Back to all announcements

Quantitative Analysts Have Fastest Linear Programming Optimisation, Classification Neural Networks, and Visualisation Tools from Visual Numerics® IMSL™ C# and JMSL™ Numerical Libraries for Portfolio Modelling

HOUSTON, Texas – June 14, 2006 – Visual Numerics, Inc., celebrating 35 years of producing leading numerical analysis and visualisation software, today announced the availability of the IMSL C# Numerical Library version 4.0 and JMSL Numerical Library for Java™ Applications version 4.0. These libraries now include the world’s fastest, most robust high performance dense linear programming optimiser in pure Java and C#, the most sophisticated classification neural network technology in a broad numerical library for data mining and business intelligence, and 3D charting capabilities. The combination of these features is optimal for capital market companies who need to build advanced applications capable of performing portfolio optimisation, trading systems analysis and risk modelling on a wide variety of computing platforms.

Many companies in the financial services industry do not have access to standard methods or algorithms for quantitative model development. As a result, quantitative engineers must go through a lengthy trial and error process to build their own models. To prevent quantitative engineers from spending years designing just one mathematical and statistical algorithm, Visual Numerics now offers up to 100 fully-tested classes in version 4.0 of the IMSL C# and JMSL Numerical Libraries to help engineers build applications that solve real world problems, and get those solutions to market faster.

New features of IMSL C# 4.0 and JMSL 4.0 include:

· High-performance Dense Linear Programming Optimiser — For state-of-the-art constrained dense linear programming optimisation. Both IMSL C# 4.0 and JMSL 4.0 now offer the world’s fastest, most robust optimiser of its kind in pure Java or C# for use in financial and business analytics. In tests against two established Fortran-based Linear Programming Optimiser competitors, the JMSL library 4.0 and IMSL C# library 4.0 performed on par with both of them - an impressive feat for a Java-based algorithm.

· Classification Neural Nets — A key algorithm in developing business intelligence applications, and used for classical data mining. This binary classification and multi-classification algorithm complements Visual Numerics’ existing forecasting functionality in the JMSL Library. The feature categorises data to identify groups that are more likely to perform certain behaviours - for example, identifying mobile phone users who are more likely to move to a higher priced plan, or identifying credit card usage that is more likely to be fraudulent

· Mersenne Twister Algorithm — New random number generator that improves accuracy in Monte Carlo simulations, and is widely used in financial applications

· 3D Charting surface plotting features in JMSL – A feature set requested by many customers that includes surface and 3D scatter charts

· 2D Charting in IMSL C# (Available in Q3’06)— Aligns with JMSL 2D charting and includes pure C# managed code for .NET environments

“Financial services companies are under incredible pressure to synthesize massive amounts of data to produce real-time forecasts and investment recommendations for their clients,” said Phil Fraher, president and CEO of Visual Numerics. “By offering highly advanced mathematical and statistical libraries with the fastest, most robust high performance dense linear programming optimiser and charting capabilities, we’re giving portfolio managers a powerful business intelligence toolset for making more accurate, up-to-the-minute financial recommendations.”