glcCluster is a global optimization algorithm, developed by Prof.Dr. Kenneth Holmström. It is a hybrid algorithm, combining the DIRECT global optimization algorithm, a clustering algorithm and local search. glcCluster is using one of the following DIRECT algorithms: glcDirect (default), glcFast or glcSolve, for global search (Step 1). Step 2 is an adaptive clustering algorithm to find a suitable number of clusters, where the best point in each cluster is then used as an initial point for a local search (Step 3). The 4th step is to run the DIRECT algoirithm once again, to possibly improve. If the DIRECT algorithm improves the best point, a local search is finally made as Step 5 with the new best point(s) as starting points. The routine glcCluster implements an extended version of DIRECT that handles problems with both nonlinear and integer constraints.
References in zbMATH (referenced in 1 article )
Showing result 1 of 1.
- Rios, Luis Miguel; Sahinidis, Nikolaos V.: Derivative-free optimization: a review of algorithms and comparison of software implementations (2013)