Simulated Annealing and Stochastic Machines

Abstract

Since the publication of the seminal paper by Kirkpatrick, Gelatt Jr. and Vecchi [98], who were inspired by an earlier work on statistical mechanics [108], simulated annealing has been embraced as’ the “savior” for solving many difficult combinatorial optimization problems, and has been applied to many areas such as computer-aided design of very large scale integrated (VLSI) circuits [100], [146], [164], image processing and computer vision [18], [159], [170], telecommunications [15], [37], [51], [55], [139], [165], [166], [167], [168], economics [63], [171], and other engineering and scientific disciplines [104], [115], [133], [152], [157]. A new class of computing machines known as stochastic machines has also emerged through the application of the annealing concept to neural networks. This chapter provides an overview on simulated annealing and the derivation of various stochastic machines. Key properties of simulated annealing are highlighted, and readers are referred to [2], [58], [59], [123], [163], for further detailed analyses.