This paper presents an improved adaptive cross approximation (ACA)approach developed in conjunction with the Hierarchical format matrix and theGMRES solver. A novel scheme to generate the cluster tree (based upon preliminaryconsiderations of the prescribed boundary conditions) and an improved ACAalgorithm (approximating the system matrix for mixed Robin conditions) are described.The asymptotic smoothness property of a kernel generated by a linearcombination of two asymptotic smooth kernels is demonstrated. Numerical resultsshow the new approach to be up to 50% faster than the conventional ACAapproach.
|Numero di pagine||24|
|Rivista||COMPUTER MODELING IN ENGINEERING & SCIENCES|
|Stato di pubblicazione||Published - 2011|
All Science Journal Classification (ASJC) codes