Geek Answers Handbook

Java eigensolvers

Does anyone from eigensolver know in Java that can give me only a few of the smallest eigenvectors without computing the entire eigendecomposition function (namely, the second smallest EV)? I looked at Colt, Jama, MTJ, UJMP, but these packages compute all eigenvectors.

+6
java linear-algebra
source share
3 answers

I don't know if this will help you, but this math library is not on your list.

Apache commons math

0
source share

Can you describe your matrix in more detail? Is it rare? In general, sparse linear algebra packets have methods for calculating only a few of the smallest or largest eigenpairs. For example, you can try using ARPACK from Java.

Another idea is to simply write your own version of Power Method, which is good at finding a few extreme eigenvalues ​​very quickly. For example, see the Book of eigenvalue templates (Hermitian) if your matrix is ​​Hermitian or the Book of eigenvalue templates (non-Hermitian) if your matrix is ​​not Hermitian.

+2
source share

MTJ includes netlib-java and has a wrapper for using arpack, so you can solve a given number of eigenvalues ​​and select attributes.

See https://github.com/fommil/matrix-toolkits-java/blob/master/src/test/java/no/uib/cipr/matrix/sparse/ArpackSymTest.java

and http://static.javadoc.io/com.googlecode.matrix-toolkits-java/mtj/1.0.4/no/uib/cipr/matrix/sparse/ArpackSym.html

0
source share

More articles:


All Articles