常用数值计算库

转载自：http://www.xuebuyuan.com/1706852.html

最近在使用数值库进行稀疏矩阵线性方程组的求解，接触了一些数值计算库，主要包括：MKL, GMM++等库。

MKL库是intel的一个运算库，大名鼎鼎的matlab中都使用MKL进行数值运算，但唯一的缺点是它是需要收费的。MKL有30天的试用版，其功能与完整的版本一样。

GMM++库是一个GNU开源库，其无需安装，直接在程序中包含相对应的头文件即可。

基本概念：BLAS( Basic Linear Algebra Subprograms); LAPACK( Linear Algebra PACKage).

Intel Math Kernel Library

1.基本线形代数运算(BLAS) 向量与向量、向量与矩阵、矩阵与矩阵的运算

2.稀疏线形代数运算

3.快速傅立叶变换(单精度/双精度)(fftw)

4.LAPACK(求解线形方程组、最小方差、特征值、Sylvester方程等)

5.向量数学库(VML)

6.向量统计学库(VSL)

7.高级离散傅立叶变换

 

IMSL

软件名称 IMSL C Numerical Library(不兼容vc6编译器)

程序设计语言 C, Forton, C#, Java

资源网址 http://www.vni.com/

功能概述 分为统计库和数学库两部分. 数学库包含应用数学和特殊函数.IMSL 程序库 - 已成为数值分析解决方案的工业标准。 IMSL 程序库提供最完整与最值得信赖的函数库。 IMSL 数值程序库提供目前世界上最广泛被使用的 IMSL 算法，有超过 370 验证过、最正确与 thread-safe 的数学与统计程序。 IMSL FORTRAN 程序库提供新一代以 FORTRAN 90 为程序库基础的程序，能展现出最佳化的演算法能力应用于多处理器与其它高效能运算系统。

 

lapack

软件名称 Linear Algebra Package

程序设计语言 Fortran 77

资源网址 http://www.netlib.org/lapack

功能概述 线性代数计算子程序包

 

lapack++

软件名称 Linear Algebra Package in c++

程序设计语言 c++

资源网址 http://math.nist.gov/lapack++/

功能概述 c++版的线性代数计算子程序包

 

BLAS

软件名称 Basic Linear Algebra Subroutines

程序设计语言 Fortran 77

主要开发者 Kagstrom B. ，Ling P. ，Van Loan C.

资源网址 http://www.netlib.org/blas

功能概述 Blas是执行向量和矩阵运算的子程序集合。

 

uBLAS

BLAS in C++ with expression templates. 表达式模版形式的 C++中的BLAS ，

 

gsl

软件名称 GNU Scientific Library (linux)

程序设计语言 C , C++ compable

资源网址 http://www.gnu.org/software/gsl/

功能概述 范围广泛, 包括数值分析的常见内容

 

Blitz++

软件名称　Blitz++ (不兼容vc6编译器)

资源网址 http://sourceforge.net/project/showfiles.php?group_id=63961

功能概述　The current versions provide dense arrays and vectors, random number generators, and small vectors and matrices.是一个高效率的数值计算函数库，它的设计目的是希望建立一套既具像 C++ 一样方便，同时又比 Fortran 速度更快的数值计算环境。通常，用 C++ 所写出的数值程序，比 Fortran 慢 20% 左右，因此Blitz++ 正是要改掉这个缺点。方法是利用 C++ 的 template 技术，程序执行甚至可以比 Fortran 更快。

 

MTL

软件名称　Matrix Template Library(兼容vc6编译器)

资源网址 http://www.osl.iu.edu/research/mtl/

功能概述　The Matrix Template Library (MTL) is a high-performance generic component library that provides comprehensive linear algebra functionality for a wide variety of matrix formats. MTL专注于线性代数相关的计算任务，如各种形式矩阵的生成(对角，共轭，稀疏，对称等)，相关的计算，变换，以及与一维向量的运算

 

 

 

GAUL: Genetic Algorithm Utility Library

 

C语言, GPL授权

 

帮助你用 基因算法 写程序的库. 支持多处理器, MPI和OpenMP并行计算. 即将有C++的wraper.

 

主页: http://gaul.sourceforge.net//

 

 

Freely available libraries

 

* Linear Algebra

o PyMat, a matrix package for the Python programming language.

o JAMA, an elementary linear algebra package for Java.

o MTL, the Matrix Template Library. Dense and sparse matrices and vectors; banded, symmetric, triangular matrices; basic algorithms. C++.

o uBLAS, BLAS in C++ with expression templates.

o tvmet, a C++ library for "tiny" vectors and matrices with expression templates.

o GMM++, generic C++ template library for sparse, dense and skyline matrices, with solvers from ITL.

o dnAnalytics, C# numerics library, linear algebra, BLAS interface.

o MET, a C++ matrix library with expression templates, which eliminates the overhead of overloaded operators.

o SL++, the Scientific Library project. Will provide matrices, random numbers, complex, quaternions, plotting, and FFTs. C++.

o Seldon, C++ library for linear algebra with BLAS interface. Many matrix types (sparse, symmetric, hermitian, etc.) are supported.

o ALP, linear and polynomial algebra. Vectors, matrices, polynomials.

o SVMT: E. Robert Tisdale's proposal for a standard C++ Scalar, Vector, Matrix and Tensor Class Library (with implementation). Note: this is a proposal, not an official standard.

o GNUSSL [ftp only], the GNU Scientific Software Library. Linear algebra and arrays. C++.

o CPPLapack, C++ wrapper for BLAS and LAPACK.

o Lapack++, C++ wrapper for BLAS and LAPACK.

o IML++ A C++ template library for numerical iterative methods.

o MV++ Numerical Matrix/Vector Classes in C++

o SparseLib++ A library for sparse matrix computations, including the Sparse BLAS (Basic Linear Algebra Subprograms). C++.

o ISIS++, an object-oriented framework for solving sparse linear systems of equations. C++.

o ARPACK++, a C++ template library for solving large-scale standard and generalized eigenvalue problems.

o The Template Numerical Toolkit (TNT) for linear algebra is a successor to the Lapack++, Sparselib++, IML++, and MV++ packages. Its goal is to integrate these ideas into a generic algorithmic library, supporting generic user-defined data types, and increasing its functionality. C++.

o LinAlg, basic linear algebra and optimization classes. C++.

o CAM C++ Class Library (Matrix, vector, and graphics classes)

o Newmat, a C++ matrix library (docs, download)

o CLHEP includes matrix classes, random number generators for the High Energy Physics (HEP) community. C++.

o BPKIT, Block Preconditioning Toolkit for iterative solution of linear systems. Callable from C++, C, or FORTRAN.

 

* Arrays and Images

o FreePOOMA framework for scientific computing on sequential and parallel computers. C++.

o The Blitz++ class library: Array and Vector classes which rival Fortran's performance. C++.

o The AIPS++ Array and Image Classes (Astronomical Information Processing System). C++.

o Daixtrose, a general-purpose expression template engine.

o PETE, an expression templates library -- add expression templates to your own array class.

o SCTL (BlueSail), C++, arrays, matrics, vectors, sparse, rotations.

o NumPy, Numerical extensions to Python.

o VIGRA, generic computer vision/image processing library.

o CPPIMA A C++ image processing library

o LIMP, Large Image Manipulation Program

o Image Restoration and Inpainting, C++ library for image restoration.

o valarray

[ftp only], approximation of the valarray class described in Ch. 26 of the ANSI/ISO C++ Standard. Uses expression templates for efficient evaluation.

o Image Understanding Environment (IUE), a DARPA project. C++.

o WAILI, a wavelet transform library in C++.

 

* Neural Networks, genetic algorithms, machine learning, data mining

o PDP++, a neural-network simulation system written in C++

o EO -- Evolutionary Computation Framework

o CONICAL, C++ classes for building Neural Networks

o GALib, a C++ library for genetic algorithms

o Xelopes data mining library (Java, C++, C#)

o MLC++, Machine learning algorithms and data mining.

 

* High-Energy Physics and Quantum Chemistry

o QC++, quantum chemistry software in C++, supporting MNDO, AM1 and PM3 models.

o FTensor, C++ class library for tensors.

o GluCat, Clifford algebra template library.

o Computational Thermodynamics Library

o Tech-X has made available C++ libraries related to particle accelerator design.

 

* Multiprecision, arbitrary precision data types

o NTL, arbitrary length integers, vectors/matrices/polynomials over integers and over finite fields.

o EXTNUM, version of double which has the same number of mantissa bits as IEEE 754, but extends the exponent to have range from 10^-646456993 to 10^646456992.

o CLN, an extensive number library. Arbitrary precision integer, float, rational, polynomials, complex, modular integers, transcendental functions, assembly language kernels for some CPUs.

o MUNTL, Multiprecision unsigned number template library (C++).

o MPFUN++, a multiple precision floating point computation package in C++.

o LiDIA, A library for computational number theory. Provides a collection of highly optimized implementations of various multiprecision data types and time-intensive algorithms.

o Apfloat, a C++ High Performance Arbitrary Precision Arithmetic Package

o hfloat, An arbitrary precision package, optimized for very large (> 1000) (decimal) digit numbers.

o fPoint, a C++ class which helps convert floating-point arithmetic to fixed-point arithmetic by recommending range and precision requirements.

o doubledouble, a quad-precision (approximately 30 decimal place) floating point arithmetic class.

 

* Differential Equations

o Rheoolef, finite element environment in C++.

o EXPDE, a C++ library for solving partial differential equations on semi-unstructured grids. Parallel.

o PZ, a C++ library for finite elements. 1-3 dimensions with hp-adaptivity, continuous or discontinuous, variety of matrix formats.

o MBDyn, multibody dynamics analysis in C++. Built-in parallelization via MPI/Metis.

o P2MESH, 2D finite volume/finite elements, C++ library designed for fast prototyping of high-performance PDE solvers.

o Femlisp, a Common Lisp framework for Finite Element Methods.

o GETFEM++, a C++ finite element library, generic, arbitrary dimensions.

o MOUSE, a C++ library for finite volume computations on unstructured grids.

o DEAL, a C++ library for adaptive finite elements and error estimation. Supports SMPs.

o ODE++, a class library for ordinary differential equations. Explicit and linear-implicit ODE systems, IVP and BVP parameters, various solvers (Runge-Kutta, multistep, BDF). [English] [Deutsch]

o PETSc, object-oriented software for partial differential equations (programmed in C)

o Diffpack, partial differential equations in C++. ($)

o TIDE: Classes for Ordinary Differential Equations. Provides vectors and matrices, extrapolation integrator for ODEs, sparse matrix classes, eigenvalues, shooting for two point BVPs, nonlinear solver, continuation

o BoxLib/CCSE Applications Suite (docs), a class library supporting adaptive mesh refinement (AMR) schemes for computational fluid dynamics.

o OVERTURE, a C++ class library for solving PDEs in complicated domains. Includes adaptive mesh refinement and overlapping grids.

o FEMLIB (ftp, www) A Finite Element package [ftp only] by Michael Tiller (BROKEN LINK, no longer supported)

 

* Automatic differentiation and interval arithmetic

o Gaol, C++ library for interval arithmetic. Includes methods for interval constraint solvers.

o FADBAD-TADIFF, a C++ package for automatic differentiation using any arithmetic (double or interval) by operator overloading.

o PROFIL/BIAS [English] [Deutsch], a C++ interval arithmetic class library.

 

* Visualization

o The Visualization Toolkit (vtk), an extensive, free C++ library for scientific visualization.

o VisAD, a Java library for interactive and collaborative visualization and analysis of numerical data

o Java 2D Graph package. Includes contour plots, animation.

o Ptplot, a 2D data plotter in Java

 

* Graph Theory/Combinatorics

o Boost Graph Library, a general purpose, generic C++ library for graph data structures and graph algorithms.

o GTL, the Graph Template Library (C++).

o LEDA, a C++ library for graph theory and combinatorial computing.

 

* Language interoperability/scripting

o CPPF77 (cppf77.zip), a utility for interfacing C++ and Fortran 77 programs.

o Paul Dubois's code for interfacing Python and C++

o SWIG, generates Perl, Python, Tcl, Java, Eiffel and Guile wrappers for C++ libraries.

o SILOON (Scripting Interface Languages for Object-Oriented Numerics), toolkits and run-time support for building scripting interfaces to existing numerical codes in C, C++, and Fortran. Generates script bindings for Perl, Tcl, and Python.

o Matwrap, a tool which generates C++ wrapper code for matrix-oriented scripting languages such as Matlab 5, Octave, and tela.

 

* Transforms

o FFTPACK++, a C++ wrapper for FFTPACK complex routines using LAPACK++ Matrix and Vector classes.

o The FXT library of transforms. FFTs, Hartley, Number theoretic, Walsh, others coded in C++.

 

* Optimization

o COOOL, an object-oriented optimization library

o OptSolve++, a C++ optimization library from Tech-X. (commercial)

o StarFLIP, optimization library for combinatorial problems with fuzzy constraints (C++)

o LM (Levenberg-Marquardt) implementation in Java for nonlinear least squares problems.

 

* Miscellaneous

o GOOSE, GNU Object-Oriented Statistics Environment (C++).

o Borneo, a dialect of the Java language designed to have true support for the IEEE 754 floating point standard.

o SDTS++, a library for manipulating SDTS datasets (geographical information systems)

o Newran, C++ library for generating streams of random numbers.

o Multivariate polynomial interpolation library in C++

o MPI-2 C++ bindings (message passing library for parallel computers)

o CNCL, Communication networks simulation/analysis library (C++).

o OOMF, Object Oriented MicroMagnetic computing Framework at ITL/NIST

o CPPF77, a utility for interfacing C++ and Fortran 77 programs.

o The Standard Template Library (STL)

o The Bench++ Benchmark Suite

o EFLIB: Extended Function Library for Object-Pascal. Generalized arithmetics, equation solvers, more.

 

* Tools

o TAU (Tuning and Analysis Utilities), explicit instrumentation of C++ libraries for profiling and tracing. For both serial and parallel codes.