Is the following proof correct? invertibility of a Fredholm operator with trivial kernel.

functional-analysisproof-verification

I'm reading the following proof that boils down to the following argument.

$A: H \rightarrow H$ compact, with $H$ Hilbert space. $B: H\rightarrow H$ bounded, invertible. $0$ not an eigenvalue of $A+B$.

Then, $A+B$ is a Fredholm operator, and since $0$ is not an eigenvalue of $A+B$, then the kernel is trivial. Therefore, $A+B$ is invertible.

My first guess is that this is false, as just injectivity is not enough, we require the Fredholm operator to be of index 0 for invertibility. (Ex. the shift operator)

But maybe a Fredholm obtained in this way is forcefully of index 0?

Best Answer

Since $B$ is invertible and $A$ is compact we have that $A+B$ is Fredholm and $ind(A+B)=ind(B)=0$. Since $ \dim (ker(A+B))=0$ we have that $ codim (Im(A+B))=0$, hence $A+B$ is surjective.

Conclusion: $A+B$ is bijective.

Related Question