基于贝叶斯理论的高斯过程代码，包含高斯过程回归分析，以及相关噪声处理和高斯过程分类，提供数据进行测试，等等
Gauss procedure code based on Bayesian theory, including Gaussian process regression analysis, and related processing and noise Gaussian process classification, to provide data for testing, etc. (2016-11-22, Visual C++, 20405KB, 下载5次)
Dijkstra(迪杰斯特拉)算法是典型的单源最短路径算法，用于计算一个节点到其他所有节点的最短路径。主要特点是以起始点为中心向外层层扩展，直到扩展到终点为止。
Dijkstra (Dijkstra) algorithm is a typical single-source shortest path algorithm for calculating a node to all other nodes in the shortest path. The main characteristic is the starting point as the center outward layers expand, to extend until the end point. (2016-07-18, Visual C++, 257KB, 下载2次)
迪杰斯特拉算法计算最短路径Dijkstra算法是由E.W.Dijkstra于1959年提出，又叫迪杰斯特拉算法，它应用了贪心算法模式，是目前公认的最好的求解最短路径的方法。算法解决的是有向图中单个源点到其他顶点的最短路径问题，其主要特点是每次迭代时选择的下一个顶点是标记点之外距离源点最近的顶点。但由于dijkstra算法主要计算从源点到其他所有点的最短路径，所以算法的效率较低。
Dijkstra shortest path algorithm Dijkstra algorithm is proposed in 1959 by the E.W.Dijkstra, also known as Dijkstra algorithm, which applied the greedy algorithm mode, it is widely recognized as the best method for solving the shortest path. Algorithm is a directed graph point to a single source shortest path problem other vertices, the next vertex whose main characteristic is selected at each iteration is outside the distance markers nearest source vertex. But because the main dijkstra algorithm to calculate the shortest path source to point to all other points, the low efficiency of the algorithm. (2016-05-10, Visual C++, 1KB, 下载3次)
算法当向该树插入一个元素时，若树中已存在该元素的关键码相同的结点，则该结点的count域加1；否则生成新的结点，并置该结点的count域为1
Algorithm when inserting an element to the tree, if the tree already exists the elements of the key code to the same node, then the node count domain plus 1 otherwise it produces a new node, the juxtaposition of the node count domain 1 (2016-04-14, Visual C++, 66KB, 下载1次)
现代设计理论中涉及的一些常用的优化设计方法，如一维搜索（黄金分割、二次差值等）、无约束优化方法（梯度法、牛顿法等）、约束优化方法（复合形法、DFP、惩罚函数法等）。
Some commonly used method for optimizing the design of modern design theory involved, as one-dimensional search (golden section, the second difference, etc.), unconstrained optimization methods (gradient method, Newton method, etc.), constrained optimization methods (complex method, DFP, penalty function method, etc.). (2014-06-21, Visual C++, 17KB, 下载12次)
迪杰斯特拉算法用于求解一个有向图（也可以是无向图，无向图是有向图的一种特例）的一个点（称之为原点）到其余各点（称之为周边点）的最短路径问题。
Algorithm is used to a directed graph (can also be a undirected graph, undirected graph is a directed graph of a special case) of a point (called the origin) to the rest of the various points (called the perimeter points) of the shortest path problem (2014-02-25, Visual C++, 1KB, 下载10次)
ISODAT模糊聚类虽然能够对数据聚类挖掘，但是由于网络入侵特征数据维数较多，不同入侵类别间的数据差别较小，不少入侵模式不能被准确分类。本案例采用结合模糊聚类和广义神经网络回归的聚类算法对入侵数据进行分类.rar
Although capable of clustering data clustering mining, but due to network intrusion feature data dimension of more and different types of data between invading small differences, many intrusion patterns can not be accurately classified. This case uses combined with fuzzy clustering and generalized regression neural network clustering algorithm for intrusion data classification. Rar (2013-09-10, Visual C++, 6KB, 下载2次)
最速下降法，用于解决二维情况下的函数极值问题。通过确定初始点，通过迭代将初始点加至最小点。步长和方向为迭代的元素，每次先去步长和其点的负梯度。
The steepest descent method for solving the function extremum problems in the two-dimensional case. Iteration by determining the initial point initial point was added to the minimum point. Step iteration elements, each time to go to the step and its point of negative gradient. (2013-01-13, Visual C++, 7KB, 下载5次)
卷积码译码霍夫曼编码及数学概率论贝叶斯公式；
卷积译码软判决通过3位量化可以比硬判决提高约2dB信噪比；
了解最大似然译码准则；
Soft Decoding Scheme of Convolution Codes Combined with Huffman Coding.Simulation study of the performance of ternary line codes under Viterbi decoding & Soft-decision decoding of some ternary line codes. (2011-07-12, Visual C++, 43KB, 下载16次)
问题的提出：17世纪法国数学家加斯帕的一本《数学的游戏问题》描述了许多有趣问题，约瑟问题就是其中之一：
15名基督教徒和15名异教徒同乘一船航行，途中风浪大作，危机万分，领航者告诉大家，只要将全船的一半人投入海中，其余人就能幸免。大家都同意这个办法，并协定者30人围成一圈；由第一个人起报数，每数至第9人便把他投入海中，下一个接着从1开始报数，第9人又被投入海中，依次循环，直至剩下15人为止。问题是如何排法使投入海中的人全为异教徒？
编程的思路：此问题有人把它的答案隐示于下列詩句中：
From numbers’ aid and art, (依靠数字的和魔力，)
Never will fame depart! （声誉决不会离你而去！）
这句诗中元音字母依次为：oueaiaaeeiaeea，分别用1、2、3、4、5代替a、e、i、o、u，便得一排数（划线得数字表示基督教徒，不划线的表示异教徒）。
4 5 2 1 3 1 1 2 2 3 1 2 2 1 故所求排法诗4个基督教徒，5个异教徒，再2个基督教徒，1个异教徒，…可以证明，这种排法所有的基督教徒都是安全的。
Of the problem: the 17th century French mathematician Gaspard s a "mathematical game problem " describes the many interesting questions, Joseph is one of them:
15 Christians and 15 pagans sail with the boat ride, the way storm comes up, extreme crisis, the leader tell you that as long as half of the whole ship into the sea, the rest can be spared. We all agree with this approach, and there were 30 agreements circle the first person from the reported number, the number to 9 for each person put into the sea in his next start and then reported from a number of people has been put into 9 The sea, in cycles, until the remaining 15 people so far. The question is how to arrange method to make the people into the sea are all infidels?
Programming idea: the problem was the answer to it implicitly in the following verse:
From numbers aid and art, (relying on numbers and magic)
Never will fame depart! (reputation will never leave you!)
Poem vowel sentence were: oueaiaaeeiaeea, were re (2010-12-19, Visual C++, 12KB, 下载3次)
高光谱探测技术已成为一种重要的军事侦察手段。异常检测算法能够在没有先验光谱信息的情况下检测到与周围环境存在光谱差异的目标，具有较强的实用性，并成为了一个重要的研究热点。然而数据量大、数据维高、目标小等因素也给检测带来了很大困难。在这样的背景下，本文进行了以下几方面研究。
gyjvyi (2010-05-04, Visual C++, 766KB, 下载37次)
多道程序设计中，经常是若干个进程同时处于就绪状态，为了使系统中的各进程有条不紊地运行，必须选择某种调度策略，以选择一个进程占用处理机。本次实验按照轮转算法模拟单处理机调度的过程，以加对处理机调度的算法的理解。
Multi-channel program design, often at the same time a number of processes in the ready state, in order to make the system processes in an orderly manner, you must select a scheduling policy to select a process consumes processor. The experiment simulated a single processor in accordance with Round Robin scheduling process in order to add to the processor scheduling algorithm understanding. (2009-11-25, Visual C++, 2KB, 下载2次)
双螺旋矩阵的定义如下，矩阵的最中心是1，往上是2，右拐3，向下4，然后依次5、6，7...构成一条顺序增大的螺旋线，此外，如果从中心往下走的话，也是一条对称的螺旋线。题目是给定一个矩阵维度N，将其打印出来
Double helix is defined as the matrix, the matrix is 1, up 2, right 3, down 4, to be followed by a sequence 5,6,7 ... constitute the helix increases, in addition, if Center, then go down as well as a symmetrical spiral. The subject is given a matrix dimension of N, to print out (2009-07-26, Visual C++, 1KB, 下载6次)
这是一个维数不定的两矩阵相乘的程序，该程序先检测输入的数组是否构成矩阵，再判断这两个矩阵是否满足相乘条件，若满足则进行相乘运算，并求出相乘后的结果！
This is an indefinite number of two-dimensional matrices of the program to detect whether an array of input matrix, and then determine whether the two matrix multiply to meet the conditions, if the multiplication were carried out to meet the computing and calculated multiplication the results! (2009-05-14, Visual C++, 185KB, 下载4次)
//Euler 函数前n项和
/*
phi(n) 为n的Euler原函数
if( (n/p) ％ i == 0 ) phi(n)=phi(n/p)*i
else phi(n)=phi(n/p)*(i-1)
对于约数：divnum
如果i|pr[j] 那么 divnum[i*pr[j]]=divsum[i]/(e[i]+1)*(e[i]+2) //最小素因子次数加1
否则 divnum[i*pr[j]]=divnum[i]*divnum[pr[j]] //满足积性函数条件
对于素因子的幂次 e[i]
如果i|pr[j] e[i*pr[j]]=e[i]+1 //最小素因子次数加1
否则 e[i*pr[j]]=1 //pr[j]为1次
对于本题：
1. 筛素数的时候首先会判断i是否是素数。
根据定义，当 x 是素数时 phi[x] = x-1
因此这里我们可以直接写上 phi[i] = i-1
2. 接着我们会看prime[j]是否是i的约数
如果是，那么根据上述推导，我们有：phi[ i * prime[j] ] = phi[i] * prime[j]
否则
phi[ i * prime[j] ] = phi[i] * (prime[j]-1)
(其实这里prime[j]-1就是phi[prime[j]]，利用了欧拉函数的积性)
经过以上改良，在筛完素数后，我们就计算出了phi[]的所有值。
我们求出phi[]的前缀和
*/
err (2008-12-05, Visual C++, 1KB, 下载30次)
// 带有列主元的高斯消元法
// 功能： 求解线性方程组 Ax = b
// 参数： A － 指向n*n系数矩阵的指针
// b － 常数向量的指针
// n － 方程组的维数
// 返回值：0 － 如果成功。线性方程组的解保存在 b 中
// 1 － 求解失败
//Carry out principal component of the Gaussian elimination method// function: solving linear equations Ax = b// parameters: A- point n* n coefficient matrix of the pointer// b- constant vector pointer// n- Equations dimension// return value: 0- if successful. Linear equations the solution stored in the b in// 1- Solving the failure of (2007-07-16, Visual C++, 1KB, 下载24次)
现代谱估计用莱文森-德宾(Levinson-Durbin)算法求解尤利-沃克(Yule-Walker)方程。
形参说明：
r:双精度实型一维数组，存放Yule-Walker方程的元素r(0),r(1),...r(p)。
p:AR模型阶数。
a:AR模型系数a(0),a(1),...a(p)。
v:预测误差功率
The modern spectrum estimated (Levinson-Durbin) the algorithm
solves with 鑾辨枃妫?- the German guests especially benefits- 娌冨厠
(Yule-Walker) the equation. The shape senate explained that, R: The
double precision solid unidimensional array, deposits Yule-Walker the
equation element r (0), r (1).. r (p). P:AR model step number. A:AR
model coefficient a (0), a (1).. a (p). V: Forecasts the erroneous
power (2005-09-29, Visual C++, 23KB, 下载318次)
计算ARMA(p，q)模型的功率谱密度。
形参说明：
b——双精度实型一维数组，长度为(q+1),存放ARMA(p，q)模型的滑动平均系数。
a——双精度实型一维数组，长度为(p+1),存放ARMA(p,q)模型的自回归系数。
q——整型变量,ARMA(p,q)模型的滑动平均阶数。
p——整型变量,ARMA(p,q)模型的自回归阶数。
sigma2——双精度实型变量,ARMA(p,q)模型白噪声激励的方差。
fs——双精度实型变量,采样频率(Hz)。
x——双精度实型一维数组,长度为len。当sign＝0时，存放功率谱密度；当sign＝
1时，存放用分贝表示的功率谱密度。
freq——双精度实型一维数组，长度为len。存放功率谱密度所对应的频率。
len——整型变量，功率谱密度的数据点数。
sign——整型变量,当sign＝0时，计算功率谱密度；当sign＝1时，计算用分贝表
示的功率谱密度。
calculated ARMA (p, q) model of the power spectral density.- Participation : b-- double-precision real one-dimensional arrays, length (q 1), storage ARMA (p, q) model of the average coefficient of sliding. A-- double-precision real one-dimensional arrays, length (p 1), storage ARMA (p, q) model of regression coefficients. Q-- integer variables, ARMA (p, q) moving average model of order. P-- integer variables, ARMA (p, q) model of order since the reunification. Sigma2-- double precision real-type variables, ARMA (p, q) model white noise variance. Fs-- double precision real-type variables, sampling frequency (Hz). X-- double-precision real one-dimensional array, the length of len. When the sign = 0, storage power spectral density; When the sign = 1:00, storage decibel levels, said th (2005-09-11, Visual C++, 23KB, 下载396次)
用Burg算法估计AR模型参数，进而实现功率谱估计.
形参说明：
x——双精度实型一维数组，长度为n，存放随机序列。
n--整型变量,随机序列的长度。
p--整型变量,AR模型的阶数。
a--双精度实型一维数组，长度为(p十1)。存放AR模型的系数a(0),a(1),...,a(p)。
v--双精度实型指针,它指向预测误差功率，即AR模型激励白噪声的方差。
with Burg algorithm estimates AR model parameters, thereby realizing the power spectrum estimation.- Participation : x-- double-precision real one-dimensional arrays, length n, storage random sequence. N-- integer variables, random sequence length. P-- integer variables, AR model of order. A-- double-precision real one-dimensional arrays, length (p 10). AR storage coefficient of a model (0), a (1 ),..., a (p). V-- double precision real-pointer, it forecast errors at power, AR model that inspired the white noise variance. (2005-09-11, Visual C++, 22KB, 下载478次)
多维数组（二维以下和数组）在内存中也是按顺序存放的，因此也可以用这种方式传递。但是，如果要确切知道每个元素在数组中的位置，就应给出各维的最大索引值
multi-dimensional arrays (two-dimensional array and below) in memory is stored sequentially, so you can pass this way. However, if we know exactly each element in the array position, it should be given the largest peacekeeping index (2004-09-23, Visual C++, 49KB, 下载30次)