Ojalgo: Handling of singular matrices

Question

I am using ojalgo to solve a (NxN) Ax = b System in java. Since there are chances that A will be a singular matrix, I would like my code to know this somehow. Is there a way? (documentation states that the solve() method returns one possible solution if the problem is under-qualified and the invert() method does not throw an exception).

Any help would be greatly appreciated. Thanks in advance.

score 2 · Accepted Answer · answered May 21 '16 at 15:32

2

Sounds like you're using the solve- and invert-methods on a BasicMatrix directly.

Switch to using an LU-decompostion instead. After you have decomposed the matrix, but before attempting the solve, you can call:

lu.isSquareAndNotSingular();

answered May 21 '16 at 15:32

apete

1,250
1
10
16

2

Hi Apete, many thanks for your swift response. Could you please provide a simple code where you solve the Ax = b problem with LU-Decomposition? Thank you very much! – Nick Simos May 22 '16 at 10:18

score 0 · Answer 2 · answered Apr 01 '22 at 22:43

To solve using LU-Decomposition try something like this.

        MatrixStore M = ...equations
        MatrixStore v = ...rhs
        LU<Double> lu = LU.PRIMITIVE.make(M);
        boolean decompose = lu.decompose(M);
        MatrixStore solution;
        if(lu.isSolvable()){
            solution = lu.getSolution(v);
        // and/or check that solution is good enough
            double norm = M.multiply(solution).subtract(v).norm();
        }

Ojalgo: Handling of singular matrices

2 Answers2