Magnetization Reversal by Electrical Spin Injection in Ferromagnetic (Ga,Mn)As-Based Magnetic Tunnel Junctions

We report current-induced magnetization reversal in a ferromagnetic semiconductor-based magnetic tunnel junction (Ga,Mn)As/AlAs/(Ga,Mn)As prepared by molecular beam epitaxy on a p-GaAs(001) substrate. A change in magneto-resistance that is asymmetric with respect to the current direction is found with the excitation current of 10⁶ A/cm². Contributions of both unpolarized and spin-polarized components are examined, and we conclude that the partial magnetization reversal occurs in the (Ga,Mn)As layer having smaller magnetization with the spin-polarized tunneling current of 10⁵ A/cm².     

基于多Agent的混合智能学习算法及

本文提出了基于多Agent的混合智能学习算法，将个体学习和群体学习有效地结合起来，并给出了该算法在RoboCup足球机器人仿真系统中的具体应用．实验结果表明了算法的可行性与有效性．     

一种基于遗传算法的图像矢量量化方法

描述了一种基于遗传算法的图像矢量量化方法。遗传算法基于自然进化原理可得到全局优化结果，把遗传算法用于码本设计，运算简单，而且可避免传统方法如Ｋ－均值技术等所带来的局部优化局限。同时，由于其结构的规则性和并行性，较适于ＶＬＳＩ实现。文中主要讨论了遗传算法用于图像矢量量化的适应度函数及操作过程，并与传统算法作了比较，最后给出了实验结果。     

求解部分可观测马氏决策过程的强化学习算法

针对部分可观测马氏决策过程(POMDP)中，由于感知混淆现象的存在，利用Sarsa等算法得到的无记忆策略可能发生振荡的现象，研究了一种基于记忆的强化学习算法——CPnSarsa(λ)学习算法来解决该问题．它通过重新定义状态，Agent结合观测历史来识别混淆状态．将CPnSarsa(λ)算法应用到一些典型的POMDP，最后得到的是最优或近似最优策略，与以往算法相比，该算法的收敛速度有了很大提高．     

基于JPEG的SAR图像压缩的C语言实现

在对通用软件的提炼、修改基础上，用标准C语言实现基于JPEG的SAR图像数的压缩，实验结果证明相对其它可实现SAR图像压缩的方法，JPEG压缩性能最佳，同时硬件也容易实现。     

在钢筋混凝土结构设计及施工中钢筋工程容易忽视的若干问题及解决对策

钢筋工程是钢筋混凝土工程中很关键的一个分项，钢筋的设计、加工及安装的好坏，直接影响工程的质量。在钢筋工程的设计与施工中，工程技术人员一般研读标准图集不够，或对钢筋工程的设计意图及施工工艺了解不是十分透彻，结果在钢筋的设计、加工及安装中出现许多质量隐患，不能引起工程界的足够重视。章针对钢筋工程在结构主梁与次梁的连接、框架柱与梁及剪力墙的连接、梁与板的连接、柱子纵向钢筋的加工、现浇板负筋的绑扎等部位和钢筋工程经常忽视的若干问题进行了分析，并提出了解决对策，供设计和施工人员参考。     

Examination of the Response of Regularly Packed Specimens of Spherical Particles Using Physical Tests and Discrete Element Simulations

Significant insight into the response of granular materials can be gained by coupling accurately controlled physical tests with complementary discrete element simulations. This paper discusses a series of triaxial and plane strain laboratory compression tests on steel spheres with face-centered-cubic and rhombic packings, as well as discrete element simulations of these tests. The tests were performed on specimens of uniform-sized steel balls and on specimens of steel balls with specified distributions of ball diameters. The packing configurations are ideal and differ considerably from real sand specimens, however, studies of such idealized granular materials can yield considerable insight into the response of granular materials and the capability of discrete element simulations to capture the response. The differences in response for the two packing configurations considered illustrate the importance of fabric. The numerical simulations captured the observed laboratory response well if the particle configurations, particle sizes, and boundary conditions were accurately represented. However, the postpeak response is more difficult to capture, and it is shown to be sensitive to the coefficient of friction assumed along the specimen boundaries. The simulations of the tests on the nonuniform-sized specimens demonstrated a clear correlation between strength and coordination number.     

增强蚁群算法的机器人最优路径规划

为解决复杂环境中机器人最优路径规划问题,本文结合增强学习和人工势场法的原理,提出一种基于增强势场优化的机器人路径规划方法,引入增强学习思想对人工势场法进行自适应路径规划.再把该规划结果作为先验知识,对蚁群算法进行初始化,提高了蚁群算法的优化效率,同时克服了传统人工势场法的局部极小问题.仿真实验结果表明,该方法在复杂环境中,对机器人的路径规划效果令人满意.