Section XI.3

The Fredholm Index

[8/0%]
LaTeX
Problem XI.3.1

Prove Lemma 3.6: let M\mathcal{M} and N\mathcal{N} be two closed subspaces of the Hilbert space H\mathcal{H}.

?
(a)

If M∩N=(0)\mathcal{M} \cap \mathcal{N} = (0) and dim⁡N=∞\dim \mathcal{N} = \infty, then dim⁡M⊥=∞\dim \mathcal{M}^\perp = \infty.

(b)

If dim⁡M⊥=∞\dim \mathcal{M}^\perp = \infty and dim⁡N<∞\dim \mathcal{N} < \infty, then dim⁡(M+N)⊥<∞\dim (\mathcal{M} + \mathcal{N})^\perp < \infty.

Problem XI.3.2

If A∈B(H)A \in \mathcal{B}(\mathcal{H}) and ran A\mathrm{ran}\ A is closed, show that ran A(∞)\mathrm{ran}\ A^{(\infty )} is closed. If A∈SFA \in \mathcal{SF} and ker A=(0)\mathrm{ker}\ A = (0), show that A(∞)∈SFA^{(\infty )} \in \mathcal{SF} and ind⁡A(∞)=−∞\operatorname {ind} A^{(\infty )} = -\infty or 00.

?
Problem XI.3.3

Does the unilateral shift of multiplicity 11 have a square root?

?
Problem XI.3.4

Show that for every nn in Z∪{±∞}\mathbb {Z} \cup \left\{ \pm \infty \right\} there is an operator AA in SF\mathcal{SF} such that ind⁡A=n\operatorname {ind} A = n.

?
Problem XI.3.5

If A∈SFA \in \mathcal{SF}, then for every n≥1n \ge 1, An∈SFA^n \in \mathcal{SF} and ind⁡An=n(ind⁡A)\operatorname {ind} A^n = n(\operatorname {ind} A).

?
Problem XI.3.6

If A:H→H′A: \mathcal{H} \to \mathcal{H}' is a left semi-Fredholm operator, then there is a finite rank operator F:H→H′F: \mathcal{H} \to \mathcal{H}' such that ker(A+F)=(0)\mathrm{ker}(A + F) = (0) and ind⁡(A+F)=ind⁡A\operatorname {ind}(A + F) = \operatorname {ind} A.

?
Problem XI.3.7

If AA is a Fredholm operator in B(H)\mathcal{B}(\mathcal{H}), prove that the following statements are equivalent.

  1. ind⁡A=0\operatorname {ind} A = 0.

  2. There is a compact operator KK such that A+KA + K is invertible.

  3. There is a finite rank operator FF such that A+FA + F is invertible.

?
Problem XI.3.8

If UU is the unilateral shift of multiplicity 11 and π:B(H)→B(H)/B0(H)\pi : \mathcal{B}(\mathcal{H}) \to \mathcal{B}(\mathcal{H})/\mathcal{B}_0(\mathcal{H}) is the natural map, show that π(U)\pi (U) is normal in B(H)/B0(H)\mathcal{B}(\mathcal{H})/\mathcal{B}_0(\mathcal{H}) but there is no normal operator NN such that U−NU - N is compact. (See Exercise IX.8.14.)

?