Speaker
Dr
Mark Amo-Boateng
(Univerity of Energy and Natural Resources)
Description
This article presents a new general purpose stochastic algorithm for large-scale optimization problems on mobile phones and embedded systems. Application of the algorithm to the Griewank function was possible in up to 100 million decision variables in double precision on a Samsung Note 4 mobile phone, and 10 million decision variables on a Raspberry Pi 3. Solving the Griewank function for 1,000 decision variables took less than 1 second on both the Samsung mobile and Raspberry Pi. In Africa where High-Performance Computing hardware is limited and expensive to setup, this algorithm opens up new possibilities to the advancement of computational sciences and research with already existing and cheap infrastructure.
Primary author
Dr
Mark Amo-Boateng
(Univerity of Energy and Natural Resources)
