The Pulickal Algorithm: A Multi-Directional Block-Parallel Elimination Framework for Dense Linear Systems

Read full text Summarise with VIRAT
Authors: Pearl Bipin
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Traditional direct solvers for dense linear systems, such as Gaussian elimination and LU decomposition, are inherently bottlenecked by strict sequential dependencies. The critical path of standard elimination forces a unidirectional data flow from the top-left to the bottom-right of the coefficient matrix, severely limiting parallel scalability on modern multi-core and distributed architectures. This paper introduces the Pulickal Algorithm, a novel multi-directional elimination framework that shatters this sequential dependency. By utilizing a bi-directional twisted block factorization, the algorithm initializes concurrent execution fronts at opposite polarities of the matrix. These independent fronts perform localized trian-gularizations and boundary updates that converge simultaneously on a centralized reduced system, termed the Pulickal Core. We outline the mathematical formulation of this dual-Schur domain decomposition, detail the algorithmic architecture, and provide a theoretical analysis of its computational complexity. This framework offers a highly scalable alternative for high-performance computing, maximizing parallel throughput and minimizing idle processor time during the dense factorization phase.

Fields of Study & Subjects

Matrix Theory and Algorithms Parallel Computing and Optimization Techniques Numerical Methods and Algorithms

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
20 March 2026
Language
EN
Document type
Preprint