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

Download:

Current browse context:

stat.ML

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

Statistics > Machine Learning

Title: Coarse-grained and emergent distributed parameter systems from data

Abstract: We explore the derivation of distributed parameter system evolution laws (and in particular, partial differential operators and associated partial differential equations, PDEs) from spatiotemporal data. This is, of course, a classical identification problem; our focus here is on the use of manifold learning techniques (and, in particular, variations of Diffusion Maps) in conjunction with neural network learning algorithms that allow us to attempt this task when the dependent variables, and even the independent variables of the PDE are not known a priori and must be themselves derived from the data. The similarity measure used in Diffusion Maps for dependent coarse variable detection involves distances between local particle distribution observations; for independent variable detection we use distances between local short-time dynamics. We demonstrate each approach through an illustrative established PDE example. Such variable-free, emergent space identification algorithms connect naturally with equation-free multiscale computation tools.
Comments: specified the corresponding author
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Computational Physics (physics.comp-ph); Data Analysis, Statistics and Probability (physics.data-an)
MSC classes: 93C20
Cite as: arXiv:2011.08138 [stat.ML]
  (or arXiv:2011.08138v2 [stat.ML] for this version)

Submission history

From: Hassan Arbabi [view email]
[v1] Mon, 16 Nov 2020 18:02:01 GMT (1468kb,D)
[v2] Tue, 17 Nov 2020 02:32:05 GMT (1468kb,D)

Link back to: arXiv, form interface, contact.