Statistical mechanics of complex systems for pattern identification

Shalabh Gupta, Asok Ray

Research output: Contribution to journalArticlepeer-review

27 Scopus citations

Abstract

This paper presents a statistical mechanics concept for identification of behavioral patterns in complex systems based on measurements (e.g., time series data) of macroscopically observable parameters and their operational characteristics. The tools of statistical mechanics, which provide a link between the microscopic (i.e., detailed) and macroscopic (i.e., aggregated) properties of a complex system are used to capture the emerging information and to identify the quasi-stationary evolution of behavioral patterns. The underlying theory is built upon thermodynamic formalism of symbol sequences in the setting of a generalized Ising model (GIM) of lattice-spin systems. In this context, transfer matrix analysis facilitates construction of pattern vectors from observed sequences. The proposed concept is experimentally validated on a richly instrumented laboratory apparatus that is operated under oscillating load for identification of evolving microstructural changes in polycrystalline alloys.

Original languageEnglish (US)
Pages (from-to)337-364
Number of pages28
JournalJournal of Statistical Physics
Volume134
Issue number2
DOIs
StatePublished - Jan 1 2009

All Science Journal Classification (ASJC) codes

  • Statistical and Nonlinear Physics
  • Mathematical Physics

Fingerprint Dive into the research topics of 'Statistical mechanics of complex systems for pattern identification'. Together they form a unique fingerprint.

Cite this