AI‐enhanced iterative solvers for accelerating the solution of large‐scale parametrized systems
From MaRDI portal
(Redirected from Publication:6148519)
Abstract: Recent advances in the field of machine learning open a new era in high performance computing. Applications of machine learning algorithms for the development of accurate and cost-efficient surrogates of complex problems have already attracted major attention from scientists. Despite their powerful approximation capabilities, however, surrogates cannot produce the `exact' solution to the problem. To address this issue, this paper exploits up-to-date ML tools and delivers customized iterative solvers of linear equation systems, capable of solving large-scale parametrized problems at any desired level of accuracy. Specifically, the proposed approach consists of the following two steps. At first, a reduced set of model evaluations is performed and the corresponding solutions are used to establish an approximate mapping from the problem's parametric space to its solution space using deep feedforward neural networks and convolutional autoencoders. This mapping serves a means to obtain very accurate initial predictions of the system's response to new query points at negligible computational cost. Subsequently, an iterative solver inspired by the Algebraic Multigrid method in combination with Proper Orthogonal Decomposition, termed POD-2G, is developed that successively refines the initial predictions towards the exact system solutions. The application of POD-2G as a standalone solver or as preconditioner in the context of preconditioned conjugate gradient methods is demonstrated on several numerical examples of large scale systems, with the results indicating its superiority over conventional iterative solution schemes.
Recommendations
- Parametric model reduction with convolutional neural networks
- Machine learning for preconditioning elliptic equations in porous microstructures: a path to error control
- Machine learning acceleration for nonlinear solvers applied to multiphase porous media flow
- Semi-supervised invertible neural operators for Bayesian inverse problems
- Isogeometric neural networks: a new deep learning approach for solving parameterized partial differential equations
Cites work
- A comparison of neural network architectures for data-driven reduced-order modeling
- A comprehensive deep learning-based approach to reduced order modeling of nonlinear time-dependent parametrized PDEs
- A Deflated Version of the Conjugate Gradient Algorithm
- A low-cost, goal-oriented `compact proper orthogonal decomposition' basis for model reduction of static systems
- A method of finite element tearing and interconnecting and its parallel solution algorithm
- A multi-element group preconditioned GMRES algorithm for nonsymmetric systems arising in finite element analysis
- A multigrid preconditioner for stabilised discretisations of advection-diffusion problems
- A multilevel Schwarz preconditioner based on a hierarchy of robust coarse spaces
- A New Look at Proper Orthogonal Decomposition
- A new perspective on the solution of uncertainty quantification and reliability analysis of large-scale problems
- A Restricted Additive Schwarz Preconditioner for General Sparse Linear Systems
- A review of algebraic multigrid
- Adaptively Preconditioned GMRES Algorithms
- Algebraic multigrid based on element interpolation (AMGe)
- Algebraic multigrid by smoothed aggregation for second and fourth order elliptic problems
- Algebraic multigrid for higher-order finite elements
- An artificial neural network framework for reduced order modeling of transient flows
- An efficient algebraic multigrid preconditioned conjugate gradient solver.
- An efficient multigrid method for graph Laplacian systems
- Analysis of Augmented Krylov Subspace Methods
- Block Preconditioning for the Conjugate Gradient Method
- Data driven approximation of parametrized PDEs by reduced basis and neural networks
- Deflated and Augmented Krylov Subspace Techniques
- Diffusion maps-aided neural networks for the solution of parametrized PDEs
- Domain decomposition methods for large linearly elliptic three-dimensional problems
- Efficient preconditioners for boundary element matrices based on grey-box algebraic multigrid methods
- Fast iterative solution of the optimal transport problem on graphs
- Fourier Analysis of GMRES(m) Preconditioned by Multigrid
- GMRES: A Generalized Minimal Residual Algorithm for Solving Nonsymmetric Linear Systems
- Grassmannian diffusion maps-based dimension reduction and classification for high-dimensional data
- scientific article; zbMATH DE number 2113718 (Why is no real title available?)
- Interpolation of inverse operators for preconditioning parameter-dependent equations
- Krylov subspaces recycling based model order reduction for acoustic BEM systems and an error estimator
- Krylov-subspace recycling via the POD-augmented conjugate-gradient method
- Machine learning in adaptive domain decomposition methods -- predicting the geometric location of constraints
- Meta-mgnet: meta multigrid networks for solving parameterized partial differential equations
- Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
- Multi space reduced basis preconditioners for large-scale parametrized PDEs
- Multi-level convolutional autoencoder networks for parametric prediction of spatio-temporal dynamics
- Multiscale two-stage solver for Biot's poroelasticity equations in subsurface media
- Non intrusive reduced order modeling of parametrized PDEs by kernel POD and neural networks
- Novel preconditioners for the iterative solution to FE-discretized coupled consolidation equations
- On POD-based deflation vectors for DPCG applied to porous media problems
- On the fixed-stress split scheme as smoother in multigrid methods for coupling flow and geomechanics
- On the verification of model reduction methods based on the proper generalized decomposition
- Parameter-robust discretization and preconditioning of Biot's consolidation model
- POD-DL-ROM: enhancing deep learning-based reduced order models for nonlinear parametrized PDEs by proper orthogonal decomposition
- Preconditioned Conjugate Gradient Method for Optimal Control Problems with Control and State Constraints
- Preconditioned iterative methods for fractional diffusion equation
- Preconditioning parametrized linear systems
- Proper generalized decomposition of multiscale models
- Recycling Krylov Subspaces and Truncating Deflation Subspaces for Solving Sequence of Linear Systems
- Reduced order models based on local POD plus Galerkin projection
- Reduced-order fluid/structure modeling of a complete aircraft configuration
- Reliability-based structural optimization using neural networks and Monte Carlo simulation
- Robust approximate inverse preconditioning for the conjugate gradient method
- Square and stretch multigrid for stochastic matrix eigenproblems
- Structural reliability analysis of elastic-plastic structures using neural networks and Monte Carlo simulation
- Superlinearly convergent CG methods via equivalent preconditioning for nonsymmetric elliptic operators
- Superlinearly convergent PCG algorithms for some nonsymmetric elliptic systems
- The LATIN multiscale computational method and the proper generalized decomposition
- The mosaic of high performance domain decomposition methods for structural mechanics: formulation, interrelation and numerical efficiency of primal and dual methods.
- Truncation Strategies for Optimal Krylov Subspace Methods
Cited in
(5)- Fusing nonlinear solvers with transformers for accelerating the solution of parametric transient problems
- A domain decomposition method employing displacement-only partitioned equations for quasi-static structural analysis
- Generative downscaling of PDE solvers with physics-guided diffusion models
- DeepONet based preconditioning strategies for solving parametric linear systems of equations
- Neural Green's operators for parametric partial differential equations
This page was built for publication: AI‐enhanced iterative solvers for accelerating the solution of large‐scale parametrized systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6148519)
