SOLVING A DETERMINISTIC GLOBAL OPTIMISATION PROBLEM

摘要

A computer implemented method for solving a Deterministic Global Optimisation (DGO) problem is provided. The method involves defining an input optimisation problem based on the DGO problem. An Al machine 201 then generates an updated configuration of an algorithm for solving the input optimisation problem, the updated configuration comprising a plurality of tasks. Generation of the updated configuration is based on metadata. The Al machine 203 distributes the plurality of tasks to different processing units of a DGO engine 203 for processing, based on the updated configuration. The results of the processing and metadata representing information about the processing are received by the Al machine 201 from the DGO engine 203, and, until it is proven that a global solution to the DGO problem has been found within a user-defined tolerance, the configuration of the algorithm is further updated using the received metadata and the processing is repeated.