ADJOINT Flashcards
PROPERTIES SELFADJOINT
- real eigenvalues
- eigenvalues corresponding to different eigenvalues are orthogonal
- are diagonalisable
EXISTENCE OF ADJOINT THEOREM
let X,Y be Hilbert spaces and T∈B(X,Y)
there exists a unique adjoint operator T*∈B(X,Y) such that
- <Tx,y> = <x,T*y>, ∀x∈X, ∀y∈Y
- ||T*||<=||T||
PROPERTIES OF ADJOINTS
- (T*)* =T
- ||T*|| = ||T||
- ||T*T|| <= ||T||^2
- T,S ∈B(X,Y)
⟹ (λT + μS)* = λ(-)T*+ μ(-)S* - T ∈B(X,Y) and S∈B(Y,Z)
⟹ (ST)* = T*S* - T∈K(X,Y) ⟹ T*∈K(Y,X)
where X,Y,Z hilbert spaces
INVERSE OF ADJOINT
(T*)^(-1) = (T^(-1))*
if T∈B(X) invertible, then so is T*
SPECTRUM OF ADJOINT
ρ(T*) = (ρ(T))(-)
and
δ(T*) = (δ(T))(-)
for T∈B(X)
ORTHOGONALITY OF ADJOINT
T∈B(X)
(ran(T-λ))^(⊥) = ker(T*-λ(-))
(ran(T*-λ(-)))^(⊥) = ker(T-λ)
ORTHOGONALITY OF ADJOINT
T∈B(X,Y)
(ran(T))^(⊥) = ker(T*)
(ran(T*))^(⊥) = ker(T)
NORMAL OPERATOR
if TT*=T*T
SELFADJOINT OPERATOR
if T=T*
UNITARY OPERATOR
if T*T = I_X and TT* = I_Y
PROPERTIES OF NORMAL OPERATORS
- ||Tx|| = ||T*x||, ∀x∈X
- ker(T-λ) = ker(T*-λ(-)), ∀λ∈K
- ρ(T) = {λ∈K : ∃c>0: ||(T-λ)x||>=c||x||, ∀x∈X}
- δ(T) = {λ∈K : ∃(xn): ||xn||=1 and (T-λ)xn->0}
- Tx= λx and Ty= μy with λ ≠ μ
⟹ <x,y>=0
ORTHOGONAL PROJECTION
P∈B(X) if
1. P^2=P
2. kerP ⊥ ranP
*X hilbert
PROJECTION P∈B(X) IS ORTHOGONAL PROJECTION ⟺
⟺ P*=P
*X hilbert
NONNEGATIVE OPERATOR
T∈B(X) if <Tx,x> >= 0, ∀x∈X
* in particular T is self-adjoint since K=C
*X hilbert
T>=0 ⟹
- ||Tx||^2 <= ||T||<Tx,x>, ∀x∈X
- ||T||= sup_(||x||=1) <Tx,x>