Can an invertible matrix have an eigenvalue equal to 0?

1 Answer
Nov 9, 2016

No.


A matrix is nonsingular (i.e. invertible) iff its determinant is nonzero.

To prove this, we note that to solve the eigenvalue equation

Avecv = lambdavecv,

we have that

lambdavecv - Avecv = vec0

=> (lambdaI - A)vecv = vec0

and hence, for a nontrivial solution,

|lambdaI - A| = 0.

Let A be an NxxN matrix. If we did have lambda = 0, then:

|0*I - A| = 0

|-A| = 0

=> (-1)^n|A| = 0

Note that a matrix inverse can be defined as:

A^(-1) = 1/|A| adj(A),

where |A| is the determinant of A and adj(A) is the classical adjoint, or the adjugate, of A (the transpose of the cofactor matrix).

Clearly, (-1)^(n) ne 0. Thus, the evaluation of the above yields 0 iff |A| = 0, which would invalidate the expression for evaluating the inverse, since 1/0 is undefined.

So, if the determinant of A is 0, which is the consequence of setting lambda = 0 to solve an eigenvalue problem, then the matrix is not invertible.