Simulated Annealing
Simulated annealing (SA) is a method of global optimization, and was first developed for solving combinatorial systems such as the travelling salesman problem and chip placement.
The Simulated Annealing search option in Slide2 employs a hybrid simulated annealing (HSA) algorithm for the search of general failure surfaces. The method couples a Very Fast Simulated Annealing algorithm (VFSA) with an efficient searching technique which we will refer to as Local Monte-Carlo (LMC). The VFSA is a state-of-art SA algorithm, relying on a probabilistic random walk and an exponentially decreasing schedule. The LMC relies on local exploration of each vertex, with a step-reduction mechanism as the search approaches the global minimum.
A comparison with other global optimization methods demonstrates that HSA has higher precision and employs significantly less iterations to find the global minimum. The hybrid algorithm couples the robustness of a global optimization algorithm with the speed and refinement of a local optimizer.
The development of the Simulated Annealing search algorithm used in Slide2 is documented in the following paper:
Optimize Surfaces
By default the Optimize Surfaces option is enabled for Simulated Annealing. This applies an additional optimization to the minimum safety factor surface generated by the Simulated Annealing Search and usually results in a lower safety factor. It is recommended that this option is always enabled for Simulated Annealing. See the Optimize Surfaces topic for details.
Convex Surfaces Only
See the Block Search topic for a description of the Convex Surfaces Only option.
Surface Filter
See the Surface Options topic for details about the Minimum Elevation, Minimum Depth, Minimum Area, Minimum Weight slip surface filter options.