New strategies in differential evolution - Design principle

Differential Evolution, a quite recent evolutionary optimization algorithm, is gaining more and more popularity among evolutionary algorithms. Proposed as a method for the global continuous optimization, Differential Evolution has been easily modified for mechanical engineering purposes and for handling nonlinear constraints. In this paper we introduce a new type of strategies which increase stability of the algorithm reducing its computational expenses. Also we propose a new principle of strategies' design. Theoretical discussions lead us to a tradeoff that helps to choose the better strategy. The strategies are illustrated, tested and compared on a classical test suite. We present a part of the testing results.