Robust analysis and global optimization (Q5919267): Difference between revisions

The author considers the minimization of a lower semicontinuous function f: \(S\to R\) over a compact subset S of a topological space X endowed with a suitable measure. The global minimum value \(\bar c\) is characterized through the mean values of f over some of its level sets and an algorithm is proposed for finding \(\bar c.\) The central hypothesis states that f is relatively robust to S at every global minimum point, i.e., every global minimum point belongs to \(cl(int(F_ c))\) for all \(c>\bar c\) where \(F_ c=\{x\in S\); \(f(x)<c\}\). Numerical tests are also provided.