ADAPTIVE SEARCH IN QUASI MONTE-CARLO OPTIMIZATION

Motivated by a linear time-complexity result for an adaptive Monte Car lo algorithm, we propose and analyze an adaptive deterministic algorit hm. We restrict a grid search to nested subregions that promise to pro vide improvement of the current solution, and we obtain an exponential rate of convergence in function evaluations. For proving the main res ult we restrict ourselves to functions on hypercubes. In a final secti on we outline how to extend the method to the general case and give so me numerical examples.