Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 10 additions & 10 deletions src/09_Operators_Spectra.jl
Original file line number Diff line number Diff line change
Expand Up @@ -163,7 +163,7 @@ For example, for $\opA = - \Delta$ defined on all of $\mathbb{R}^3)$
# ╔═╡ b86c0bf2-cea9-4f6d-89bd-30634e0a2523
md"""
A subtle point in the definition of operators is that the **choice of the operator domain** $V$
(for a given "receipe" to evaluate the operator such as $\opA = - \Delta$)
(for a given "recipe" to evaluate the operator such as $\opA = - \Delta$)
has a crucial influence on the obtained spectrum $\sigma(\mathcal{A})$.
For example even in the case of physical operators such as
the Schrödinger operators $\opH = - Δ / 2 + V$ with smooth potential $V$,
Expand Down Expand Up @@ -235,16 +235,16 @@ Foldable("Properties of matrices that hold for operators", md"""
md"""
## Domains of operators

Picking up on the subtle point of operator domains, let's first establish som standard choices.
Picking up on the subtle point of operator domains, let's first establish some standard choices.
"""

# ╔═╡ 5f979847-44af-4457-9220-6840b3b1bd11
md"""
In the following we will consider the specific case of linear operators on a Hilbert space $\hilbert$, i.e. operators of the form $\opA: D(\opA) \rightarrow \hilbert$. The **domain** $D(\opA) \leq \hilbert$ is a dense subspace of $V$. Unless otherwise noted the standard choice of the domain is
In the following we will consider the specific case of linear operators on a Hilbert space $\hilbert$, i.e. operators of the form $\opA: D(\opA) \rightarrow \hilbert$. The **domain** $D(\opA) \leq \hilbert$ is a dense subspace of $\hilbert$. Unless otherwise noted the standard choice of the domain is
```math
D(\mathcal{A})=\{f \in \hilbert \mid \opA f \in \hilbert \},
```
i.e. the largest possible subspace of $\hilbert$, such that applying the operator does not take us ot of $\hilbert$. Unless otherwise noted we will always employ this domain.
i.e. the largest possible subspace of $\hilbert$, such that applying the operator does not take us out of $\hilbert$. Unless otherwise noted we will always employ this domain.

- Other choices of the domain are well possible and sometimes useful. Moreover and perhaps surprisingly it has a decisive influence on the properties of an operator. For example it is related to the operator being self-adjoint or not --- which is in turn related to the question whether the operator spectrum is physical or not, as we will discuss.

Expand All @@ -256,15 +256,15 @@ i.e. the largest possible subspace of $\hilbert$, such that applying the operato

# ╔═╡ 15012412-d229-45e4-8b12-0dc89d3baaa2
md"""
Similar to matrices we can also define norms of operators making refernce to the norm of $\hilbert$:
Similar to matrices we can also define norms of operators making reference to the norm of $\hilbert$:

!!! note "Definition (Operator norm)"
Let $\left(V,\|\cdot\|_{V}\right)$ and $\left(W,\|\cdot\|_{W}\right)$ denote two Banach spaces. The corresponding operator norm is
```math
\|\opA \|_{\boundedoperators (V, W)} \coloneqq \sup _{0 \neq x \in V} \frac{\| \opA (v)\|_{W}}{\| v \|_{V}}.
```
When dealing with an operator $\opA: D(\opA) \rightarrow \hilbert$
and whent the underlying Hilbert space is clear, we frequently denote this as
and when the underlying Hilbert space is clear, we frequently denote this as
```math
\| \opA \|_\text{op} \equiv \sup _{0 \neq f \in D(\opA)} \frac{\| \opA f\|_{\hilbert}}{\|f\|_{\hilbert}},
```
Expand Down Expand Up @@ -513,7 +513,7 @@ md"""

# ╔═╡ feb88cc7-66ac-463c-81b4-a223902c5897
md"""
In infinite dimensions one can easly construct examples where conditions (1), (2) or (3) fail separately, so the **spectrum can contain elements**, which are **not eigenvalues**. Similarly **$\sigma(\opA)$ can be empty**. See the Foldable below for explicit constructions of some of these cases.
In infinite dimensions one can easily construct examples where conditions (1), (2) or (3) fail separately, so the **spectrum can contain elements**, which are **not eigenvalues**. Similarly **$\sigma(\opA)$ can be empty**. See the Foldable below for explicit constructions of some of these cases.
"""

# ╔═╡ 8d7c4bd7-fc1e-4dc4-869c-ae33a27b015c
Expand Down Expand Up @@ -576,7 +576,7 @@ Foldable("Lemma 2: Discussion Analyticity of resolvent (optional)", md"""
md"""
## Self-adjoint operators

As discussed in principle multiple choices for the operator domain $D(\opA)$ can sometimes be reasonable and thus seems to leave considerable ambiguity.
As discussed in principle multiple choices for the operator domain $D(\opA)$ can sometimes be reasonable and thus seem to leave considerable ambiguity.

- As the Schrödinger operator example shows, there is clearly some form of a maximal domain, since we clearly need $\varphi \in D(\opH)=H^{2} (\mathbb R^{d})$ to assure $\opH \varphi \in L^{2} (\mathbb{R}^{d} )$.
But what about choosing a smaller domain, e.g. $D(H)=C_{0}^{\infty} (\mathbb{R}^{d} )$ ?
Expand Down Expand Up @@ -706,7 +706,7 @@ md"""

We provide some examples of self-adjoint operators on $\hilbert=L^{2}(\mathbb R^{d})$ and state the spectra of some of them (Proofs are given
[in the second appendix on Weyl sequences](#Example-spectra-of-operators)).
We will note that the previously introduced **Sobolev spaces arise naturally** as the the correct **domain to ensure self-adjointness** of physical operators.
We will note that the previously introduced **Sobolev spaces arise naturally** as the correct **domain to ensure self-adjointness** of physical operators.
"""

# ╔═╡ a53f5693-6a82-481c-a1ba-cb10652e96ae
Expand Down Expand Up @@ -2087,7 +2087,7 @@ where
```math
\im\Big(\mathbf 1 _{\{λ_i\}} (M)\Big) = \mathop{\mathrm{Ker}}(M - \lambda_i)
```
is the the eigenspace of eigenvalue $\lambda_i$
is the eigenspace of eigenvalue $\lambda_i$
and $\mathbf 1 _{\{λ_i\}} (M)$ is the projector into the eigenspace
of $λ_i$.

Expand Down