|
Project Information
|
Inspired by nature, transcend the nature We learn from the nature. But for science exploration, we need to do something new. Global Optimization We concerns global optimization (GO) problems, in which one wants to get the globally best solution of an objective function f(x) on a given domain, where f might be multi-modal, non-differentiable, discontinuous, or even worse black-box type. Difficulties: 1. Derivative-based methods are easily get trapped into local optimum. 2. Derivative of objective function is unavailable, or hard to compute, or unreliable (numerically unstable, e.g., in the presence of noise). So derivative-free methods, including evolutionary algorithms, are preferred. Some efficient evolutionary algorithms for GO: 1. Real-coded genetic algorithm (RGA) 2. Genetic programming (GP) 3. Particle swarm optimization (PSO) 4. Differential evolution (DE) 5. Low dimensional simplex evolution (LDSE) |