Abstract:The convergence of the conjugate gradient method for solving large-scale and sparse linear equation systems depends on the spectral properties of the system matrix, which can be improved by preconditioning. In this paper, we develop a computationally efficient data-driven approach to accelerate the generation of effective preconditioners. We, therefore, replace the typically hand-engineered preconditioners by the output of graph neural networks. Our method generates an incomplete factorization of the matrix and is, therefore, referred to as neural incomplete factorization (NeuralIF). Optimizing the condition number of the linear system directly is computationally infeasible. Instead, we utilize a stochastic approximation of the Frobenius loss which only requires matrix-vector multiplications for efficient training. At the core of our method is a novel message-passing block, inspired by sparse matrix theory, that aligns with the objective of finding a sparse factorization of the matrix. We evaluate our proposed method on both synthetic problem instances and on problems arising from the discretization of the Poisson equation on varying domains. Our experiments show that by using data-driven preconditioners within the conjugate gradient method we are able to speed up the convergence of the iterative procedure. The code is available at <a class="link-external link-https" href="https://github.com/paulhausner/neural-incomplete-factorization" rel="external noopener nofollow">this https URL</a>.

A Class of Inexact Block Factorization Preconditioners for Indefinite Matrices with a Three-by-three Block Structure.

High Performance Block Incomplete LU Factorization

Incomplete Factorization Preconditioning for Linear Least Squares Problems

Parallel Incomplete Factorization Preconditioning of Block Tridiagonal Linear Systems with 2-D Domain Decomposition

CIMGS: an Incomplete Orthogonal FactorizationPreconditioner

Block-splitting preconditioners for indefinite least squares problem

Some preconditioning techniques for a class of double saddle point problems

Practical Restrictively Preconditioned Conjugate Gradient Methods for a Class of Block Two-by-Two Linear Systems

Parallelization of incomplete factorization preconditioning of block tridiagonal linear systems with 1-D domain decomposition

A Preconditioned MINRES Method for Block Lower Triangular Toeplitz Systems

Block [math]-Circulant Preconditioners for Parabolic Optimal Control Problems

Adaptive Factored Incomplete Inverse Matrices

A sparse-sparse iteration for computing a sparse incomplete factorization of the inverse of an SPD matrix

An approximate block factorization preconditioner for mixed-dimensional beam-solid interaction

On the improvement of shift-splitting preconditioners for double saddle point problems

A Parallelization Technique Based on Factor Combination and Graph Partitioning for General Incomplete LU Factorization

A New Matrix Truncation Method for Improving Approximate Factorisation Preconditioners

Neural incomplete factorization: learning preconditioners for the conjugate gradient method

A CONSTRAINT PRECONDITIONER FOR SOLVING SYMMETRIC POSITIVE DEFINITE SYSTEMS AND APPLICATION TO THE HELMHOLTZ EQUATIONS AND POISSON EQUATIONS