A comparison of selected SLAE solvers for Epilot

A comparison of selected SLAE solvers for Epilot

Paper in proceedings outside WoS and Scopus

The paper is focused on the search for a suitable solver of systems of linear algebraic equations obtained from the numerical solution of the thermal behavior of the energy pile, performed by the finite volume method. In these cases, sparse, diagonally dominant matrices in the order of hundreds to thousands of elements are typically generated. The paper aims to compare selected algorithms from the open-source Eigen library (QR decomposition of a sparse matrix, LU factorization for a sparse matrix, Conjugate Gradient method, BiConjugate Gradient stabilized method) with a focus on the required computational time and the number of iterations. The numerical solution is presented on several data sets representing stationary (steady) and non-stationary (transient) thermal states of the energy pilot. For the given purposes, the Conjugate Gradient method seems to be the most advantageous in terms of the required computational time.

The paper is focused on the search for a suitable solver of systems of linear algebraic equations obtained from the numerical solution of the thermal behavior of the energy pile, performed by the finite volume method. In these cases, sparse, diagonally dominant matrices in the order of hundreds to thousands of elements are typically generated. The paper aims to compare selected algorithms from the open-source Eigen library (QR decomposition of a sparse matrix, LU factorization for a sparse matrix, Conjugate Gradient method, BiConjugate Gradient stabilized method) with a focus on the required computational time and the number of iterations. The numerical solution is presented on several data sets representing stationary (steady) and non-stationary (transient) thermal states of the energy pilot. For the given purposes, the Conjugate Gradient method seems to be the most advantageous in terms of the required computational time.

computer modeling, numerical computing, iterative methods, system of algebraic equation, sparse matrix

computer modeling, numerical computing, iterative methods, system of algebraic equation, sparse matrix

SEHNALOVÁ, P.; ŠIKULA, O.

2021

05.11.2020

IBPSA-CZ

Brno

978-80-907423-1

Simulace budov a techniky prostředí 2020. Sborník 11. konference IBPSA-CZ

111

114

4