Double diffusion maps and their latent harmonics for scientific computations in latent space

DOI10.1016/J.JCP.2023.112072arXiv2204.12536OpenAlexW4362606235MaRDI QIDQ6158080FDOQ6158080

Authors: Nikolaos Evangelou, Felix Dietrich, Eliodoro Chiavazzo, Daniel Lehmberg, Marina Meilă, I. G. Kevrekidis 

Publication date: 31 May 2023

Published in: Journal of Computational Physics (Search for Journal in Brave)

Abstract: We introduce a data-driven approach to building reduced dynamical models through manifold learning; the reduced latent space is discovered using Diffusion Maps (a manifold learning technique) on time series data. A second round of Diffusion Maps on those latent coordinates allows the approximation of the reduced dynamical models. This second round enables mapping the latent space coordinates back to the full ambient space (what is called lifting); it also enables the approximation of full state functions of interest in terms of the reduced coordinates. In our work, we develop and test three different reduced numerical simulation methodologies, either through pre-tabulation in the latent space and integration on the fly or by going back and forth between the ambient space and the latent space. The data-driven latent space simulation results, based on the three different approaches, are validated through (a) the latent space observation of the full simulation through the Nystr"om Extension formula, or through (b) lifting the reduced trajectory back to the full ambient space, via Latent Harmonics. Latent space modeling often involves additional regularization to favor certain properties of the space over others, and the mapping back to the ambient space is then constructed mostly independently from these properties; here, we use the same data-driven approach to construct the latent space and then map back to the ambient space.

Full work available at URL: https://arxiv.org/abs/2204.12536

Recommendations

• Grassmannian diffusion maps based surrogate modeling via geometric harmonics
• Geometric diffusions as a tool for harmonic analysis and structure definition of data: multiscale methods
• Vector diffusion maps and the connection Laplacian
• Geometric diffusions as a tool for harmonic analysis and structure definition of data: diffusion maps
• Diffusion operators for multimodal data analysis
• Diffusion maps
• Spectral Convergence of Diffusion Maps: Improved Error Bounds and an Alternative Normalization
• Approximately-isometric diffusion maps
• Diffusion estimation from multiscale data by operator eigenpairs

zbMATH Keywords

dynamical systemsmanifold learningscientific computingdiffusion maps

Mathematics Subject Classification ID

Smooth dynamical systems: general theory (37Cxx) Artificial intelligence (68Txx) Parabolic equations and parabolic systems (35Kxx)

Cites Work

• Diffusion maps
• Equation-free, coarse-grained multiscale computation: enabling microscopic simulators to perform system-level analysis
• Inertial manifolds for nonlinear evolutionary equations
• Time-scale separation from diffusion-mapped delay coordinates
• Approximate inertial manifolds for the Kuramoto-Sivashinsky equation: Analysis and computations
• A survey of projection-based model reduction methods for parametric dynamical systems
• Geometric harmonics: a novel tool for multiscale out-of-sample extension of empirical functions
• Low-dimensional behaviour in the complex Ginzburg-Landau equation
• Explicit construction of an inertial manifold for a reaction diffusion equation
• Integral manifolds and inertial manifolds for dissipative partial differential equations
• Title not available (Why is that?)
• Low-dimensional models for complex geometry flows: Application to grooved channels and circular cylinders
• Dimension of the attractors associated to the Ginzburg-Landau partial differential equation
• Graph Laplacian Tomography From Unknown Random Projections
• Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Finite-dimensional asymptotic behavior for the Swift-Hohenberg model of convection
• Parsimonious representation of nonlinear dynamical systems through manifold learning: a chemotaxis case study
• Data-driven discovery of coordinates and governing equations

Cited In (3)

• Physics-agnostic and physics-infused machine learning for thin films flows: modelling, and predictions from small data
• Time-series forecasting using manifold learning, radial basis function interpolation, and geometric harmonics
• Slow invariant manifolds of fast-slow systems of ODEs with physics-informed neural networks

This page was built for publication: Double diffusion maps and their latent harmonics for scientific computations in latent space

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6158080)