Metaheuristics Based on Sciences
This chapter introduces dozens of metaheuristic optimization algorithms that are related to physics, natural phenomena, chemistry, biogeography, and mathematics.
- PDF / 254,513 Bytes
- 20 Pages / 439.37 x 666.142 pts Page_size
- 93 Downloads / 235 Views
18
This chapter introduces dozens of metaheuristic optimization algorithms that are related to physics, natural phenomena, chemistry, biogeography, and mathematics.
18.1 Search Based on Newton’s Laws Gravitational Search Algorithm Gravitational search algorithm [45] is a stochastic optimization technique inspired by the metaphor of the Newton theory of gravitational interaction between masses. The search agents are a collection of objects having masses, and their interactions are based on the Newtonian laws of gravity and motion. The force causes a global movement of all objects toward the objects with heavier masses. Hence, masses cooperate through gravitational force. Each agent represents a solution. Each agent has four specifications: position, inertial mass, active gravitational mass, and passive gravitational mass. The position of an agent corresponds to a solution of the problem, and its gravitational and inertial masses are determined using a fitness function. Heavy masses, which correspond to good solutions, move more slowly than lighter ones; this guarantees the exploitation step of the algorithm. The algorithm is navigated by properly adjusting the gravitational and inertia masses. Masses are attracted by the heaviest object, corresponding to an optimum solution in the search space. Gravitational search algorithm is somewhat similar to PSO in the position and velocity update equations. However, the velocity update is based on the acceleration obtained by the gravitational law of Newton. Consequently, position of each agent is updated using the modified velocity. The gravitational constant adjusts the accuracy of the search, so it speeds up the solution process. Furthermore, gravitational search algorithm is memoryless, it works efficiently like algorithms with memory, and it can be considered as an adaptive learning algorithm. © Springer International Publishing Switzerland 2016 K.-L. Du and M.N.S. Swamy, Search and Optimization by Metaheuristics, DOI 10.1007/978-3-319-41192-7_18
295
296
18 Metaheuristics Based on Sciences
In gravitational search algorithm, the algorithmic gravitational forces lead directly to changes in the position of search points in a continuous space. In most cases, gravitational search algorithm provides superior or at least comparable results with PSO and central force optimization [23]. Gravitational search algorithm is easier to implement in parallel with Open-MP compared to central force optimization. In binary gravitational search algorithm [46], trajectories are changes in the probability that a coordinate will take on a zero or one value depending on the forces. Artificial Physics Optimization Inspired by the second Newton’s force law, artificial physics optimization [60] is a stochastic population-based global optimization algorithm. Each entity is treated as a physical individual with attributes of mass, position, and velocity. The relationship between an individual’s mass and its fitness is constructed. The better the objective function value, the bigger is the mas
Data Loading...
