.NET Matrix Library 5.0
.NET Matrix Library 5.0 Ranking & Summary
.NET Matrix Library 5.0 description
.NET Matrix Library 5.0 is regarded as an effective solution for you whether you are developing in VB.NET, C#, C++, J#, or Delphi and you need to solve systems of simultaneous equations, find least squares solutions of linear systems, multivariate regressions, solve singular value decompositions, eigenvalues and eigenvectors problems.
- Get the best of two worlds
- While exposing an easy to use and powerful interface, NML does not sacrifice any performance. Highly optimized BLAS and the standard LAPACK routines are used within the library and provide fast execution and accurate calculations.
- NML has been developed as a mixed mode C++ project, combining together managed and unmanaged code and delivering the best of both worlds; the speed of native C++ code and the feature-rich and easy to use environment of the .NET Framework.
- Internally it uses highly optimized code at processor level. This means that the processor type is detected at runtime and different brunches of code are executed in order to achieve optimal performance.
- Sparse Matrices
- Starting from version 5.0, NML contains classes for sparse matrices which can be used in combination with the dense matrix and vector classes.
- Additionally NML contains high performance sparse solver classes for solving large sparse symmetric and unsymmetric linear systems of equations.
- Parallelism: NML makes full use of the capabilities of modern processors. It achieves impressive performance gains through parallelism provided by the newest multi-core architectures. Applications using NML can benefit from its multithreading capabilities in either of the following ways:
- On single threaded applications, NML is by default configured to distribute the computational workload on all physical cores of the machine, dramatically reducing the execution time especially when it comes to operations on large matrices.
- Being thread safe, NML allows its use in threaded applications. Threads defined in client applications can use objects of the library and execute methods independently of other threads.
- Starting from version 4.1, new static methods have been introduced allowing a more precise control of the threading behavior of the library.
- Native 64-bit support
- NML is available in separate builds each one targeting either 32-bit or 64-bit platforms.
- The 32-bit version of NML will run on 32-bit versions of Windows and also will run as a 32-bit process on 64-bit Windows
- The 64-bit version of NML will run as a native 64-bit process on 64-bit Windows offering an additional 10 to 30% performance boost especially on large matrices.
- Support for sparse matrices has been added to NML. The SparseMatrix and CSparseMatrix classes have been added and they are used to represent real and complex sparse matrices.
- SparseSolver and CSparseSolver classes have been added; they are used for solving large sparse systems of linear equations.
- The FactorizationStrategy, FillInReducingMethod, MatrixType, SolverStatus, TransposeMode enumerations have been added for fine tuning the performance of the sparse solvers.
- New overloads have been added to Matrix.Multiply and CMatrix.Multiply methods.
- New overloads have been added for the multiplication operator for the Matrix and CMatrix classes.
- New overload has been added for the Vector.Add and CVector.Add methods.
- Fixed a bug which was introduced in version 4.3 regarding the CSVD.Pseudoinverse and CMatrix.Pseudoinverse methods.
- 32-bit and 64-bit versions of NML now have separate installation programs.
- Both the 32-bit and 64-bit versions of NML now share the same version number so that they can used by the same built of an application compiled as "AnyCPU".
- Compiled against the VC++ runtime libraries that come with Visual Studio 2008.
.NET Matrix Library 5.0 Screenshot
.NET Matrix Library 5.0 Keywords
Bookmark .NET Matrix Library 5.0
.NET Matrix Library 5.0 Copyright
Want to place your software product here?
Please contact us for consideration.