31.01.2015 Views

Woo Young Lee Lecture Notes on Operator Theory

Woo Young Lee Lecture Notes on Operator Theory

Woo Young Lee Lecture Notes on Operator Theory

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

CHAPTER 3.<br />

HYPONORMAL AND SUBNORMAL THEORY<br />

a pure subnormal operator with rank-<strong>on</strong>e self-commutator (pure means having no<br />

normal summand) is unitarily equivalent to a linear functi<strong>on</strong> of the unilateral shift.<br />

Morrel’s theorem can be essentially stated (also see [C<strong>on</strong>2, p.162]) that if<br />

⎧<br />

⎪⎨ (i) T is hyp<strong>on</strong>ormal;<br />

(ii) [T<br />

⎪⎩<br />

∗ , T ] is of rank-<strong>on</strong>e; and<br />

(3.8)<br />

(iii) ker [T ∗ , T ] is invariant for T ,<br />

then T −β is quasinormal for some β ∈ C. Now remember that every pure quasinormal<br />

operator is unitarily equivalent to U ⊗ P , where U is the unilateral shift and P is<br />

a positive operator with trivial kernel. Thus if [T ∗ , T ] is of rank-<strong>on</strong>e (and hence so<br />

is [(T − β) ∗ , (T − β)]), we must have P ∼ = α (≠ 0) ∈ C, so that T − β ∼ = α U, or<br />

T ∼ = α U + β. It would be interesting (in the sense of giving a simple sufficiency for<br />

the subnormality) to note that Morrel’s theorem gives that if T satisfies the c<strong>on</strong>diti<strong>on</strong><br />

(3.8) then T is subnormal. On the other hand, it was shown ([CuL2, Lemma 2.2])<br />

that if T is 2-hyp<strong>on</strong>ormal then T (ker [T ∗ , T ]) ⊆ ker [T ∗ , T ]. Therefore by Morrel’s<br />

theorem, we can see that<br />

every 2-hyp<strong>on</strong>ormal operator with rank-<strong>on</strong>e self-commutator is subnormal. (3.9)<br />

On the other hand, M. Putinar [Pu4] gave a matricial model for the hyp<strong>on</strong>ormal<br />

operator T ∈ B(H) with finite rank self-commutator, in the cases where<br />

∞∨<br />

H 0 := T ∗k( ran [T ∗ , T ] ) ∞∨<br />

has finite dimensi<strong>on</strong> d and H = T n H 0 .<br />

k=0<br />

In this case, if we write<br />

H n := G n ⊖ G n−1 (n ≥ 1) and G n :=<br />

n=0<br />

n∨<br />

T k H 0 (n ≥ 0),<br />

then T has the following two-diag<strong>on</strong>al structure relative to the decompositi<strong>on</strong> H =<br />

H 0 ⊕ H 1 ⊕ · · · :<br />

⎡<br />

⎤<br />

B 0 0 0 0 · · ·<br />

A 0 B 1 0 0 · · ·<br />

T =<br />

0 A 1 B 2 0 · · ·<br />

, (3.10)<br />

⎢<br />

⎣<br />

0 0 A 2 B 3 · · · ⎥<br />

⎦<br />

.<br />

. . . . ..<br />

k=0<br />

where<br />

dim (H n ) = dim (H n+1 ) = d (n ≥ 0);<br />

⎧⎪ ⎨<br />

[T ∗ , T ] = ([B0, ∗ B 0 ] + A ∗ 0A 0 ) ⊕ 0 ∞ ;<br />

⎪<br />

[Bn+1, ∗ B n+1 ] + A ∗ n+1A n+1 = A n A ∗ n (n ≥ 0);<br />

⎩<br />

A ∗ nB n+1 = B n A ∗ n (n ≥ 0).<br />

(3.11)<br />

We will refer the operator (3.10) to the Putinar’s matricial model of rank d. This<br />

model was also introduced in [GuP, Pu1, Xi3, Ya1], and etc.<br />

102

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!