Optimal Work Extraction and the Minimum Description Length Principle

Statistical Physics and Complexity Group meeting

Optimal Work Extraction and the Minimum Description Length Principle

  • Event time: 10:00am until 11:00am
  • Event date: 9th June 2021
  • Speaker: (The Abdus Salam International Centre for Theoretical Physics)
  • Location: Online - see email.

Event details

We establish a connection between optimal work extraction in stochastic thermodynamics and efficient universal data compression, by showing that the maximal work that can be extracted in an information engine that relies on the measurement of a quantity X equals the minimum description length complexity of the statistical model that has X as a sufficient statistics. This provides design principles for optimal information engines. In particular, it suggests that optimal coding is thermodynamically efficient, and that it is essential to drive the system into a non-equilibrium critical state in order to achieve optimal performance (see arXiv:2006.04544 or Léo Touzo et al J. Stat. Mech. (2020) 093403)