A Statistical Mechanics Approach to Optimization Problems
17 October 1986
The analogy between combinatorial optimization problems and systems with many degrees of freedom and competing interactions studied in statistical mechanics has led to the use of simulated annealing to find approximate solutions to complex optimization problems. An alternative approach uses highly interconnected networks of nonlinear analog neurons to compute good answers on a short time scale. The application of both techniques to a classic example of combinatorial optimization, the traveling salesman problem, illustrates the differences and similarities between the two methods.