Sparse-matrix-algorithm-

fortran最优化计算方法常用程序汇编 包括一维最优化部分、无约束优化部分、约束优化部分 0.618法，分数法、二次插值法、三次插值法 共轭梯度法、DFP变尺法（导数）、DFP变尺法（差分）、阻尼最小二乘法、鲍威尔法、模式搜索法、单纯形法 混合罚函数法（SUMT调用DFP）、混合罚函数法（SUMT调用鲍威尔法）、综合约束函数双下降法、可变容差法、复合形法、网格法、随机试验法、解线性规划的单纯形法(fortran optimization calculation method used to assemble Including one-dimensional optimization section, part of unconstrained optimization, constrained optimization part 0.618, fractional method, quadratic interpolation, cubic interpolation Conjugate gradient method, DFP becomes ruler Act (derivative), DFP variable scale method (differential), damped least squares method, Powell method, pattern search method, the simplex method Mixed penalty function method (SUMT call DFP), mixed penalty function method (SUMT call Powell Law), integrated dual descent constraint functions, variable tolerance method, the complex method, grid method, random test, solving linear programming simplex Method)

三次B样条最小二乘拟合，附文献，Approximation of data using cubic B′ ezier curve least square fi tting，Author: M Khan(Cubic B-spline least squares fitting and literature. Approximation of data using cubic B ezier curve least square fitting, Author: M Khan,)

A FORTRAN IV function to compute the probability of a standard normal deviate

之前找不到合适精度的log2x函数的定点化源码，经过实现并验证后，发上来大家一起学习(fixed version for log2x)

下面我们将给出一些基于Galerkin有限元法的微型泵模拟结果，模拟过程在商业软件COMSOL Multiphysics 3.2中实现。数值求解采用压力修正算法——SIMPLE，它首先假设一个压力场，然后通过求解不可压缩流动的Navier-Stokes方程得到速度场。这些速度不需要满足Possion型连续方程，所以对压力场的修正也带来速度场的修正，最终满足质量守恒。求解速度场的同时计算电势场方程。这会得到Lorentz力，然后将其反馈回N-S方程并作为体积力处理。连续耦合Lorentz力和速度场直到Newton迭代收敛。 (Below we will give some micro-pump simulation results based on the Galerkin finite element method, the simulation process in the commercial software COMSOL Multiphysics 3.2. The numerical solution of pressure correction algorithm- the SIMPLE, it begins by assuming a pressure field, and then by solving the incompressible Navier-Stokes equations, the velocity field. These velocities do not need to satisfy the continuity equation of Possion type pressure field correction to bring the amendment of the velocity field, and ultimately to meet the conservation of mass. Solving the velocity field at the same time to calculate the potential field equations. This will be the Lorentz force, and then fed back to the NS equations and processing as a volume force. Continuous coupling of the Lorentz force and the velocity field until the Newton iteration convergence.)

时间积分算法在有限元动力学中的应用，可以计算结构的瞬态响应(This class implements a Newmark beta-gamma predictor-corrector time inte- gration scheme, with accelerations as primary unknowns. .A Newmark scheme assigns to the dofs the following 5 unknowns : - the displacement d and its predictor D - the velocity v and its predictor V - the acceletion a . Their expressions are calculated by the dofs in class Dof.)

随机调制信号下的模拟ppm，连续相位调制信号(CPM)产生，供做算法研究人员参考。( Random ppm modulated analog signal under Continuous phase modulation signal (CPM) to produce, Algorithm for researchers to do reference.)

基于现代时间序列分析方法的多传感器观测融合Kalman滤波器与Wiener滤波器.NH格式论文。(time seq analysis with kalman and winner filter.)

基于最小二乘法的抛物面拟合，内含有模拟数据和真是数据供参考。(Contains analog data based on the method of least squares parabolic fitting, and is really data for reference.)

Evesen开发的洛伦兹模型耦合集合卡尔曼滤波算法(Lorenz model coupled with ensembled Kalman Filter)