SOLVING THE UPDATED AND DOWNDATED ORDINARY LINEAR MODEL ON MASSIVELY PARALLEL SIMD SYSTEMS
Erricos John Kontoghiorghes, Mairi Clarke
Abstract
Erricos John Kontoghiorghes, Mairi Clarke
Abstract
Several algorithms have appeared for solving the Ordinary Linear Model (OLM), after a number of observations have been added or deleted. In this paper we employ Householder transformations and Givens rotations to solve the updated and downdated OLM, using a massively parallel SIMD computer. Some of our methods are modified versions of serial algorithms published previously while others appear for the first time. The execution time models of all algorithms are studied and compared.
OpenAlex reports 17 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Several algorithms have appeared for solving the Ordinary Linear Model (OLM), after a number of observations have been added or deleted. In this paper we employ Householder transformations and Givens rotations to solve the updated and downdated OLM, using a massively parallel SIMD computer. Some of our methods are modified versions of serial algorithms published previously while others appear for the first time. The execution time models of all algorithms are studied and compared.
Key concepts: SIMD, Massively parallel, Computer science, Parallel computing