Adjoint of Symmetric Densely-Defined Linear Operator Extends Operator
Theorem
Let $\struct {\HH, \innerprod \cdot \cdot}$ be a Hilbert space.
Let $\struct {\map D T, T}$ be a symmetric densely-defined linear operator.
Let $\struct {\map D {T^\ast}, T^\ast}$ be the adjoint of $T$.
Then $\map D T \subseteq \map D {T^\ast}$ and:
- $T x = T^\ast x$ for each $x \in \map D T$.
Proof
For each $y \in \HH$, define the linear functional $f_x : \map D T \to \Bbb F$ by:
- $\map {f_y} x = \innerprod {T x} y$ for each $x \in \map D T$.
We show that for $y \in \map D T$, $f_y$ is bounded.
Let $y \in \map D T$, then:
- $\innerprod {T x} y = \innerprod x {T y}$ for each $x \in \map D T$.
Then we have:
\(\ds \cmod {\map {f_y} x}\) | \(=\) | \(\ds \cmod {\innerprod x {T y} }\) | ||||||||||||
\(\ds \) | \(\le\) | \(\ds \norm {T y} \norm x\) | Cauchy-Bunyakovsky-Schwarz Inequality for Inner Product Spaces |
So $f_y$ is bounded and $\map D T \subseteq \map D {T^\ast}$.
Now, for $x, y \in \map D T$, we have:
- $\innerprod {T x} y = \innerprod x {T y}$
since $\struct {\map D T, T}$ is symmetric and:
- $\innerprod {T x} y = \innerprod x {T^\ast y}$
from the definition of the adjoint.
So we have:
- $\innerprod x {T y - T^\ast y} = 0$
for each $x \in \map D T$.
Taking a sequence $\sequence {x_n}_{n \mathop \in \N}$ in $\map D T$ such that $x_n \to T y - T^\ast y$.
Since:
- $\innerprod {x_n} {T y - T^\ast y} = 0$
for each $n \in \N$, we have:
- $\innerprod {T y - T^\ast y} {T y - T^\ast y} = \norm {T y - T^\ast y}^2 = 0$
from Inner Product is Continuous.
So $T y = T^\ast y$ for each $y \in \map D T$.
$\blacksquare$