What's Faster for Symmetric Tridiagonal Eigendecomp: Armadillo, LAPACK, or MATLAB?

Discussion in 'Programmer's Corner' started by Twigg, Nov 20, 2015.

  1. Twigg

    Thread Starter New Member

    Jul 25, 2015
    22
    1
    It seems like such a special case of a sparse eigenvalue problem that a homemade C++ algorithm would be most efficient, but I'm not really sure. The tridiagonal matrix is square 100 x 100. It's symmetric and even better all the off-diagonal entries are equal.

    Small example: [2, -1, 0, 0; -1, 7, -1, 0; 0, -1, 42, -1; 0, 0, -1, 3]

    I want it to be efficient because this eigendecomp is part of a fitness calculation in a genetic algorithm.
     
Loading...