Java library that provides routines for quickly finding a few eigenvalues/eigenvectors of large sparse or dense matrices.
All of the functionality can be accessed through the various eigs()
and eigsh()
functions located in the MatrixDecomposition class. All functions can either be called by utilizing la4j's various matrix interfaces or by providing the left multiplication of a matrix as lambda function. Only real matrices of type double are supported.
This library is essentially a wrapper function for the ARPACK library, written in Fortran, as well as some openBLAS functions.
- Solve eigenvalue problems for square and symmetric matrices using
eigsh()
(returns real eigenvectors/eigenvalues) or non-symmetric matrices usingeigs()
(which returns complex eigenvalues/eigenvectors). - Solve standard and general eigenvalue problems.
- Support for different modes and options such as shift-invert, buckling or cayley-transform
- LU decomposition
- Matrix inversion (real and complex).
- Supports any kind of la4j matrix (sparse, dense) or any matrix defined by the user as lambda function.