Global Optimization Portal | ||||||
Welcome to Global Optimization Portal (GOP). This portal is a continuation of RDIC on our website and it is aimed at providing efficient and practical algorithms for "Surrogate Model" based optimization applications in any dimensionality (1, 2,,,, 10, 20, 100 and more). The algorithms formulated at GOP are strictly based on mathematics. Therefore, they are independent of the nature of the application and can be applied to different areas of data analysis, for example,
In simulation, one can search for the optimal solution according to the “objective function” defined for the system. Standard numerical optimization method is used to find the parameters which typically yield local optimum. Because of the sophistication of the simulation, the objective function which exists in multi-dimensional space can demand excessively high CPU time and cost to find the optimum through the iterative optimization process. To overcome this problem, one can use the approach of interpolation theory to acquire sampling values of the function, and build an analytic function (called interpolant or interpolating function) which approximates the objective function and is simpler to compute and manage. The interpolant established is the "Surrogate Model" of the original objective function. In consequence, one can replace the complex simulation code by the "Surrogate Model" in the optimum searching routine. The resulting gain from utilizing the interpolant can far outweigh the approximation incurred. GOP is for providing speed and cost reduction to the applications of optimization. The solution is based upon data-driven surrogate modeling in arbitrary dimensionality. It is capable of using limited amount of input samples to explore and capture the overall charateristics of the design space. In low dimensionality, GOP appears similar to IDW (inverse distance weighting) method but GOP is more rigorous and not ad-hoc. It is a fast, efficient tool to process interpolation applications in the geostatistical arena such as mining reserve estimation, pollutant tracking, soil science mapping and meteorology predictions. GOP solution depends upon the interpolation employed in the formalism together with the subsequent optimization process. Successful application of GOP requires the preparation of the real world input data as well as the realistic expectation of tolerance limits. | |
 |  |
 |  |
© FANG, INC. 2021 All Rights Reserved
(All information including software provided at GOP should be used by the discretion of the user.
FANG, INC. is not responsible for the outcome of the usage of GOP.)