Reduced Basis Greedy Selection Using Random Training Sets

DOI10.1051/M2AN/2020004MaRDI QIDQ3304879zbMATH OpenOpenAlexFDO

Authors

Publication date 3 August 2020

Published in ESAIM: Mathematical Modelling and Numerical Analysis (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/1810.09344

zbMATH Keywords

entropy numbersKolmogorov \(n\)-widthsperformance boundsrandom samplingreduced basessparse high-dimensional polynomial approximation

Mathematics Subject Classification ID

Dependence of solutions to PDEs on initial and/or boundary data and/or on parameters of PDEs (35B30) Rate of convergence, degree of approximation (41A25) Manifolds of mappings (58D15) Neural nets and related approaches to inference from stochastic processes (62M45) Numerical interpolation (65D05) Complexity and performance of numerical algorithms (65Y20) Algorithmic information theory (Kolmogorov complexity, etc.) (68Q30) Computational learning theory (68Q32)

Abstract: Reduced bases have been introduced for the approximation of parametrized PDEs in applications where many online queries are required. Their numerical efficiency for such problems has been theoretically confirmed in cite{BCDDPW,DPW}, where it is shown that the reduced basis space $V_n$ of dimension $n$, constructed by a certain greedy strategy, has approximation error similar to that of the optimal space associated to the Kolmogorov $n$-width of the solution manifold. The greedy construction of the reduced basis space is performed in an offline stage which requires at each step a maximization of the current error over the parameter space. For the purpose of numerical computation, this maximization is performed over a finite {em training set} obtained through a discretization. of the parameter domain. To guarantee a final approximation error $varepsilon$ for the space generated by the greedy algorithm requires in principle that the snapshots associated to this training set constitute an approximation net for the solution manifold with accuracy or order $varepsilon$. Hence, the size of the training set is the $varepsilon$ covering number for $mathcalM$ and this covering number typically behaves like $exp(Cvarepsilo{n}^{-1/s})$ for some $C>0$ when the solution manifold has $n$-width decay $O(n^{-s})$. Thus, the shear size of the training set prohibits implementation of the algorithm when $varepsilon$ is small. The main result of this paper shows that, if one is willing to accept results which hold with high probability, rather than with certainty, then for a large class of relevant problems one may replace the fine discretization by a random training set of size polynomial in $varepsilo{n}^{-1}$. Our proof of this fact is established by using inverse inequalities for polynomials in high dimensions.

Recommendations

• Convergence rates for greedy algorithms in reduced basis methods
• Offline-enhanced reduced basis method through adaptive construction of the surrogate training set
• \textit{A priori} convergence of the greedy algorithm for the parametrized reduced basis method
• Greedy algorithms for reduced bases in Banach spaces

Cites work

• \textit{A priori} convergence of the greedy algorithm for the parametrized reduced basis method
• A priori convergence theory for reduced-basis approximations of single-parameter elliptic partial differential equations
• Analytic regularity and polynomial approximation of parametric and stochastic elliptic PDE's
• Approximation of high-dimensional parametric PDEs
• Breaking the curse of dimensionality in sparse polynomial approximation of parametric PDEs
• Convergence rates for greedy algorithms in reduced basis methods
• Discrete least squares polynomial approximation with random evaluations - application to parametric and stochastic elliptic PDEs
• Global a priori convergence theory for reduced-basis approximations of single-parameter symmetric coercive elliptic partial differential equations
• Greedy algorithms for reduced bases in Banach spaces
• scientific article; zbMATH DE number 194093 (Why is no real title available?)
• Multivariate approximation in downward closed polynomial spaces
• Reduced basis approximation and a posteriori error estimation for affinely parametrized elliptic coercive partial differential equations. Application to transport and continuum mechanics.
• Sparse polynomial approximation of parametric elliptic PDEs. Part I: affine coefficients
• The Marcinkiewicz-type discretization theorems

Cited in

(36)

• Offline-enhanced reduced basis method through adaptive construction of the surrogate training set
• A ROM-accelerated parallel-in-time preconditioner for solving all-at-once systems in unsteady convection-diffusion PDEs
• A FOM/ROM hybrid approach for accelerating numerical simulations
• A training set subsampling strategy for the reduced basis method
• A reduced order modeling method based on GNAT-embedded hybrid snapshot simulation
• How to best sample a solution manifold?
• Convergence rates for greedy algorithms in reduced basis methods
• Optimal reduced model algorithms for data-based state estimation
• Nonlinear model reduction on metric spaces. Application to one-dimensional conservative PDEs in Wasserstein spaces
• Nonlinear methods for model reduction
• Nonlinear reduced DNN models for state estimation
• On the stability and accuracy of the empirical interpolation method and gravitational wave surrogates
• Convergence rates of the POD-greedy method
• A training set and multiple bases generation approach for parameterized model reduction based on adaptive grids in parameter space
• Efficient greedy algorithms for high-dimensional parameter spaces with applications to empirical interpolation and reduced basis methods
• Machine Learning: ECML 2004
• Nonlinear reduced models for state and parameter estimation
• An iterative multi‐fidelity approach for model order reduction of multidimensional input parametric PDE systems
• A probabilistic reduced basis method for parameter-dependent problems
• Dual natural-norm a posteriori error estimators for reduced basis approximations to parametrized linear equations
• State Estimation—The Role of Reduced Models
• Multiplicative reduced bases for hyperelasticity
• Inverse problems: a deterministic approach using physics-based reduced models
• Influence of sampling on the convergence rates of greedy algorithms for parameter-dependent random variables
• Structured optimization-based model order reduction for parametric systems
• Polytope division method: a scalable sampling method for problems with high-dimensional parameters
• Randomized greedy algorithms for neural network optimization in solving partial differential equations
• Coarse reduced model selection for nonlinear state estimation
• Tree-based nonlinear reduced modeling
• Introduction: Wolfgang Dahmen's mathematical work (as of 2009)
• Compositional sparsity, approximation classes, and parametric transport equations
• A subspace method based on the Neumann series for the solution of parametric linear systems
• Structure-preserving interpolation reduced-order modelling for efficient simulation of parametric port-Hamiltonian systems
• Nonlinear manifold approximation using compositional polynomial networks
• Nonlinear model reduction for transport-dominated problems
• SABER: stable adaptive barycentric extension for reconstruction of high-dimensional functions

This page was built for publication: Reduced Basis Greedy Selection Using Random Training Sets

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3304879)