 | This articlerelies excessively onreferences toprimary sources. Please improve this article by addingsecondary or tertiary sources. Find sources: "NAG Numerical Library" – news ·newspapers ·books ·scholar ·JSTOR(April 2012) (Learn how and when to remove this message) |
TheNAG Numerical Library is acommercial software product developed and sold by The Numerical Algorithms Group Ltd. It is asoftware library ofnumerical-analysis routines, containing more than 1,900 mathematical and statistical algorithms. Areas covered by the library includelinear algebra,optimization,quadrature, the solution ofordinary andpartial differential equations,regression analysis, andtime series analysis.
Users of the NAG Library call its routines from within their applications to incorporate its mathematical or statistical functionality and to solve numerical problems - for example,finding the minimum or maximum of a function,fitting a curve or surface to data, orsolving a differential equation. The NAG Library[1] can be accessed from a variety ofprogramming languages and environments such asC/C++,[2]Fortran,[3]Python,[4]Active Directory (AD),[5]MATLAB,[6]Java,[7] and.NET.[8] The main supported systems are currentlyWindows,Linux andmacOS running on x86-64 architectures;32-bitWindows support is being phased out. Some NAG mathematical optimization solvers are accessible via the optimization modelling suite.[9]
The original version of the NAG Library was written inALGOL 60 andFortran. It contained 98 user-callable routines, and was released for theInternational Computers Limited (ICL)1906A and1906S machines on October 1, 1971. Three further Marks of the library appeared in the following five years; during this time theALGOL version was ported to ALGOL 68, with the following platforms being supported:CDC 7600/CYBER (CDCALGOL 68),IBM 360/370/AMDAHL (FLACCALGOL 68),ICL 1900 (ALGOL 68R), ICL 1906A/S (ALGOL 68R), ICL 2900 (ALGOL 68RS) andTelefunken TR440 (ALGOL 68C).
The first partiallyvectorized implementation of the NAG Fortran Library for theCray-1 was released in 1983, while the first release of the NAG Parallel Library (which was specially designed fordistributed memory parallel computer architectures) was in the early 1990s. Mark 1 of the NAG C Library was released in 1990. In 1992, the Library incorporatedLAPACK routines for the first time; NAG had been a collaborator in the LAPACK project since 1987. The first release of the NAG Library for SMP & Multicore,[10] which takes advantage of theshared memory parallelism ofSymmetric Multi-Processors (SMP) andmulticore processors, appeared in 1997 for multiprocessor machines built using theDec Alpha andSPARC architectures. The NAG Library for .NET, which is aCLI DLL assembly containing methods and objects that giveCommon Language Infrastructure (CLI) users access to NAG algorithms, was first released in 2010.
Mark 29 of the NAG Library includes mathematical and statistical algorithms organised into chapters.[11]