自动微分方法在XIAMEN软件优化中的应用

比起有限差分方法来,运用自动微分方法计算函数的梯度在计算时间和计算精度方面都具有明显的优势.使用伴随模式计算函数的梯度,在XIAMEN软件优化中得到了明显的加速效果.使用ADG系统自动生成伴随模式,大大降低了伴随模式的开发时间和难度.重点讨论了伴随模式实现的几个关键难题,并给出了几个典型应用的数值结果.     

基于OpenCL的自动微分并行实现及其应用

针对如光束平差这样的大规模优化问题,实现基于OpenCL的并行化自动微分。采用更有效的反向计算模式,实现对多参数函数的导数计算。在OpenCL框架下,主机端完成C/C++形式的函数构建以及基于拓扑排序的计算序列生成,设备端按照计算序列完成函数值以及导数的并行计算。测试结果表明,将实现的自动微分应用于光束平差的雅可比矩阵计算后,相比于采用OpenMP的Ceres Solver,运行速度提高了约3.6倍。     

微分对策理论及其应用研究的新进展

微分对策理论是控制论和对策论的重要分支，在军事对策和经济学研究领域具有非常广泛而重要的应用．对此简述了半个世纪以来微分对策理论和应用发展的历史，介绍了近年来微分对策理论发展的最新研究成果，并对微分对策理论发展的几个热点问题作了简要评述．     

基于自动微分的反应动力学程序设计

针对ReaxFF反应力场,开发了一套基于PyTorch的计算软件,功能上实现了直接读取分子动力学轨迹文件并计算分子运动模拟过程中的原子所受到的势能函数等能量项。将这些能量项在分子运动模拟各个时刻的值直观地展现在科研人员面前。使用GULP也同样地计算了这些势能函数,经过对比可以看到与经GULP计算后绘制的势能函数图像重合,证明了代码是准确且可靠的。并且引入自动微分技术,使得ReaxFF可以与神经网络结合进行优化。     

用C语言实现微分运算精确解

用C语言实现微分运算精确解的算法和程序结构,并给出了实例。     

自动微分在隐式曲线绘制中的应用

自动微分是用于计算多变量函数的导数和偏导数的一种微分技术,在给定一个多变量光滑函数值的程序代码后,可以很容易地利用自动微分来实现有关导数和偏导数的精确计算。将自动微分技术与泰勒方法相结合应用到计算机图形学领域隐式函数曲线绘制的细分算法中,并与未使用自动微分技术前的隐式曲线绘制方法作比较和分析,展示了自动微分方法在绘制隐式曲线方面的优势。     

双闭环不完全微分PID自动调速系统仿真研究

本文分析了双闭环调速系统各个环节传递函数,详细讨论了应用不完全微分PID设计系统的速度控制器的方法．并在matlab／simulink中对系统进行了仿真,得到了很好的仿真结果。     

C语言源程序的自动评判系统

自动评判系统为通过编译C语言源程序生成的可执行程序创建进程,输入适当的测试数据,对输出结果与预设答案进行比较和评判。提出了通过匿名管道进行数据输入输出的实现方法,分析了自动评判机制中的主要问题及解决方法,重点对分级评判进行了探讨,进而提出了四级评判机制。     

广义动态系统的微分算子矩阵方法

本文用微分算子矩阵方法讨论了一类广义动态系统的结构性质,并且用微分算子多项式矩阵的系数阵,给出了各类动态系统能控(观)性的直接判据.它是用微分算子矩阵方法研究广义动态系统的开端.     

C语言作业自动评测系统实现

C语言程序设计是一门公共必修课,课程作业多,教师批改压力大,且易受主观影响,造成较大误差。因此,使用PHP技术与MYSQL数据库,构建一个基于B/S结构的C语言作业自动评测系统,利用相似度匹配算法与正则表达式的结合实现程序的过程评分,加上编译评分与结果评分,最终实现程序作业的自动评分。     

自动微分转换系统及其应用

§1.引言 计算微分大致经历了从差商微分,符号微分,手写代码到自动微分几个阶段,与其它几种微分方法相比,自动微分具有代码简练、计算精度高及投入人力少等优点。自动微分实现     

基于神经网络的实时专家控制系统及其PTA工业应用

以精对苯二甲酸结晶过程为研究对象。提出一种基于神经网络模型的实时专家控制系统．该方法利用神经网络建模技术获取对象的机理知识,通过对影响模型特性的多个变量进行分析,自动得到常规专家控制系统难于获取的定性、定量知识,并按分级递阶的启发式搜索机制,实现了对工业过程对象的实时控制．实际应用表明：该方法不但克服了以往专家系统知识获取的瓶颈,而且有效实现了人机对话的功能,便于现场操作和更改专家知识库,为化工过程的多变量控制提供了新的思路．     

Automatic Differentiation and Interval Arithmetic for Estimation of Disequilibrium Models

Nonlinear estimation problems have a unknown number of stationary points. Interval arithmetic is a promising method that eliminates all but the global optimum. Automatic differentiation provides users with a convenient method of computing the gradient and Hessian of nonlinear functions. These two can be combined to provide an efficient and convenient global optimization process.     

Model adjointization and its cost

1 Introduction With great development of observation and sampling techniques, the research of inverse problems, e.g., data assimilation and remote sensing inversion, is becoming a frontier and focus in some disciplines such as atmospheric sciences and oceanography. Most optimization methods, which depend on first or higher order derivatives of consid-ered functions, are efficient ways to solve such large-scale inverse problems. Compared with other differentiation techniques, e.g., divided dif…     

Adaptive hinging hyperplanes and its applications in dynamic system identification

The model of adaptive hinging hyperplanes (AHH) is proposed in this paper. It is based on multivariate adaptive regression splines (MARS) and generalized hinging hyperplanes (GHH) and shares attractive properties of the two. By making a modification to the basis function of MARS, AHH shows linear property in each subregion. The AHH model is actually a special case of the GHH model, which has a universal representation capability for continuous piecewise linear functions. The approximation ability of the AHH model is proved. The AHH algorithm is developed similar to the MARS algorithm. It is adaptive and can be executed efficiently, hence has power and flexibility to model unknown relationships. The AHH procedure is applied to identifying two dynamic systems and its potential is illustrated.     

Comparison of Automatic and Symbolic Differentiation in Mathematical Modeling and Computer Simulation of Rigid-Body Systems

The objective of this paper is to check the efficiency and validity oftwo approaches for computing derivatives of complex functions,automatic differentiation using ADOLC and symbolicdifferentiation using MACSYMA. This has been done in three benchmarkexamples, where the gradient of a Helmholtz energy function has beencomputed for different dimensions of independent variables (Example 1)and Jacobian matrices of inverse kinematics of planar and spatialparallel robots (Examples 2 and 3) have been computed. The results havebeen evaluated under six criteria: preliminary implementation work,computation time, flexibility in applications, limits of applicability,accuracy, and memory requirements.ADOLC was superior to MACSYMA concerning preliminarywork (programming, source code generation, and compilation) andmodifications of the functions to be differentiated and thedifferentiation task to be performed. In addition, contrary toMACSYMA, no limits of applicability were observed forADOLC, even in the simulation of complex multi-body systems.On the other hand, for ADOLC the computation time of derivatives was 10 to 40 times higher than for MACSYMA. As aconsequence, differentiation by MACSYMA is better suited forreal-time applications like hardware in the loop simulation, real-timecontrol and real-time data processing than ADOLC.Both programs provide numerical results of equal accuracy.     

勾股模糊H平均算法应用于多媒体图像系统选择

提出了一种勾股模糊H平均算法应用多媒体图像系统选择问题。首先,定义了基于t-模和t-余模的勾股模糊数运算;讨论了勾股模糊Heronian平均算法的三个特征性质和经常使用的特例;然后构建了新的勾股模糊决策模型,该模型在构建过程中能够挖掘输入数据之间关联性,还提高了决策的使用范围;最后,将构建的模型应用于多媒体图像系统选择案例来验证有效性。     

Information Systems Strategy and its Relationship With Innovation Differentiation and Organizational Performance

ABSTRACT

This study, from the resource-based view, analyzes the different typologies of information systems (IS) strategy and their effects on firms’ innovation differentiation strategy and performance. Distinguishing between innovative and conservative IS strategy, results from the Spanish food industry show that an innovative IS strategy has a stronger direct effect on organizational performance, but that a conservative IS strategy can be a safer way to support the innovation differentiation strategy of firms.