The Pointer-2 strategy controls the following set of optimization techniques:
Each optimization technique succeeds or fails for different topography features, and a hybrid combination of these methods allows a pointer optimization to solve a broad range of design problems. The pointer strategy can control one technique at a time or all techniques at once. As the optimization proceeds, the strategy determines which techniques are most successful and suitable values for the internal control parameters (e.g., step sizes, numbers of iterations, number of restarts, etc.). In addition to running optimization techniques, the Pointer-2 strategy periodically generates internal approximations using data points taken from the optimization history. The Pointer-2 strategy performs an optimization on the approximation and verifies the optimum with an exact run. This approach can accelerate the convergence of the Pointer-2 strategy for design spaces that can be approximated accurately. The Pointer-2 strategy allows you to select a single technique that is used instead of the hybrid combination. You may want to select a single technique if you have successfully run an optimization with a particular technique, and you want to improve its performance. If the Pointer-2 algorithm gets stuck and fails to produce a better solution over a large number of simulations, the algorithm makes increasingly larger and larger random jumps from the current local optimum in an attempt to get unstuck. In that sense, Pointer-2 never gives up and never assumes it has converged. The Pointer-2 technique stops only when it reaches its maximum number of allowable simulations. The Pointer-2 technique is superior to the Pointer technique, especially early on in the optimization, and you should always use the Pointer-2 technique over the Pointer technique. The Pointer-2 technique performs the optimization significantly faster than the Pointer technique because it can take advantage of the parallel execution of multiple CPUs; whereas, the Pointer technique can submit only one subflow to a single CPU at a time. (Even when the optimization submits only one subflow at a time, you can still expect superior performance from the Pointer-2 technique.) The original Pointer technique is retained only for compatibility with older Optimization Process Composer models. Because of its flexibility over a wide range of problems and speed at which it reaches a solution, Pointer-2 is the default optimization technique. | |||||||||