2.1 Basic properties of Fredholm operators
Definition
Let H be a Hilbert space. A bounded linear operator T:H→H is Fredholm if kerT and cokerT are both finite-dimensional.
Exercise
T induces an isomorphism between kerT⊥ and imT. In particular, imT is closed and (if kerT=0) 0 is an isolated point in the spectrum of T.
Definition
The index of a Fredholm operator T:H→H′ is
idxT=dimkerT−dimcokerT.
Example
If H=H′ is finite-dimensional, then all operators are Fredholm with index 0, by the rank-nullity theorem.
Example
Let shift:ℓ2→ℓ2 be the map
shift(x1,x2,x3,…)=(x2,x3,x4,…).
Then this has index 1. The adjoint of shift, written shift−1, is given by
shift−1(x1,x2,x3,…)=(0,x1,x2,…).
which has index −1.
One can similar see that idx(shiftk)=k for all k∈Z. So in particular, every integer is the index of some operator.
The fact that the index of shiftk is the negative of the index of its adjoint is not a coincidence.
Exercise
Let T:H→H be Fredholm. Then so is T∗, and idxT∗=−idxT.
We shall prove some basic properties of Fredholm operators. They all follow from the following result:
Lemma
Consider the commutative diagram
If the rows are short exact, and any two of T, S and R are Fredholm, then so is the third, and
idxS=idxT+idxR.
Apply the snake lemma to obtain a long exact sequence
Corollary
If T:H→H and S:H′→H′ are Fredholm, then so is T⊕S:H⊕H′→H⊕H′, and we have
idx(T⊕S)=idxT+idxS.
Of course, we could have proved this directly.
Corollary
Let T:H→H′ and S:H′→H” be Fredholm. Then ST is Fredholm, and
idx(ST)=idxT+idxS.
Corollary
Let I:H→H be the identity and K:H→H be a compact operator. Then I+K is Fredholm and has index 0.
First consider the case where
K has finite-dimensional image. Then consider the short exact sequences
where the vertical maps are all restrictions of I+K. One sees that the maps are well-defined.
Now note that the left-hand map is one between finite-dimensional vector spaces, so is Fredholm with index 0. On the other hand, the right-hand map is in fact the identity, since we quotiented out by the image of K, and is in particular Fredholm with index 0. So we are done.
For general compact K, we can approximate K arbitrarily well by finite-rank operators. So pick K′ such that K′ has finite rank and ∥K−K′∥<1. Thus I+(K−K′) is invertible, and we can write
I+K=(I+(K−K′))(I+(I+K−K′)−1K′).
The first term is invertible, hence Fredholm with index 0. The second is I plus something of finite rank, so is Fredholm with index 0. So the general result follows.
Theorem
Let T be a bounded linear operator. Then T is Fredholm iff there exists S:H→H such that TS−I and ST−I are compact.
Note that T gives an isomorphism between (kerT)⊥ and imT. So we can make sense of (T∣imT)−1:imT→(kerT)⊥. Then we can take S to be
Then ST−I and TS−I have finite rank, as they vanish on (kerT)⊥ and imT respectively.
Using the previous corollary, we know ST and TS are Fredholm. Since ST is Fredholm, kerT is finite-dimensional. Since TS is Fredholm, cokerT is finite-dimensional. So we are done.
Corollary
If T is Fredholm and K is compact, then T+K is also Fredholm.
We can use the same
S, since the algebra of compact operators is a two-sided ideal in
End(H).