Determining Dynamical Path Distributions usingMaximum Relative Entropy

Abstract

The scientific objective was to extend MaxCal to include microscopic (data) information by deriving the explicit, general, theoretical distributions that includes both non-equilibrium, path trajectory information (as done in MaxCal) and direct, observational microscopic information. MaxCal is just The Principle of Maximum Entropy (MaxEnt) where constraints are changing in time. This simply amounts to an additional step to summarize the microstate (from MaxEnt) as it changes. The MaxEnt algorithm and therefore, MaxCal, are only applicable for information in the form of expectation values. Microscopic information does not have a place in it. However, macroscopic information is in essence a summary of microscopic information. Therefore, it makes logical sense that microscopic information would shape these distributions. To address the main objective of the project, the next task was to show that MaxCal is a special case of MrE. MaxCal is a special case of MaxEnt and MaxEnt is a special case of MrE. Therefore, MaxCal is a special case of MrE. However, the specific objective was to determine the general theoretical distributions that include both non-equilibrium path information and observational information. This is simply an application and I include several examples to illustrate this application.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 31, 2015
Accession Number
AD1001388

Entities

People

  • Adom Giffin

Organizations

  • Clarkson University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Agreements
  • Algorithms
  • Bayes Theorem
  • Bayesian Inference
  • Chemical Kinetics
  • Delta Functions
  • Department Of Defense
  • Dynamics
  • Energy
  • Engineering
  • Fluid Dynamics
  • Law
  • Mathematics
  • Probability
  • Probability Distributions
  • Students
  • Variational Principles

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Computational Fluid Dynamics (CFD)