Trade.Information

Product Name: IISS - Intelligent Iterative Solver Service

Product Description

The core of most engineering simulation applications requires a large sparse matrix solver. In particular, iterative methods may become the only choice for a very large matrix. When the iterative solver consumes 60-80% of the total CPU time: any improvement in the solver can have a substantial influence on the efficiency. However, the optimal choice of an effective iterative solver varies; these methods are also known to suffer from irregular performance in different applications. While one method may be very effective for one application, it may perform poorly for another. Additionally, even with a good solver and application match, choosing the right parameters-required for performance-can be challenging for regular users. Many of the behaviors of these methods are still open to researchers. The Boeing Math Group has mastered a collection of the state-of-the-art iterative solvers placing them under one umbrella:

• BCSLIB-EXT Iterative Solver is an industrial strength preconditioned iterative solver. It is designed for many linear systems derived from PDE or PDE systems. This solver has been successfuly used in CFD, semiconductor device simulations, ground water contamination, finance option analysis and many others.
• Hypre is a scalable software for solving large, sparse linear systems of equations on massively parallel computers.
• AZTEC is a parallel iterative library for solving linear systems, which takes advantage of advanced partitioning techniques (Chaco) and utilizes efficient dense matrix algorithms when solving block sparse matrices. The Aztec iterative solver package grew out of a specific application: modeling reacting flows.

Our solver group consists of the world-renowned experts. We provide the following unparalleled service:

• Carry out the analysis of your linear problems;Identify the best solver for your application;
• Customize the solver to fit your application; tune the solver to provide the near black box robustness of direct factorizations, but without the computation and memory demands
• Provide consultation for the best usage of each solver;
• Ensure the consistently optimal performance for your applications.