drucker-prager

EDFA的掺铒光纤长度模拟计算，附3篇相关论文，330mw的980nm泵泵浦，3db的泵浦吸收，12m最佳长度已经的到验证，可作为EDFA设计参考 (EDFA erbium-doped fiber length simulation of three related papers, 330mw of 980nm pump pump pump 3db absorption 12m optimum length has been to verify, can be used as the EDFA design reference)

完成了复数矩阵的计算,采用了并行算法原理，较快速的处理了稀疏矩阵方程结构。(Completed a complex matrix computation, parallel algorithm principle, more rapid processing of the sparse matrix equation structure.)

移动最小二乘曲线拟合原理，论文中分析了一维和二维的例子，并且同最小二乘法进行比较。(Moving least squares curve fitting principle, the paper analyzes the one-dimensional and two-dimensional case, and compared with the least squares method.)

可控源音频大地电磁一维occam反演程序(Controlled source audio magnetotelluric inversion program of one-dimensional occam)

矩量法的fortran程序，忘了那本书的源码了(Method of Moments fortran program, forgot the source of the book)

说明：  用二分法求解一元五次非线性方程的实数解，在高等电路学习中有很好的用途。(One yuan of five nonlinear equations dichotomy solving real solutions, there is a very good use in the higher circuit learning.)

有限元编程做网格划分的，对于从事有限元分析的比较有用(Programming the finite element mesh to do the, for finite element analysis of the more useful)

统计学习的入门书，通俗易懂，号称是ESL的入门版，全书没有太多数学推导，适合学工程的人不适合学统计的人读(Introduction to statistical learning books, easy to understand, known as the entry version of ESL, the book does not have too many mathematical derivations, suitable for people who are not suitable for learning the project to read statistics.)

使用Matlab编程，对教材74页例3.3.1进行编程仿真。其中，正电压A的值、噪声方差值、每个码元周期内的采样点数N自行设定（可设置为可调的变量）。噪声值可使用Matlab中产生高斯随机数的函数来进行仿真。贝叶斯检测判决式中，先验概率P(H1)=P(H0)，判断错误的代价因子设为1，判断正确的代价因子设为0. 按照上述参数的设定进行仿真，实现对仿真数据的贝叶斯检测判决；循环生成仿真数据，并进行判决结果的统计，即记录判决正确的次数，估计正确判决的概率。(The use of Matlab programming, teaching materials 74 pages of 3.3.1 programming simulation. Among them, the value of positive voltage A, the variance of noise, the number of sampling points in each symbol cycle (N) can be set by itself (adjustable variable). Noise values can be simulated using the function of generating Gauss random numbers in Matlab. In the Bias detection decision, the prior probability P (H1) =P (H0), the factor of error judgment is set to 1, and the correct cost factor is set to 0. According to the simulation parameters and realize the judgment of Bayesian detection simulation data; generating cycle simulation data, statistics and decision, which records the correct number of judgment, the estimated probability of correct decision.)

数值分析里面的：不动点迭代法、牛顿迭代法、共轭梯度法和最速下降法(Conjugate Gradient)