incorrect eigenvalues/vectors?

[Pomax]

I was trying to figure out why none of the graphics I was plotting lined up correctly, and tracked it down to numeric.js not computing the correct eigenvalues and vectors. For example:

const M = [[11200, 5600], [5600, 4800]];
const eig = numeric.eig(M);
console.log(eig.lamba.x, eig.E.x);

yields:

(2) [14449.80619863884, 1550.1938013611607]
 0: 14449.80619863884
 1: 1550.1938013611607

(2) [Array(2), Array(2)]
 0: (2) [-0.8649100931185951, 0.5019268181932333]
 1: (2) [-0.5019268181932334, -0.8649100931185951]

but these eigenvectors seem to have the wrong sign for their [0] coordinate and should instead be:

(2) [Array(2), Array(2)]
 0: (2) [0.8649100931185951, 0.5019268181932333]
 1: (2) [0.5019268181932334, -0.8649100931185951]

[HastingsGreer]

I think this code is generally not working. It gives incorrect eigenvectors for [[1, 0, 0], [0, -1, 0], [0, 0, 1]] and crashes on [[1, 0, 0], [0, 1, 0], [0, 0, 1]]

describe('numeric.js', () => { 
 beforeAll(() => { 
 // Load numeric.js into global scope 
 const numericCode = fs.readFileSync(path.resolve('./numeric.min.js'), 'utf-8') 
 eval(numericCode) 
 }) 
 
 it('computes eigenvalues of simple matrix', () => { 
 var result = numeric.eig([[1, 0, 0], [0, -1, 0], [0, 0, 1]]) 
 console.log("E:", result.E); 
 console.log(result.E.dot(numeric.T.diag(result.lambda).dot(result.E.inv()))); 
 }) 
}) 
 
~ 
~ 
~ 
~ 
~ 
~ 
TSC 5.8.3 
 RERUN mymath.test.js x38 
 
stdout | mymath.test.js > numeric.js > computes eigenvalues of simple matrix 
E: { x: [ [ 1, 0, 1 ], [ 0, 1, 0 ], [ 0, 0, 0 ] ], y: undefined } 
{ 
 x: [ [ NaN, NaN, NaN ], [ NaN, NaN, NaN ], [ NaN, NaN, NaN ] ], 
 y: undefined 
}