizations, and outline how they offer a promising route to derive the laws of thermodynamics from first principles.