A matrix-free ILU realization based on surrogates

TL;DR

This paper introduces a novel matrix-free ILU(0) smoother for finite element methods on hybrid tetrahedral grids, enabling efficient large-scale simulations with reduced memory usage.

Contribution

It proposes a memory-efficient, matrix-free ILU(0) smoother using surrogate matrix polynomials for low-order finite elements on hybrid grids, enhancing large-scale solver performance.

Findings

Effective ILU(0) smoothing on macro-tetrahedra

Surrogate polynomial evaluation reduces memory footprint

Convergence rates depend on polynomial order

Abstract

Matrix-free techniques play an increasingly important role in large-scale simulations. Schur complement techniques and massively parallel multigrid solvers for second-order elliptic partial differential equations can significantly benefit from reduced memory traffic and consumption. The matrix-free approach often restricts solver components to purely local operations, for instance, the Jacobi- or Gauss--Seidel-Smoothers in multigrid methods. An incomplete LU (ILU) decomposition cannot be calculated from local information and is therefore not amenable to an on-the-fly computation which is typically needed for matrix-free calculations. It generally requires the storage and factorization of a sparse matrix which contradicts the low memory requirements in large scale scenarios. In this work, we propose a matrix-free ILU realization. More precisely, we introduce a memory-efficient, matrix-free ILU(0)-Smoother component for low-order conforming finite elements on tetrahedral hybrid grids. Hybrid grids consist of an unstructured macro-mesh which is subdivided into a structured micro-mesh. The ILU(0) is used for degrees-of-freedom assigned to the interior of macro-tetrahedra. This ILU(0)-Smoother can be used for the efficient matrix-free evaluation of the Steklov-Poincare operator from domain-decomposition methods. After introducing and formally defining our smoother, we investigate its performance on refined macro-tetrahedra. Secondly, the ILU(0)-Smoother on the macro-tetrahedrons is implemented via surrogate matrix polynomials in conjunction with a fast on-the-fly evaluation scheme resulting in an efficient matrix-free algorithm. The polynomial coefficients are obtained by solving a least-squares problem on a small part of the factorized ILU(0) matrices to stay memory efficient. The convergence rates of this smoother with respect to the polynomial order are thoroughly studied.