Asynchronous Parallel Multiple Markov Chains Simulated Annealing Algorithm to Transient Electromagnetic Inversion
S. Liu, Huimin Sun · 2020
Summary The simulated annealing algorithm (SA) is a global optimization algorithm. It has the advantages of not relying on the initial model and being easy to jump out of the local optimum. It has many applications. However, excessively long runtimes have prevented it from being used in more complex and wide areas. Scholars have tried many acceleration methods to improve the practicality of sa algorithm. Based on research of Aarts and Korst (1986) , LEE (1996) propsed the asynchronous Multiple Markov chains Parallel SA (asynchronous MMC PSA) algorithm. Due to the high parallelism of this method, SA can be significantly speeded up. Asynchronous MMC PSA has been used to solve many optimization problems. However, no related studies on the inversion of asynchronous MMC PSA in electromagnetic data have been found. In this paper, the asynchronous MMC PSA algorithm is applied to invert a synthetic model, and we analyzes the detailed performance of the algorithm and the solution quality.