Numerical integration of Lagrange's top is generally not executed without deviation of the first integral value. In this paper, a new method that the deviation of Hamiltonian is fed back negatively to state variables θ, ρ_θ is proposed. Each feedback quantity is proportional to the corresponding partial derivative of the first integral. Simulation results show that the integration deviation of the first integral keeps zero neighborhood by the method.

抄録全体を表示

  The FDTD method is considered to be especially effective because we should consider many types of reflection and diffraction caused by furniture and walls as well as human bodies in the structure, The numerical method makes it possible to consider the entire building, including human bodies, rooms and corridors. However, most studies consider building structures without human bodies and furniture since many parameters should be considered in calculations. On the other hand, several studies have been carried out on indoor radio propagations with limited number and shaped phantoms. The phantoms are simplified cylindrical or rectangular models. However, there is no discussion on the differences in propagation loss characteristics depending upon the shapes of phantom model. In this paper, we consider the three models of phantom, a rectangular, an elliptical cylinder or a detailed shape model. Propagation loss characteristics in corridors are estimated for several space occupancy ratios and relative placements as parameters by calculations employing the FDTD technique. In addition, we discuss effects of phantom models to the propagation loss characteristics. A design of 2.4 GHz wireless LAN based on the IEEE802.11n standard with a high data transfer speed and an extensive coverage is a target of the study.

抄録全体を表示

  The standard Eisenstat preconditioning technique has been discussed based on the amount of computa-tional cost. Sometimes, however, the actual computational time differs from the theoretical estimated time. In this paper, first we propose practical implementation of Eisenstat preconditioning. Second we describe validity of estimation based on the memory access in place of estimation of theoretical computational cost. Through numerical experiments, we verify effectiveness of our discussion based on memory access for Eisenstat preconditioning.

抄録全体を表示