LinAlg
This is a wrapper for the cern.colt.matrix.impl
classes
DenseDoubleMatrix1D
and DenseDoubleMatrix2D
.
The purpose is to rewrite the interface to the blas
in more
palatable and programming like form (plusEquals, timesEquals,...
).
However every layer adds a little weight. If greatest speed is necessary in loops it is best to work with raw java arrays which can be handed to matrix constructors if linear algebra operations are absolutely necessary.
Thank God for the colt library. Numerical Linear Algebra is a nontrivial subject.