Definition

Let HH and KK be . A densely defined operator from HH to KK is a pair (D(T),T)(D(T),T), where D(T)D(T) is a of HH, in HH, and

T:D(T)KT:D(T)\longrightarrow K

is a . The domain D(T)D(T) is part of the operator: two maps given by the same formula on different domains are different operators. When H=KH=K, one says that TT is an operator in HH. No boundedness, continuity, or closedness is assumed.

The adjoint

Density is precisely what makes an adjoint single-valued. Using linear in the first variable, yKy\in K belongs to D(T)D(T^*) when there is zHz\in H such that

Tx,yK=x,zH\langle Tx,y\rangle_K=\langle x,z\rangle_H

for every xD(T)x\in D(T); then Ty=zT^*y=z. The vector zz is unique because D(T)D(T) is dense. The adjoint is always closed, although its own domain need not be dense.

Graph, closure, and examples

The graph of TT is the linear subspace {(x,Tx):xD(T)}HK\{(x,Tx):x\in D(T)\}\subseteq H\oplus K. The operator is when this graph is closed, and closable when its graph closure is again the graph of an operator. On an , differentiation on smooth compactly supported functions and multiplication by an unbounded are standard densely defined operators. Their domains cannot be discarded without changing their adjoints and closures.

Conventions and scope
References
  1. K. Schmüdgen, Unbounded Self-adjoint Operators on Hilbert Space, Springer, 2012. Springer DOI record. Relevant: Chapter 1, “Basics of Closed Operators.”