ALGO2-2

在visual studio的平台上开发CUDA程序实现并行加速计算矩阵的乘积，同时比较与串行计算的差异(Differences in the development of procedures to achieve parallel CUDA accelerated computing matrices in visual studio platform, but compared with the serial computing)

自己写了一个二维Euler方程的间断有限元程序 上次发了一个三角形单元的程序 因为不是曲边单元 所以在圆柱后面容易形成涡 现在把程序改为曲边四边形单元了 没有涡出现 单元是8节点四边形单元 节点编号顺序是 1 5 2 6 3 7 4 8 也就是四个角点依次 是1 2 3 4 然后是边的中点编号 5 6 7 8. 时间推进采用 Runge-Kutta 方法 数值通量采用全局Lax-Friedrichs通量 仍然不能捕捉激波 因为没有做重构或者加人工粘性 等这个做出来了 再发一次。 程序没有进行优化 比如说内存的消耗没有优化 比如直边单元的边界积分仍然采 用了曲边的积分方法 增加了计算量 比如面积分、线积分都是采用的是Gauss- Legendre-Lobatto积分 积分精度会比一般的Gauss-Legendre积分精度低一阶 等 等问题。 二维的 纯属交流性质 就没有考虑这些问题 ^_^ 如果物面全部是直边 那么只要改变一个参数N 就可以获得不同的计算精度 且具 有谱精度 因为单元的节点是Gauss-Legendre-Lobatto积分点。 其实就是谱元法 （物面是曲边的情况我不清楚是不是也可以通过提高基函数的阶数 也就是增加N 来提高计算精度）(Wrote a two-dimensional Euler equations with discontinuous finite element program Last made ​ ​ a triangular element of the program, not curved edge unit is so easy to form a vortex in the cylinder behind the Program to curved edge quadrilateral element vortices appear Unit is the order of 8-node quadrilateral element node number is 15,263,748 which is the four corners of the points in turn Is 1234 and then the side of the midpoint of the number 5678. Time promote the use of Runge-Kutta method Numerical flux of the overall situation of Lax-Friedrichs, flux Still can not capture the shock wave did not do the reconstruction or artificial viscosity do it Zaifayici. The program is not optimized for example, memory consumption is not optimized such as straight-edge boundary integral of the unit is still mining Integral method to increase the amount of computation such as surface integral with a curved edge, the line integral using the Gauss- The Legendr)

说明：  线性代数同济第六版,最适用的一版,有书签(Linear algebra tongji sixth edition)

天线组阵中相位差的递推最小二乘估计与滤波(Phase array antenna group recursive least squares estimation and filtering)

说明：  本书是为理工科大学各专业普遍开设的“数值分析”课程编写的教材 . 其内容包括插值与逼近 ,数值微分与数值积分 , 非线性方程与线性方程组的数值解法 , 矩阵的特征值与特征向量计算 , 常微分方程数值解法 . 每章附有习题并在书末有部分答案 ,书末还附有计算实习题和并行算法简介 . 全书阐述严谨, 脉络分明 ,深入浅出 ,便于教学 .(This book is a textbook for the course of "numerical analysis" which is widely offered in various majors of universities of science and engineering. Its contents include interpolation and approximation, numerical differentiation and numerical integration, numerical solution of non-linear and linear equations, calculation of eigenvalues and eigenvectors of matrices, numerical solution of ordinary differential equations. At the end of the book, there are some answers. At the end of the book, there are also a brief introduction of computational practice questions and parallel algorithms. The whole book elaborates rigorously, clearly, thoroughly and shallowly, which is convenient for teaching.)

说明：  一个虎门大桥涡街的udf和fluent仿真教程，适合初学者(A fluent udf example)

一个用FTT法实现相关函数的快速估计程序(A correlation function with the FTT method to achieve the rapid estimation procedures)

有限元计算基本程序，为毕业论文时编写的局部渗流计算程序。(Finite element method the basic procedures for the thesis prepared by local seepage calculation procedures.)

信号处理与通信中的凸优化理论，从纯数学的角度介绍了凸优化理论的基本原理和方法(Convex Optimization in Signal Processing and Communication)

对于给定某一线性时不变网络（不含受控源），根据网络属性及结构，通过编程实现自动生成其关联矩阵A和一个树。(Given a linear time-invariant network (excluding controlled source), according to the network properties and structure of the automatically generated by programming its associated matrix A and a tree.)