×

Fast level set based algorithms using shape and topological sensitivity information. (English) Zbl 1167.49317

Summary: A framework for descent algorithms using shape as well as topological sensitivity information is introduced. The concept of gradient-related descent velocities in shape optimization is defined, a corresponding algorithmic approach is developed, and a convergence analysis is provided. It is shown that for a particular choice of the bilinear form involved in the definition of gradientrelated directions a shape Newton method can be obtain. The level set methodology is used for representing and updating the geometry during the iterations. In order to include topological changes in addition to merging and splitting of existing geometries, a descent algorithm based on topological sensitivity is proposed. The overall method utilizes the shape sensitivity and topological sensitivity based methods in a serial fashion. Finally, numerical results are presented.

MSC:

49M15 Newton-type methods
49Q10 Optimization of shapes other than minimal surfaces
PDFBibTeX XMLCite
Full Text: EuDML