News - Xpress 2003 Release

18 July 2002

Dash Optimization Announces Xpress-MP 2003

ENGLEWOOD CLIFFS, NJ and LONDON: Dash Optimization today announced that Xpress-MP 2003 will be released to coincide with the User Meeting to be held in London in September 2002.

Xpress-MP continues to gain market share across a wide range of vertical applications and Dash Optimization has become the preferred optimization and modeling software technology partner of end-users, consultants and OEMs who require performance, reliability and ease-of-use. Xpress-MP 2003 will contain significant innovations in core technology and new layered products designed to address additional problems that frequently arise in areas such as production planning, supply chain, sourcing, resource allocation and asset management.

Optimizer Innovations

Improvements to the core technology of Xpress-MP follow the basic development principles of Dash Optimization of solving ever-larger problems, quickly and reliably. These include:

Dash Optimization has assembled over 1,000 problems from a wide range of vertical applications and now runs all new software through an automated testing and benchmarking system. Up to 100% improvements in speed have been observed between version 13 and the 2003 release. In particular, Dash is confident that independent benchmarks will show that the MIP within the Xpress-Optimizer is the more often than not the first to achieve a good integer solution (within 5% of the absolute optimum) demonstrating yet again that Xpress-MP does indeed get good solutions quickly.

Modeling Technology

Xpress-Mosel and Xpress-IVE – released in November 2001 – have been significantly enhanced and improved and many Dash customers are now upgrading from the previous 'mp-model' technology. An automatic translation tool – which has been proven on a wide variety of complex models – is available to enable customers to upgrade easily from the previous technology. This translation tool has also demonstrated that Xpress-Mosel can be several orders of magnitude faster in model formulation. The new version of Xpress-Mosel will have an open interface enabling user written or third party programs to be integrated into optimization applications. This complements the unique capabilities of Xpress-Mosel to build optimizer applications incorporating user written algorithms and heuristics. The new version of Xpress-IVE has a rich set of visualization tools including solution path and matrix visualization, making building, testing and tuning applications simpler and easier.

Non-Linear Optimization

Xpress-MP 2003 includes Xpress-SLP, a non-linear optimization code with mixed-integer capability (MINLP) which is currently available in Beta form. Customers in areas such as the process industry, finance, marketing and retail have driven this development. Xpress-SLP will be the World's first large-scale, globally supported MINLP component and is able to solve a much wider range of optimization problems. Xpress-SLP will have full modeling support within Xpress-Mosel and Xpress-IVE, and the full range of embedding and interfacing capabilities enjoyed by the LP/MIP/QP and MIQP optimizers already available within the Xpress-MP family. End-users, consultants and OEMs will have the full range of productivity and software tools with which to express, solve and deploy non-linear optimization models.

Stochastic Optimization

At the September User Meeting, Dash will also be making a demonstration of Stochastic Optimization software applied to several real-world problems. Initially prototyped in Xpress-Mosel, this software will be further developed during 2003 with a view to a product release, Xpress-SP, early in 2003. Dash believes that Stochastic Optimization is an essential technology for applications using sophisticated statistics and prediction technology. The combination of the two will release significant additional value in a wide range of vertical sectors including energy, process, retail and finance.

Simon Bragg, ARC's European Research Director adds: "Dash is the innovator in the mathematical programming arena. SLP is particularly useful. Programmers have traditionally devoted considerable ingenuity in order to handle non-linear variables and constraints. Xpress-SLP saves this bother. Secondly, planners, investors and analysts are learning the hard way that the future is not just a simple extrapolation of the past. Stochastic Optimization will help them understand multiple scenarios, and create a robust strategy that can withstand surprises. Maybe, now, those that execute plans will start to respect what these planners produce."