Learning, Information Theory, and Nonequilibrium Statistical Mechanics

 
 

Berkeley workgroup on

       Learning, Information Theory, & Nonequilibrium

       Thermodynamics

 

Coordinates

EMAIL  lineq (at) lists.berkeley.edu

LOCATION  560 Evans Halls, UC Berkeley

TIme  3:30 PM every other Friday (kinda)

WEB https://calmail.berkeley.edu/manage/list/listinfo/lineq@lists.berkeley.edu

Charles Frye (UC Berkeley): The Statistical Physics of Hype? A Paper Discussion

6 November 2015

Two papers by the England group at MIT have made waves and drawn attention even outside physics, as expected for papers with titles like "Statistical Physics of Self-Replication" [1] and "Statistical Physics of Adaptation" [2]. They have been met, however, with reasonable skepticism, in part precisely for the grandiosity of their claims. We'll be walking through the derivations of their "macroscopic Crooks relation" -- Eqn (8) in [1] -- and "expanded Helmholtz equation" -- Eqn (8) in [2] -- with an eye to critical evaluation of the assumptions, implicit and explicit. If time permits, we'll discuss applications to learning, which have been preliminarily presented in a talk format ([3], ~50 min).

References:

[1] Statistical physics of self-replication. Jeremy L. England, J. Chem. Phys. 139, 121923 (2013), URL: http://scitation.aip.org/content/aip/journal/jcp/139/12/10.1063/1.4818538.

[2] Statistical physics of adaptation. Nikolai Perunov, Robert Marsland, and Jeremy England, arXiV (2014), URL: http://arxiv.org/abs/1412.1875.

[3] What is life? Jeremy England, Talk at URL: https://www.youtube.com/watch?v=e91D5UAz-f4.