An algorithm for the generation of PL integral operator is presented. This algorithm provides a virtually trivial determination of the operator elements. Sparseness and decomposition of the integral operator are also discussed.

By use of a planar grating on a dielectric slab where some strips of different width are placed in one period, polarization- and frequency-selectivities are improved when the thickness of the dielectric slab, its relative permitivity and the location of the strips are chosen appropriately.

Reflection and transmission when a Gaussian beam is launched into an anisotropic slab whose optical axis has arbitrary direction is described. This problem is solved by using Fourier transforms. Solutions of reflected and transmitted beams having integral form are numerically calculated by using FFT.

The problem is deduced to scattering from four conducting rectangular cylinders by using the electric image method and analyzed by Point Matching Method. This letter gives not only near field distribution but also model experiments using a line source at 9.6 GHz. Numerical result is in nearly good agreement with the experimental result.

The spatial network method can analyze a three-dimensional electromagnetic field, including time response. But, no paper has discussed a matching condition of a microstrip line in detail. In this paper, we have clarified a generalized procedure to obtain the maching condition using resistor and capacitor at the terminal plane of the microstrip line.

A new method is presented for designing robust stabilizing controllers for SISO plants with any bounded parameter and/or degree variations in some known ranges It is shown that, if the plants are of minimum phase and have (any) fixed relative degrees and gains with fixed sign, then, linear time-invariant controllers can always be designed to stabilize robustly such plants. The order of a controller designed by this method is equal to the relative degree of the plant, and is independent of the order of the plant.

This paper presents and effective acceleration technique for the homotopy method using a rectangular subdivision (which will be called the rectangular algorithm" in this paper). The rectangular algorithm is an efficient solution algorithm that is globally convergent for a large class of nonlinear resistive networks. In the rectangular algorithm, the restart technique is usually used in order to improve the accuracy of the solution. In this paper, we show that certain realization of the restart technique achieves local quadratic convergence. In other words, if we determine the grid size of the rectangular subdivision pertinently in each restart, the sequence of the approximate solutions generated by the algorithm converges to the exact solution quadratically. And in this case, the computational work involved in each restart is almost identical to that of Newton's method. Therefore, our algorithm becomes as efficient as Newton's method when it reaches sufficiently close to the solution.

A new type of phase comparator for DPLL (Digital Phase-Locked Loop), named Adaptive Multilevel-Quantized Phase Comparator (abbreviated as AMPC), is proposed. The characteristics of this proposed comparator AMPC are changed adaptively to reduce the frequency deviation and the phase jitter of the input signals, whereas the conventional phase comparator has constant characteristics whatever signals come. When the offset between input and output signal frequency exists, the amount of frequency control is increased by shifting up or down its characteristics in order to decrease this deviation. When the loop is in the steady-state, the amount of phase control is decreased by varying the scale of them to suppress the input jitter. Performance characteristics of AMPC and the loop which uses AMPC are analyzed theoretically and verified by computer simulation. As a result, the loop which uses AMPC has much wider locking-range and much better jitter suppresion effect than those of the conventional loops, and steady-state phase error is also reduced by using AMPC.

This paper concerns the study on the Ku band multi-beam direct radiating array antenna technology, which provides frequency reuse and reconfigurable multi-beam capabilities in satellite communication system. The study consists of two dimensional pattern synthesis, beam forming network analysis and design, and fabrication and measurement of an experimental model. The measurement results have shown that the experimental model with 64 array elements including 14 dummy elements has a low loss of less than 2 dB in average and high sidelobe and polarization isolation of more than 27 dB and 29 dB, respectively, over 1 GHz bandwidth in the Ku band. As a result of careful evaluation, it has been confirmed that the design method established here is valid and useful for multi-beam direct radiating array antenna design.

Degradation of the two-dimensional images transmitted through a graded-index (GRIN) optical fiber is evaluated by considering the fourth and sixth order terms of the refractive index distribution, and the spatial inverse filter and Wiener filter are investigated in order to restore the degraded images. When a point image is transmitted through a practical GRIN fiber, the output image is degraded by the modal dispersion, i.e. the phase differences between the modes. This degraded image at the output plane of the GRIN fiber is restored by the spatial inverse filter and Wiener filter which are designed so that the filtered coherent transfer function of the GRIN fiber has almost flat spatial frequency spectrum. The transform function of the inverse filter is calculated by using only the on-axis point spread function (PSF) of the fiber and is found to be effective in a region near the propagation axis, because the fiber is space-invariant around the axis. By considering the difference between the on-axis and off-axis PSFs, the transfer function of the optimum Wiener filter is obtained. The optimum Wiener filter restores the off-axis PSF as well as that of the on-axis. The waists of the main peaks at which the intensities of the filtered on-axis and off-axis PSFs become 1/e² of their maximum values, are 0.75 and 0.52 of those without filtering. The resolution of the PSFs corresponding to the point images transmitted in the range of 0-25 µm are increased. The levels of the small peaks in the field intensity distributions of the filtered on-axis and off-axis PSFs are also reduced and these peaks are low enough to obtain high contrast images at the output for the considered range.

Bennett proved that any irreversible Turing machine can be simulated by a reversible one. However, Bennett's reversible machine uses 3 tapes and many tape symbols. Previously, Gono and Morita showed that the number of symbols can be reduced to 2. In this paper, by improving these methods, we give a procedure to convert an irreversible machine into an equivalent 1-tape 2-symbol reversible machine. First, it is shown that the state-degeneration degree" of any Turing machine can be reduced to 2 or less. Using this result and some other techniques, a given irreversible machine is converted into a 1-tape 32-symbol (i.e., 5-track 2-symbol) reversible machine. Finally the 32-symbol machine is converted into a 1-tape 2-symbol reversible machine. From this result, it is seen that a 1-tape 2-symbol reversible Turing machine is computation universal.

A sequence α(a₁,,a_n) is k-sorted if and only if α satisfies the condition: for all i, j in {1,,n}, ijk implies a_ia_j. A local characterization of k-sorted sequences is given. An algorithm for sorting k-sorted sequences is designed, and its time efficiency is analysed. For each k1, a lower bound on the number of comparisons needed to sort k-sorted sequences is obtained by an estimate of the number of k-sorted sequences of length n and a decision tree argument. We show that 0.6947 n and 1.1265 n are average case lower bounds on the numbers of comparisons for sorting 1-sorted and 2-sorted sequences of length n, respectively.

First, we give a method by which any k-valued logic function with n variables can be realized with a cellular array of (k1)k^n-1 rows and k^n-1 columns. Next, by encoding multi-value to binary one and decoding binary value to multivalue, we give another method by which any k-valued logic function with n variables can be realized with log₂k cellular arrays each of which is of k^n-1 rows and k^n-1 columns.