Acceleration Techniques For Light Ray Optimization Algorithms

The proposed LRO algorithm is based on the principle proposed by Fermat: the light between two points always propagates along the path that takes the shortest time.The algorithm simulates the propagation of light in different media.The function value is used as the velocity of the light.The feasible domain is divided into a number of rectangular small meshes according to a given step size,and is filled in each rectangular small mesh according to the function value.Uniform medium,given the initial point and initial direction randomly,the light is locally optimized by refraction in the feasible domain,and global optimization search is performed by reflection.This paper first proves the four theorems of the LRO algorithm.The theorem reflects some problems in the search,and then proposes an acceleration method for these problems.The acceleration method includes parameter adaptation and initial direction optimization.Through analysis,the paper gives the form of adaptive parameters,the form of initial direction optimization parameters,and the flow of the improved algorithm.Parameter adaptation is to use the incident velocity and the reflection velocity as independent variables to obtain the dependent variable parameter.The parameter adjusts its own size according to the change of the function value,and finally achieves the purpose of reducing the reflection condition.The optimization of the initial direction uses Boolean algebra to control the reflection by Boolean algebra.The condition is such that when the initial direction becomes larger,the direction of the search function value becomes smaller by reflection,thereby optimizing the initial direction.The addition of the two methods reduces the number of iteration steps to a certain extent,and solves the problem that the deflection angle becomes smaller and the reflection rarely occurs when the maximum value of the function is maximum,thereby achieving the number of steps in fewer iterations.Internally,achieve global convergence.Again,the paper proposes an efficiency estimation function,which is used to estimate the search success rate of the improved algorithm in the case of reducing the number of iteration steps.Finally,the paper verifies the effect of the improved algorithm through numerical simulation experiments.By comparing the time taken by the improved algorithm to achieve global convergence,the iterative path of light,and the histogram distribution histogram required to achieve global convergence,it is intuitively displayed.The effect of the improved algorithm.Finally,the numerical simulation experiment is used to verify the search effect of the efficiency estimation function based on the improved algorithm after reducing the number of iteration steps.