We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat.stat-mech

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Condensed Matter > Statistical Mechanics

Title: Entropy production estimation with optimal current

Abstract: Entropy production characterizes the thermodynamic irreversibility and reflects the amount of heat dissipated into the environment and free energy lost in nonequilibrium systems. According to the thermodynamic uncertainty relation, we propose a deterministic method to estimate the entropy production from a single trajectory of system states. We explicitly and approximately compute an optimal current that yields the tightest lower bound using predetermined basis currents. Notably, the obtained tightest lower bound is intimately related to the multidimensional thermodynamic uncertainty relation. By proving the saturation of the thermodynamic uncertainty relation in the short-time limit, the exact estimate of the entropy production can be obtained for overdamped Langevin systems, irrespective of the underlying dynamics. For Markov jump processes, because the attainability of the thermodynamic uncertainty relation is not theoretically ensured, the proposed method provides the tightest lower bound for the entropy production. When entropy production is the optimal current, a more accurate estimate can be further obtained using the integral fluctuation theorem. We illustrate the proposed method using three systems: a four-state Markov chain, periodically driven particle, and multiple bead-spring model. The estimated results in all examples empirically verify the effectiveness and efficiency of the proposed method.
Comments: 10 pages, 4 figures
Subjects: Statistical Mechanics (cond-mat.stat-mech)
Cite as: arXiv:2001.07131 [cond-mat.stat-mech]
  (or arXiv:2001.07131v1 [cond-mat.stat-mech] for this version)

Submission history

From: Tan Vu Van [view email]
[v1] Mon, 20 Jan 2020 14:59:44 GMT (266kb,D)

Link back to: arXiv, form interface, contact.