Adaptive multiple super fast simulated annealing for stochastic microstructure reconstruction

Abstract

Fast microstructure reconstruction from statistical information is critical to create high resolution image with large domain for materials property prediction. Large structural data generated from multimodality chemical imaging instrumentation also calls for fast microstructure reconstruction for evaluation and interpretation. Stochastic methods in microstructure reconstruction from two-point correlation function have been applied widely in materials and mechanical engineering community. The main challenge is to increase the efficiency of large data reconstruction and reconstruction. A novel simulated annealing method, adaptive multiple super fast simulated annealing, is proposed for fast solution to improve efficiency. Combining the advantage of very fast cooling schedules, dynamic adaption and parallelisation, the new simulation annealing algorithm increases the efficiencies by several orders of magnitude, making the large domain image fusion feasible.