Support vector machine (SVM) is one of the most widely used algorithms in parameter inversion of rough surface. However, the penalty parameters (C) and kernel function parameters (G) in SVM affects accuracy of results. If the parameters are not used properly, the model will lead to "over learning"or "less learning", which reduces greatly prediction accuracy. Several optimization algorithms for C and G of SVM are shown, such as K-fold cross validation (K-CV), genetic algorithm (GA) and particle swarm optimization (PSO). An improved PSO algorithm based on K-CV and GA (GA-CV-PSO) is proposed. The training set and test set are constructed with rough surface backscattering coefficient obtained by the moment method (MoM). Inversion precision and calculation time of different optimization algorithms are compared. It shows that GA-CV-PSO algorithm overcomes shortcomings of single optimization algorithms, with more accurate inversion precision and stronger generalization ability.

A frequency domain electromagnetic algorithm boundary element method is applied for computation of Casimir forces between arbitrary materials with arbitrary geometry.Considering electric and magnetic surface current distributions,Casimir force of two objects in terms of interactions of surface currents is obtained.Casimir effects between dielectric objects embedded in dielectric fluid are presented and numerical conditions of repulsive Casimir force are investigated.Non-zeropoint energy Casimir force calculation method is provided.It can be used for design of realistic MEMS.

Euler-Hamilton equations are provided using Hamiltonian function of Maxwell's equations.High order symplectic schemes of three-dimensional time-domain Maxwell's equations are constructed with symplectic integrator technique combined with high order staggered difference.The method is used to analyzing stability and numerical dispersion of high order time-domain methods and symplectic schemes with matrix analysis and tensor product.It confirms accuracy of the scheme and super ability compared with other time-domain methods.

Multi-step high-order finite difference schemes for infinite dimensional Hamiltonian systems with split operators are proposed to solve time domain Maxwell's equations. Maxwell's equations are discretized in time direction and in spatial direction with different order of sympleetie schemes and fourth-order finite difference approximations, respectively. Stability analysis and numerical results are presented. A five-stage fourth-order multi-step high-order finite difference scheme proves accurate and efficient, and applicable to long time simulations.