|
Numerical Libraries Enhanced
Linear programming optimization, classification neural networks, and visualization tools key changes in V4.0 IMSL C# and JMSL.
by DE Editors | Published June 1, 2006
On June 14, Visual Numerics, Inc. (VNI; Houston, TX) released the 4.0 versions of its IMSL C#Numerical Library and JMSL Numerical Library for Java applicationdevelopment. Both libraries, VNI asserts, now include fast, and robusthigh-performance dense linear programming optimizer in pure Java andC#, sophisticated classification neural network technology, andenhanced charting capabilities. This combination of features, says VNI,is optimal for companies building advanced applications such riskmodeling.
V4.0 of the IMSL C# and JMSL libraries offers up to 100 fully-testedclasses that provide standard methodologies as well as mathematical andstatistical algorithms for quantitative model development. Both IMSL C#4.0 and JMSL 4.0 now offer high-performance dense linear programmingoptimizer. VNI claims the optimizer is the fastest and most robustoptimizer in pure Java or C# of its kind. Backing this claim, VNI citestests against a pair of Fortran-based linear programming optimizersshowing that JMSL 4.0 and IMSL C# 4.0 performed on par with eachFortran optimizer.
A binary classification and multi-classification algorithm used in datamining known as Classification Neural Nets now complements the'existing forecasting functionality in the JMSL Library. TheClassification Neural Nets feature categorizes data to identify groupsthat are more likely to perform certain behaviors. Also new is aVNI-supported—as opposed to freeware—Mersenne Twister algorithm forMonte Carlo simulations.
JMSL also has been enhanced with new 3D Charting surface plottingfeatures, including surface and 3D scatter charts. Later this year, 2DCharting will be offered for IMSL C#.
Both libraries are available immediately. A single development seat ofthe IMSL C# Numerical Library 4.0 is $995. The JMSL Numerical Libraryfor Java 4.0 costs $2,895 per development seat. For complete details, click here.
|
