What's New for the IMSL® C Numerical Library Version 6.0


New World-Class High Performance Linear Programming Optimizer
  • New, state of the art constrained dense Linear Programming Optimizer
  • World’s fastest speed in a general mathematical library
  • Includes MPS format readers to facilitate usage with existing users of Linear Programming
  • Well suited for capital markets customers doing portfolio optimization, trading systems and risk modeling
  • Studies using 91 dense LP problems from Netlib produced the following robustness and performance results:



Advanced Forecasting Package

  • New Neural Network engine for data mining and forecasting
    • A technique that takes advantage of repetitive learning through repeated forecasts and comparisons against actual outcomes
    • Powerful for situations where data contains missing values, short time periods, inconsistent reporting of data and where the user has multiple items to predict
    • Performance: The table below compares neural network forecast to the actual order for DRAM chips over a 2-year period
  • Auto_ARIMA
    • An “expert system” time series forecasting preconditioner and engine for forecasting situations where automated forecasting is desirable
    • Useful for data containing seasonality, level shifts, and outliers
    • Automatic selection of model parameters for straightforward usage

  • Time series forecasting preprocessing algorithms for
    • Outlier detection
    • Outlier classification
    • Missing value estimation
Differential Equations package for financial engineering and scientific computing
  • Useful in customized options pricing algorithm development
  • dea_petzold_gear algorithm solves a first order differential-algebraic system of equations using Petzold-Gear method
  • pde_1d_mg – solution of systems for PDEs in one dimension, method of lines with variable griddings
Mersenne Twister random number generator technique
  • Provides for fast generation of very high quality pseudorandom numbers

New LAPACK functionality
  • Conveniently accessible for users coding in C


CompanyProducts & ServicesSolutionsSuccessesSupportDownloadsEmail this page
© Copyright 2008 Visual Numerics, Inc. All Rights Reserved Legal Privacy