Wolfram ResearchProductsPurchasingServices & ResourcesAbout UsOur Sites
What is New in Mathematica 5.1 Products
-----
 /
Mathematica
*String Manipulation
*DatabaseLink
*Binary Data I/O
*Additional Import/Export Formats
*Web Services
*Large-Scale Graph Layout
*Array Plotting
<Enhanced Linear Algebra Performance
*Piecewise
*Integration over Regions
*Event Handling
*Symbolic Differential Equations
*Cluster Analysis
*EquationTrekker
*MathematicaMark2004
*GUIKit
*Summary of Features
*Ask about this page
PreviousNext

Numerical Linear Algebra Performance Enhancements

See What's New in Wolfram Mathematica 6

Building on the dramatic optimizations of Mathematica 5.0, there are significant further enhancements in numerical linear algebra.

For example, using a novel iterative algorithm LinearSolve now achieves still higher performance for dense high-precision matrices, and dramatic speed increases for sparse high-precision matrices.

LinearSolve and other linear algebra functions use sophisticated automatic algorithm-selection methods to analyze their input and choose the algorithms to use.




More Details

The following graph compares the performance of LinearSolve for sparse random matrices of 100-digit numbers in Mathematica 5.1 versus Mathematica 5.0.

LinearSolve is increasingly faster in Mathematica 5.1 as matrix sizes climb.


Related Links
Built-In Functions Reference Guide Built-In Functions Reference Guide
LinearSolve
SparseArray
  
In-Depth Advanced Documentation In-Depth Advanced Documentation
Linear Algebra in Mathematica
  
Additional Information Additional Information
Key Technology: Automatic Algorithm Selection
Technical Presentation: Recent Advances in Linear Algebra Functionality


Back to Top PreviousNext



 © 2008 Wolfram Research, Inc.  Terms of Use  Privacy Policy | [ja]