Developed by Brown University in the US, Numerica is a new optimisation technique that can help engineers and academics solve global resolution algorithms. The technique has been implemented into a new desktop engineering calculator called ILOG Numerica that can speed up the modelling and solving process for engineers in the mechanical, robotics and kinematics industries. With it, processes and problems can be modelled using non-linear constraints with global optimisation techniques. It can apparently solve a benchmark problem, such as the Ebers and Moll Transistor modelling problem, in a matter of 40min on a standalone PC. In the past, it took a network of 30 Sun Sparc-1 workstations more than fourteen months to perform the same task.
Well, until now, mathematical software tools have only been able to search `locally’ for solutions, resulting in a choice which, although mathematically correct, cannot always be applied in practice. ILOG Numerica allows engineers to extend this search and find the optimum solution that can be applied to a physical problem. For example, engineers can calculate the optimum reach of a robotic arm or the correct combination of chemical elements needed to reach optimum production.
Numerica combines recent advances in the fields of constraint-based programming and numerical analysis in a desktop software tool. Hence, highly non-linear problems can be solved on a standard PC.
Numerica was developed by Dr Pascal van Hentenryck at Brown University and is described in the book Numerica by Dr van Hentenryck, Laurent Michel and Yves Deville.