9700 South Cass Avenue

Title: 9700 South Cass Avenue

Research Question: How can optimization algorithms be designed and implemented to effectively solve large-scale problems on parallel architectures?

Methodology: The study uses the Toolkit for Advanced Optimization (TAO), a component-based optimization software designed for large-scale applications. It focuses on the Gradient Projection Conjugate Gradient (GPCG) algorithm, an optimization method for solving bound-constrained quadratic programming problems. The GPCG algorithm was implemented on a parallel architecture using object-oriented techniques and the PETSc library for linear algebra support.

Results: The implementation of GPCG on a parallel architecture showed promising results. The algorithm's performance and scalability were analyzed, revealing that the scalability is limited by the sizes of the matrices involved in the optimization process. The study found that the GPCG algorithm is a prime candidate for a case study on the performance and scalability of optimization algorithms in parallel architectures.

Implications: The results suggest that the GPCG algorithm, when implemented on a parallel architecture, can be an effective tool for solving large-scale optimization problems. The study also highlights the importance of object-oriented techniques and linear algebra support in designing and implementing optimization algorithms for parallel architectures.

Link to Article: https://arxiv.org/abs/0101018v1 Authors: arXiv ID: 0101018v1