Search

Relaxation oscillations in [Er.sup.3+]-doped and [Yb.sup.3+]/[Er.sup.3+] co-doped fiber grating lasers were investigated. Intensity noise models were established for both cases in which the fluctuations of pump power and cavity loss were analyzed. Simulation indicates that the relaxation oscillation is induced by the pump power fluctuation, and fluctuation of the cavity loss can broaden the relaxation oscillation peak and enhance the amplitude. The relaxation oscillation of a laser with [Er.sup.3+]-doped fiber exhibits a lower frequency and larger amplitude than that of an [Yb.sup.3+]/[Er.sup.3+] co-doped fiber grating laser. Distributed Bragg reflector fiber grating lasers with [Er.sup.3+]-doped fiber and [Yb.sup.3+]/[Er.sup.3+] co-doped fiber were fabricated, and the relaxation oscillations were measured. For the former, the relaxation oscillation frequency is at the magnitude of several hundreds of kilohertz and lineally shifts toward the higher frequency with the pump power. For the latter, no obvious relaxation oscillation was measured. OCIS codes: 060.2310, 060.2370, 060.2340.

Support vector machines (SV machines, SVMs) have many merits that distinguish themselves from many other machine-learning algorithms, such as the nonexistence of local minima, the possession of the largest distance from the separating hyperplane to the SVs, and a solid theoretical foundation. However, SVM training algorithms such as the efficient sequential minimal optimization (SMO) often produce many SVs. Some scholars have found that the kernel outputs are frequently of similar levels, which insinuate the redundancy of SVs. By analyzing the overlapped information of kernel outputs, a succinct separating-hyperplane-securing method for pruning the dispensable SVs based on crosswise propagation (CP) is systematically developed. The method also circumvents the problem of explicitly discerning SVs in feature space as the SVM formulation does. Experiments with the famous SMO-based software LibSVM reveal that all typical kernels with different parameters on the data sets contribute the dispensable SVs. Some 1% ~ 9% (in some scenarios, more than 50%) dispensable SVs are found. Furthermore, the experimental results also verify that the pruning method does not alter the SVMs' performances at all. As a corollary, this paper further contributes in theory a new lower upper bound on the number of SVs in the high-dimensional feature space. Index Terms--Architectures, crosswise propagation (CP), high-dimensional feature space, pruning, support vector machines (SVMs), upper bound.

With the ability to understand linkages and feedbacks between land use dynamics and human–land relationships, cellular automata (CA) are extensively applied in land use/cover change (LUCC) simulation. However, with complex transition rules and a growing volume of spatial data, conventional serial CA models cannot meet the demands of efficient computation. In this article, a Tensor‐CA model using vectorization and Graphics Processing Unit (GPU) technology based on a tensor computation framework for optimizing multiple LUCC simulations is presented. Complex transition rules of LUCC‐CA models are vectorized and formalized to tensor operations which are effectively solved by GPU. The proposed Tensor‐CA model was applied to LUCC simulations in the Pearl River Delta of China. The experimental results indicate that the proposed model effectively improved the performance compared to Serial‐CA, Parallel‐CA, and GPU‐CA. [ABSTRACT FROM AUTHOR]