In order to find the plural number of optimal solutions for the module placement problem, which is of particular significance in the design of VLSI, a new Genetic Algorithm (GA) based on Imanishi's evolutionary theory is proposed. First of all, a phenotypic distance between two solutions, or individuals, is defined as the shortest Hamming distance between sets of isomorphic genotypes. If the phenotypic distance between two individuals is shorter than a threshold value, they are regarded as the same species. Then, a new generation alternation model that is analogous to the habitat segregation is presented. Since each individual in the population represents an isolating species, the diversity of the population is maintained spontaneously. Even though the selection of individuals based on their fitness is not employed, excellent individuals are created effectively by using the harmonic crossover operation combined with a local optimization method.
This paper addresses a sort of non-convex relaxation problem for the minimization problem of a quadratic function. We define relaxation problems by generalizing the feasible region of the original problem to the space consisting of hypercomplex numbers. Computational experiments for 0-1 quadratic minimization problems reveal the effectiveness of two proposed algorithms based on the derived properties. Fundamental properties of the relaxation problem for a more general class of quadratic minimization problems are also discussed.
A design procedure for interpolators is developed based on sampled-data control theory. The procedure provides an interpolator which minimizes the L2-induced norm of the error system between the interpolator and a time-delay, and the L2/l2-induced norm of the system between the quantization noise and the output of. the interpolator. While the system is multirate and has delay elements, the design problem can be reduced to a finite-dimensional discrete-time problem using the FSFH (fast-sample and fast-hold) approximation. Numerical examples are presented to illustrate the effectiveness of the proposed method.
In this paper, an effective tracking method for a maneuvering target moving at unknown variable speed on the 2-dimensional plane is proposed. First, dynamics are derived for the target which moves at variable speed by introducing the first-order dynamics for the target's jerk motion as a modified Singer model. Next, in order to get highly accurate tracking, the kinematic constraints are incorporated into measurement equation as pseudomeasurements. Based on the extended Kalman filter algorithm, the tracking filter is derived from the linear dynamics and the augmented nonlinear measurement equation. The efficacy of this method is shown by simulations using real data.
A congestion-driven placement technique for LSI cells through hybrid genetic algorithm is presented. In particular, the procedure consists of a two-level hierarchical placement procedure and a hybrid genetic algorithm, in which the algorithm is combined with the approach of searching locally for an optimal solution. For selection control, new objective function are introduced to each phase for dispersing congestion, and a parallel processing technique suited to hierarchical placement is proposed as an effective approach to accelerating the processing speed of genetic algorithm-based placement. Regarding to total virtual wire length and wire congestion, the ratio of the proposed approach to the conventional one is 0.76 and 0.9, respectively. As a result, the effectiveness of the suggested approach is shown through a comparison with the conventional one.