Random search methods, including simulated annealing and genetic algorithms, are popular search methods for solving global optimization problems with no known structure to the function. While random search methods are relatively easy to implement, their performance is difficult to analyze, and setting algorithmic parameters so that the samplings converge to the optimum effectively and efficiently remains challenging. This talk will summarize some theoretical results regarding performance, and then discuss a new meta-control methodology that adaptively guides the temperature parameter of an interacting-particle algorithm to achieve desired performance characteristics (e.g., quality of the final outcome, algorithm running time, etc.). Instead of selecting a cooling schedule a priori, the meta-control methodology dynamically heats and cools the temperature based on observed behavior of the algorithm. An application in engineering design of composites structures for aircraft fuselage, such as the new 787 Boeing composite aircraft, will be mentioned.