Compare commits
3
Commits
6dbdd2396d
...
343843f281
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
343843f281 | ||
|
|
d698b9b7f1 | ||
|
|
cb83d2d061 |
@@ -369,4 +369,160 @@ A significant property of Hilbert spaces is that every closed subspace is comple
|
||||
\end{proof}
|
||||
|
||||
|
||||
\begin{definition}[Quadratic Function]
|
||||
\label{definition:hilbert-quadratic-function}
|
||||
Let $H$ be a complex Hilbert space and $q: H \to \complex$, then $q$ is \textbf{quadratic} if:
|
||||
\begin{enumerate}
|
||||
\item For each $\lambda \in \complex$ and $\xi \in H$, $q(\lambda x) = |\lambda|^2 q(x)$.
|
||||
\item (\textbf{Parallelogram Law}) For each $\xi, \eta \in H$,
|
||||
\[
|
||||
q(\xi + \eta) + q(\xi - \eta) = 2q(\xi) + 2q(\eta)
|
||||
\]
|
||||
\end{enumerate}
|
||||
|
||||
and \textbf{bounded} if
|
||||
\begin{enumerate}[start=2]
|
||||
\item There exists $M \ge 0$ such that $|q(\xi)| \le M\norm{\xi}_H^2$ for all $\xi \in H$.
|
||||
\end{enumerate}
|
||||
\end{definition}
|
||||
|
||||
\begin{proposition}
|
||||
\label{proposition:quadratic-function-sesquilinear}
|
||||
Let $H$ be a complex Hilbert space, $q: H \to \complex$ be a bounded \hyperref[quadratic function]{definition:hilbert-quadratic-function}, and
|
||||
\[
|
||||
H \times H \to \complex \quad (\xi, \eta) \mapsto \dpn{\xi, \eta}{q} := \frac{1}{4}\sum_{k = 0}^3i^kq(\xi + i^k\eta)
|
||||
\]
|
||||
|
||||
then
|
||||
|
||||
|
||||
\begin{enumerate}
|
||||
\item $\dpn{\cdot, \cdot}{q}$ is the unique bounded sesquilinear form on $H$ such that $\dpn{\xi, \xi}{q} = q(\xi)$ for all $\xi \in H$.
|
||||
\item There exists a unique $T \in B(H)$ such that $q(\xi) = \dpn{T\xi, \xi}{H}$ for all $\xi \in H$.
|
||||
\end{enumerate}
|
||||
\end{proposition}
|
||||
\begin{proof}
|
||||
(1): Uniqueness is given by \hyperref[polarisation]{proposition:polarisation-complex}. Existence requires a large number of checkpoints, which are detailed below. Let $M \ge 0$ such that $|q(\xi)| \le M\norm{\xi}_H^2$ for all $\xi \in H$.
|
||||
\begin{enumerate}[label=(1.\roman*)]
|
||||
\item For each $\xi, \eta, \theta \in H$,
|
||||
\[
|
||||
\dpn{\xi, \eta}{q} + \dpn{\theta, \eta}{q} = 2 \angles{\frac{\xi + \theta}{2}, \eta}_{q}
|
||||
\]
|
||||
|
||||
and
|
||||
\[
|
||||
\dpn{\xi, \eta}{q} + \dpn{\xi, \theta}{q} = 2 \angles{\xi, \frac{\eta + \theta}{2}}_{q}
|
||||
\]
|
||||
\item For each dyadic rational number $\lambda \in \mathbb{D}$ with $\lambda \in [0, 1]$,
|
||||
\[
|
||||
\dpn{\lambda\xi, \eta}{q} = \lambda\dpn{\xi, \eta}{q} = \dpn{\xi, \lambda \eta}{q}
|
||||
\]
|
||||
|
||||
\item For each $\xi, \eta, \theta \in H$,
|
||||
\[
|
||||
\dpn{\xi + \theta, \eta}{H} = \dpn{\xi, \eta}{H} + \dpn{\theta, \eta}{H}
|
||||
\]
|
||||
|
||||
and
|
||||
\[
|
||||
\dpn{\xi, \eta + \theta}{H} = \dpn{\xi, \eta}{H} + \dpn{\xi, \theta}{H}
|
||||
\]
|
||||
|
||||
\item For each dyadic rational number $\lambda \in \mathbb{D}$ with $\lambda \in \mathbb{D}$,
|
||||
\[
|
||||
\dpn{\lambda\xi, \eta}{q} = \lambda\dpn{\xi, \eta}{q} = \dpn{\xi, \lambda \eta}{q}
|
||||
\]
|
||||
|
||||
\item For each $\xi, \eta \in H$, $|\dpn{\xi, \eta}{q}| \le M(\norm{\xi}_H^2 +\norm{\eta}_H^2)$.
|
||||
\item For each $\xi, \eta \in H$, $|\dpn{\xi, \eta}{q}| \le 3M \norm{\xi}_H\norm{\eta}_H$.
|
||||
|
||||
\item $\dpn{\cdot, \cdot}{q}$ is a sesquilinear form on $H$.
|
||||
\item For all $\xi \in H$, $\dpn{\xi, \xi}{q} = q(\xi)$.
|
||||
\end{enumerate}
|
||||
|
||||
|
||||
(1.i): Let $\xi, \eta, \theta \in H$, then
|
||||
\[
|
||||
4\dpn{\xi, \eta}{q} + 4 \dpn{\theta, \eta}{q} = \sum_{k = 0}^3 i^k\braks{q(\xi + i^k\eta) + q(\theta + i^k\eta)}
|
||||
\]
|
||||
|
||||
For each $0 \le k \le 3$, by the Parallelogram Law applied to the pair $((\xi + \theta)/2 + i^k\eta, (\xi - \eta)/2)$,
|
||||
\[
|
||||
q(\xi + i^k\eta) + q(\theta + i^k\eta) = 2\braks{q\paren{\frac{\xi + \theta}{2} + i^k\eta} + q\paren{\frac{\xi - \theta}{2}}}
|
||||
\]
|
||||
|
||||
so
|
||||
\begin{align*}
|
||||
4\dpn{\xi, \eta}{q} + 4 \dpn{\theta, \eta}{q} &= \sum_{k = 0}^3 i^k\braks{q(\xi + i^k\eta) + q(\theta + i^k\eta)} \\
|
||||
&= 2\sum_{k = 0}^3 i^k\braks{q\paren{\frac{\xi + \theta}{2} + i^k\eta} + q\paren{\frac{\xi - \theta}{2}}} \\
|
||||
&= 2\sum_{k = 0}^3 i^k{q\paren{\frac{\xi + \theta}{2} + i^k\eta}} = 2 \angles{\frac{\xi + \theta}{2}, \eta}_{q}
|
||||
\end{align*}
|
||||
|
||||
Similarly, for each $0 \le k \le 3$, by the Parallelogram Law applied to the pair $(\xi + i^k(\eta + \theta)/2, i^k(\eta - \theta)/2)$,
|
||||
\[
|
||||
q(\xi + i^k\eta) + q(\xi + i^k\theta) = 2\braks{q\paren{\xi + i^k\frac{\eta + \theta}{2}} + q\paren{i^k\frac{\eta - \theta}{2}}}
|
||||
\]
|
||||
|
||||
and
|
||||
\[
|
||||
\dpn{\xi, \eta}{q} + \dpn{\xi, \theta}{q} = 2\angles{\xi, \frac{\eta + \theta}{2}}_q
|
||||
\]
|
||||
|
||||
|
||||
so $\dpn{\cdot, \cdot}{q}: H \times H \to \complex$ is a $\real$-bilinear form.
|
||||
|
||||
(1.ii): By definition of $\dpn{\cdot, \cdot}{q}$, the claim holds for $0$ and $1$. Let $n \in \natz$ and suppose inductively that the claim holds for all $\lambda \in \mathbb{D}_n \cap [0, 1]$. Take $\lambda \in \mathbb{D}_{n+1} \cap [0, 1]$, then there exists $\mu, \nu \in \mathbb{D}_n \cap [0, 1]$ such that $\lambda = (\mu + \nu)/2$. In which case, by (1.i),
|
||||
\begin{align*}
|
||||
\dpn{\lambda \xi, \eta}{q} &= \angles{\frac{\mu + \nu}{2}\xi, \eta}_{q} = \frac{1}{2}\dpn{\mu\xi, \eta}{q} + \frac{1}{2}\dpn{\nu\xi, \eta}{q} \\
|
||||
&= \frac{\mu + \nu}{2}\dpn{\xi, \eta}{q} = \lambda \dpn{\xi, \eta}{q}
|
||||
\end{align*}
|
||||
|
||||
By induction, $\dpn{\lambda \xi, \eta}{q} = \lambda \dpn{\xi, \eta}{q}$ for all $\lambda \in \mathbb{D} \cap [0, 1]$.
|
||||
|
||||
Similarly, $\dpn{\xi, \lambda \eta}{q} = \lambda \dpn{\xi, \eta}{q}$ for all $\lambda \in \mathbb{D} \cap [0, 1]$.
|
||||
|
||||
(1.iii): Let $\xi, \eta, \theta \in H$, then by (1.i) and (1.ii),
|
||||
\[
|
||||
\frac{1}{2}(\dpn{\xi, \eta}{q} + \dpn{\theta, \eta}{q}) = \angles{\frac{\xi + \theta}{2}, \eta}_q = \frac{1}{2}\dpn{\xi + \theta, \eta}{q}
|
||||
\]
|
||||
|
||||
(1.iv): By (1.ii) and repeated application of (1.iii),
|
||||
\[
|
||||
\dpn{\lambda \xi, \eta}{q} = \lambda\dpn{\xi, \eta}{q} = \dpn{\xi, \lambda \eta}{q}
|
||||
\]
|
||||
|
||||
for all $\lambda \in \mathbb{D}$.
|
||||
|
||||
(1.v): Let $\xi, \eta \in H$, then
|
||||
\begin{align*}
|
||||
|\dpn{\xi, \eta}{q}| &\le \frac{1}{4}\sum_{k = 0}^3 |q(\xi + i^k\eta)| = \frac{M}{4}\sum_{k = 0}^3\normn{\xi + i^k\eta}_H \\
|
||||
&\le \frac{M}{4}\sum_{k = 0}^3 [\norm{\xi}_H^2 + \norm{\eta}_H^2 + i^{-k}\dpn{\xi, \eta}{H}] = M(\norm{\xi}_H^2 + \norm{\eta}_H^2)
|
||||
\end{align*}
|
||||
|
||||
|
||||
|
||||
(1.vi): Let $\xi, \eta \in H$ and assume without loss of generality that $\xi, \eta \ne 0$. Since $\norm{\eta}_H/\norm{\xi}_H > 0$, there exists $n \in \integer$ such that $\norm{\eta}_H/\norm{\xi}_H \in [2^n, 2^{n+1}]$. In which case, by (1.iv),
|
||||
\begin{align*}
|
||||
|\dpn{\xi, \eta}{q}| &= 2^{-n}|\dpn{2^n\xi, \eta}{q}| \le 2^{-n}M(4^n \norm{\xi}_H^2 + \norm{\eta}_H^2) = 2^nM\norm{\xi}_H^2 + 2^{-n}M\norm{\eta}_H^2 \\
|
||||
&\le M\norm{\xi}_H\norm{\eta}_H + 2M\norm{\xi}_H\norm{\eta}_H = 3M\norm{\xi}_H\norm{\eta}_H
|
||||
\end{align*}
|
||||
|
||||
(1.vii): By (1.iv) and (1.vi), $\dpn{\cdot, \cdot}{q}$ is a $\real$-bilinear form on $H$. For any $\xi, \eta \in H$, by absolute homogeneity of $q$,
|
||||
\begin{align*}
|
||||
\dpn{i\xi, \eta}{q} &= \frac{1}{4}\sum_{k = 0}^3 i^k q(i\xi + i^k\eta) = \frac{1}{4}\sum_{k = 0}^3 i^k q(\xi + i^{k-1}\eta) \\
|
||||
&= \frac{i}{4}\sum_{k = 0}^3 i^{k-1} q(\xi + i^{k-1}\eta) = \dpn{\xi, \eta}{q} \\
|
||||
\dpn{\xi, i\eta}{q} &= \frac{1}{4}\sum_{k = 0}^3 i^k q(\xi + i^{k+1}\eta) = \frac{1}{4}\sum_{k = 0}^3 i^k q(i^{-1}\xi + i^{k}\eta) \\
|
||||
&= \dpn{-i\xi, \eta}{q} = -i \dpn{\xi, \eta}{q}
|
||||
\end{align*}
|
||||
|
||||
Therefore $\dpn{\cdot, \cdot}{q}$ is a sesquilinear form.
|
||||
|
||||
(1.viii): Let $\xi \in H$, then by absolute homogeneity of $q$,
|
||||
\begin{align*}
|
||||
\dpn{\xi, \xi}{q} &= \frac{1}{4}\braks{q(\xi + \xi) - q(\xi - \xi) + iq(\xi + i\xi) - iq(\xi - i\xi)} \\
|
||||
&= \frac{1}{4}\braks{4q(\xi) + iq(\xi + i\xi) - iq(i\xi + \xi)} = q(\xi)
|
||||
\end{align*}
|
||||
|
||||
(2): By the \hyperref[Riesz Representation Theorem]{theorem:riesz-hilbert}.
|
||||
\end{proof}
|
||||
|
||||
|
||||
@@ -34,60 +34,6 @@ The following is a slight generalisation of one form of the above example, but s
|
||||
Let $A$ be a $C^*$-algebra, $S, T \subset A$ be operator systems, and $\phi: S \to T$. If there exists $x \in A$ such that $\phi(s) = x\phi(s)x^*$ for all $s \in S$, then $\phi$ is completely positive.
|
||||
\end{proposition}
|
||||
|
||||
|
||||
|
||||
\begin{proposition}
|
||||
\label{proposition:completely-positive-norm}
|
||||
Let $A, B$ be $C^*$-algebras with $A$ being unital, $S \subset A$ be an operator system, and $\phi: S \to B$ be a completely positive map, then
|
||||
\[
|
||||
\norm{\phi}_{\text{cb}} = \norm{\phi}_{L(S; B)} = \norm{\phi(1_A)}_B
|
||||
\]
|
||||
\end{proposition}
|
||||
\begin{proof}
|
||||
After taking a \hyperref[unitisation]{proposition:c-star-unitisation}, assume without loss of generality that $B$ is unital.
|
||||
|
||||
It is sufficient to show that $\norm{\phi}_{\text{cb}} \le \norm{\phi(1_A)}_B$. To this end, let $T \in M_n(S)$ with $\norm{T}_{M_n(S)} \le 1$, then
|
||||
\[
|
||||
\begin{bmatrix}
|
||||
1_{M_n(A)} &T \\
|
||||
T^* &1_{M_n(A)}
|
||||
\end{bmatrix} \ge 0
|
||||
\]
|
||||
|
||||
in $M_{2}(M_n(S)) = M_{2n}(S)$ by \autoref{lemma:norm-matrix-test}. Since $\phi$ is completely positive,
|
||||
\[
|
||||
\phi_{2n}\begin{bmatrix}
|
||||
1_{M_n(A)} &T \\
|
||||
T^* &1_{M_n(A)}
|
||||
\end{bmatrix} = \begin{bmatrix}
|
||||
\phi_n(1_{M_n(A)}) &\phi_n(T) \\
|
||||
\phi_n(T)^* & \phi_n(1_{M_n(A)})
|
||||
\end{bmatrix} \ge 0
|
||||
\]
|
||||
|
||||
First suppose that $\phi_n(1_{M_n(A)})$ is invertible, then
|
||||
|
||||
\begin{align*}
|
||||
\phi_n(1_{M_n(A)}) &\le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)} \cdot 1_{M_n(B)} \\
|
||||
1_{M_n(B)} &\le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)} \cdot \phi_n(1_{M_n(A)})^{-1} \\
|
||||
\normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^{-1} &\le \phi_n(1_{M_n(A)})^{-1}
|
||||
\end{align*}
|
||||
|
||||
By the \hyperref[Schur complement condition]{proposition:schur-complement-positive},
|
||||
\begin{align*}
|
||||
\phi_n(1_{M_n(A)}) &\ge \phi_n(T)\phi_n(1_{M_n(A)})^{-1}\phi_n(T)^* \\
|
||||
&\ge \normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^{-1}\phi_n(T)\phi_n(T)^* \\
|
||||
\normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^2 &\ge \normn{\phi_n(T)\phi_n(T)^*}_{M_n(B)} = \normn{\phi_n(T)}_{M_n(B)}^2
|
||||
\end{align*}
|
||||
|
||||
and $\normn{\phi_n(T)}_{M_n(B)} \le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)}$. Therefore $\norm{\phi}_{\text{cb}} \le \normn{\phi(1_A)}_B$.
|
||||
|
||||
Now suppose that $\phi$ is an arbitrary completely positive map. Let $\psi \in S(A)$ be a state and $\eps > 0$, then $\phi + \eps 1_B\psi$ is also a completely positive map. In which case,
|
||||
\[
|
||||
\norm{\phi}_{\text{cb}} \le \inf_{\eps > 0}\norm{\phi + \eps1_B\psi}_{\text{cb}} \le \inf_{\eps > 0}\norm{\phi(1_A) + \eps 1_B}_B = \norm{\phi(1_A)}_B
|
||||
\]
|
||||
\end{proof}
|
||||
|
||||
\begin{proposition}
|
||||
\label{proposition:commutative-mn-facts}
|
||||
Let $X$ be a compact Hausdorff space, then:
|
||||
|
||||
@@ -180,7 +180,58 @@
|
||||
\end{proof}
|
||||
|
||||
|
||||
% Omitted
|
||||
\begin{proposition}
|
||||
\label{proposition:completely-positive-norm}
|
||||
Let $A, B$ be $C^*$-algebras with $A$ being unital, $S \subset A$ be an operator system, and $\phi: S \to B$ be a completely positive map, then
|
||||
\[
|
||||
\norm{\phi}_{\text{cb}} = \norm{\phi}_{L(S; B)} = \norm{\phi(1_A)}_B
|
||||
\]
|
||||
\end{proposition}
|
||||
\begin{proof}
|
||||
After taking a \hyperref[unitisation]{proposition:c-star-unitisation}, assume without loss of generality that $B$ is unital.
|
||||
|
||||
It is sufficient to show that $\norm{\phi}_{\text{cb}} \le \norm{\phi(1_A)}_B$. To this end, let $T \in M_n(S)$ with $\norm{T}_{M_n(S)} \le 1$, then
|
||||
\[
|
||||
\begin{bmatrix}
|
||||
1_{M_n(A)} &T \\
|
||||
T^* &1_{M_n(A)}
|
||||
\end{bmatrix} \ge 0
|
||||
\]
|
||||
|
||||
in $M_{2}(M_n(S)) = M_{2n}(S)$ by \autoref{lemma:norm-matrix-test}. Since $\phi$ is completely positive,
|
||||
\[
|
||||
\phi_{2n}\begin{bmatrix}
|
||||
1_{M_n(A)} &T \\
|
||||
T^* &1_{M_n(A)}
|
||||
\end{bmatrix} = \begin{bmatrix}
|
||||
\phi_n(1_{M_n(A)}) &\phi_n(T) \\
|
||||
\phi_n(T)^* & \phi_n(1_{M_n(A)})
|
||||
\end{bmatrix} \ge 0
|
||||
\]
|
||||
|
||||
First suppose that $\phi_n(1_{M_n(A)})$ is invertible, then
|
||||
|
||||
\begin{align*}
|
||||
\phi_n(1_{M_n(A)}) &\le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)} \cdot 1_{M_n(B)} \\
|
||||
1_{M_n(B)} &\le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)} \cdot \phi_n(1_{M_n(A)})^{-1} \\
|
||||
\normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^{-1} &\le \phi_n(1_{M_n(A)})^{-1}
|
||||
\end{align*}
|
||||
|
||||
By the \hyperref[Schur complement condition]{proposition:schur-complement-positive},
|
||||
\begin{align*}
|
||||
\phi_n(1_{M_n(A)}) &\ge \phi_n(T)\phi_n(1_{M_n(A)})^{-1}\phi_n(T)^* \\
|
||||
&\ge \normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^{-1}\phi_n(T)\phi_n(T)^* \\
|
||||
\normn{\phi_n(1_{M_n(A)})}_{M_n(B)}^2 &\ge \normn{\phi_n(T)\phi_n(T)^*}_{M_n(B)} = \normn{\phi_n(T)}_{M_n(B)}^2
|
||||
\end{align*}
|
||||
|
||||
and $\normn{\phi_n(T)}_{M_n(B)} \le \normn{\phi_n(1_{M_n(A)})}_{M_n(B)}$. Therefore $\norm{\phi}_{\text{cb}} \le \normn{\phi(1_A)}_B$.
|
||||
|
||||
Now suppose that $\phi$ is an arbitrary completely positive map. Let $\psi \in S(A)$ be a state and $\eps > 0$, then $\phi + \eps 1_B\psi$ is also a completely positive map. In which case,
|
||||
\[
|
||||
\norm{\phi}_{\text{cb}} \le \inf_{\eps > 0}\norm{\phi + \eps1_B\psi}_{\text{cb}} \le \inf_{\eps > 0}\norm{\phi(1_A) + \eps 1_B}_B = \norm{\phi(1_A)}_B
|
||||
\]
|
||||
\end{proof}
|
||||
|
||||
|
||||
\begin{proposition}
|
||||
\label{proposition:unital-2-positive-contractive}
|
||||
|
||||
@@ -51,7 +51,7 @@
|
||||
|
||||
so $|\dpn{T\xi, \xi}{H}| \le 1$, and $w(T) \le 1$.
|
||||
|
||||
(3): For any $\xi \in H$ with $\norm{\xi}_H = 1$, $|\dpn{T\xi, \xi}{H}| \le \norm{T}_{B(H)}$, so $w(T) \le \norm{T}_H$. On the other hand, if $w(T) \le 1$, then for each $\xi, \eta \in H$ with $\dpn{T\xi, \eta}{H} \in \real$,
|
||||
(3): For any $\xi \in H$ with $\norm{\xi}_H = 1$, $|\dpn{T\xi, \xi}{H}| \le \norm{T}_{B(H)}$, so $w(T) \le \norm{T}_{B(H)}$. On the other hand, if $w(T) \le 1$, then for each $\xi, \eta \in H$ with $\dpn{T\xi, \eta}{H} \in \real$,
|
||||
\begin{align*}
|
||||
\dpn{T\xi, \eta}{H} &= \frac{1}{4}\sum_{k = 0}^3 i^k\dpn{T(\xi + i^k\eta), \xi + i^k\eta}{H} \\
|
||||
&= \frac{1}{4}(\dpn{T(\xi + \eta), \xi + \eta}{H} - \dpn{T(\xi - \eta), \xi - \eta}{H}) \\
|
||||
@@ -65,7 +65,7 @@
|
||||
\norm{T\xi}_H^2 &\le \norm{\xi}_H^2
|
||||
\end{align*}
|
||||
|
||||
(4): For any $\xi \in H$ with $\norm{\xi}_H = 1$, $|\dpn{T\xi, \xi}{H}| \le \norm{T}_{B(H)}$, so $w(T) \le \norm{T}_H$. The sharp case is given by $I_H$.
|
||||
(4): For any $\xi \in H$ with $\norm{\xi}_H = 1$, $|\dpn{T\xi, \xi}{H}| \le \norm{T}_{B(H)}$, so $w(T) \le \norm{T}_{B(H)}$. The sharp case is given by $I_H$.
|
||||
|
||||
On the other hand, if $w(T) \le 1$, then $2 + T + T^* = 2 + 2\text{Re}(T) \ge 0$ and $2 - T - T^* = 2 - 2\text{Re}(T) \ge 0$ imply that $\norm{\text{Re}(T)}_{B(H)} \le 1$. Similarly, $\norm{\text{Im}(T)}_{B(H)} \le 1$ as well, so $\norm{T}_{B(H)} \le \norm{\text{Re}(T)}_{B(H)} + \norm{\text{Im}(T)}_{B(H)} \le 2$. Therefore $\norm{T}_{B(H)} \le 2w(T)$.
|
||||
|
||||
@@ -97,7 +97,7 @@
|
||||
|
||||
\begin{proposition}
|
||||
\label{proposition:cyclic-forward-shift-radius}
|
||||
Let $H$ be a complex Hilbert space, $T \in B(H)$, $R = \sum_{j = 2}^nE_{(j+1)j} \otimes T \in M_n(B(H))$, $\bracsn{e_j}_0^{n-1} \subset \complex^n$ be the canonical basis for $\complex^n$, and
|
||||
Let $H$ be a complex Hilbert space, $T \in B(H)$, $R = \sum_{j = 1}^{n-1}E_{(j+1)j} \otimes T \in M_n(B(H))$, $\bracsn{e_j}_0^{n-1} \subset \complex^n$ be the canonical basis for $\complex^n$, and
|
||||
\[
|
||||
S_n: \complex^n \to \complex^n \quad S_ne_j = e_{j+1 \mod n}
|
||||
\]
|
||||
@@ -127,7 +127,7 @@
|
||||
(2): Let $\bracsn{f_j}_{0}^{n-1} \subset \complex^n$ be an orthonormal eigenbasis for $S_n$, with eigenvalues $\bracsn{\omega^k}_0^{n-1}$. For any $\bracsn{\xi_j}_0^{n-1} \subset H$, let $\xi = \sum_{j = 0}^{n-1}\xi_j \otimes f_j$, then
|
||||
\begin{align*}
|
||||
\dpn{(T \otimes S_n)\xi, \xi}{H^n} &= \sum_{j, k = 0}^{n-1}\dpn{(T \otimes S_n)(\xi_j \otimes f_j), \xi_k \otimes f_k}{H^n} \\
|
||||
&= \sum_{j, k = 0}^{n-1}\dpn{T\xi_j, \xi_k}{H} \dpn{S_nf_j, f_k}{H} = \sum_{j = 0}^{n-1}\omega^j\dpn{T\xi_j, \xi_j}{H} \\
|
||||
&= \sum_{j, k = 0}^{n-1}\dpn{T\xi_j, \xi_k}{H} \dpn{S_nf_j, f_k}{\complex^n} = \sum_{j = 0}^{n-1}\omega^j\dpn{T\xi_j, \xi_j}{H} \\
|
||||
|\dpn{(T \otimes S_n)\xi, \xi}{H^n}| &\le \sum_{j = 0}^{n-1}|\omega^j| \cdot w(T) \cdot \norm{\xi_j}_{H}^2 = w(T)\norm{\xi}_{H^n}^2
|
||||
\end{align*}
|
||||
|
||||
@@ -140,7 +140,7 @@
|
||||
|
||||
(3): Let $\xi = (\xi_1, \cdots, \xi_{n}) \in H^n$, then
|
||||
\begin{align*}
|
||||
\dpn{R\xi, \xi}{H^n} &= \sum_{j = 2}^{n}\dpn{(E_{(j+1)j}\otimes T)\xi, \xi}{H^n} = \sum_{j = 1}^{n-1} \dpn{T\xi_j, \xi_{j+1}}{H} \\
|
||||
\dpn{R\xi, \xi}{H^n} &= \sum_{j = 1}^{n-1}\dpn{(E_{(j+1)j}\otimes T)\xi, \xi}{H^n} = \sum_{j = 1}^{n-1} \dpn{T\xi_j, \xi_{j+1}}{H} \\
|
||||
&= \dpn{T0, \xi_1}{H} + \dpn{T\xi_n, 0}{H} + \sum_{j = 1}^{n-1} \dpn{T\xi_j, \xi_{j+1}}{H} \\
|
||||
&= \dpn{(T \otimes S_{n+1})(\xi_1, \cdots, \xi_n, 0), (\xi_1, \cdots, \xi_n, 0)}{H^{n+1}}
|
||||
\end{align*}
|
||||
|
||||
@@ -7,15 +7,15 @@
|
||||
Let $H$ be a complex Hilbert space, $A = \sum_{i, j = 1}^n E_{ij} \otimes a_{ij} \in M_n(B(H))$ be positive, and $\Lambda = \sum_{i, j = 1}^n \lambda_{ij}E_{ij} \in M_n(\complex)$ be positive, then $S_A(\Lambda) = \sum_{i, j = 1}^n \lambda_{ij}E_{ij} \otimes a_{ij}$ is also positive.
|
||||
\end{lemma}
|
||||
\begin{proof}
|
||||
For any $\xi \in \complex^n$, the projection map $P_\xi$ onto its span is given by $\sum_{i, j = 1}^n \xi_i E_{ij} \ol{\xi_j}$. In which case,
|
||||
For any unit vector $\xi \in \complex^n$, the projection map $P_\xi$ onto its span is given by $\sum_{i, j = 1}^n \xi_i E_{ij} \ol{\xi_j}$. In which case,
|
||||
\begin{align*}
|
||||
S_A(P_\xi) &= \sum_{i, j = 1}^n \xi_i \ol{\xi_j} E_{ij} \otimes a_{ij} \\
|
||||
&= \braks{\sum_{i = 1}^n \xi_{i}E_{ii}}\braks{\sum_{i, j = 1}^n \xi_i \ol{\xi_j} E_{ij} \otimes a_{ij}}\braks{\sum_{j = 1}^n {\xi_j}E_{jj}}^* \ge 0
|
||||
&= \braks{\sum_{i = 1}^n \xi_{i}E_{ii}}\braks{\sum_{i, j = 1}^n E_{ij} \otimes a_{ij}}\braks{\sum_{j = 1}^n {\xi_j}E_{jj}}^* \ge 0
|
||||
\end{align*}
|
||||
|
||||
|
||||
|
||||
Using the \hyperref[Spectral Theorem]{definition:linfty-functional-calculus}, let $\bracsn{\xi^{(j)}}_1^n \subset \complex^n$ be an orthonormal eigenbasis for $\Lambda$ with eigenvectors $\seqf{\lambda_j} \subset \complex$. For each $1 \le j \le n$, let $P_j \in M_n(\complex)$ be the rank one projection onto the span of $\xi^{(j)}$, then
|
||||
Using the \hyperref[Spectral Theorem]{definition:linfty-functional-calculus}, let $\bracsn{\xi^{(j)}}_1^n \subset \complex^n$ be an orthonormal eigenbasis for $\Lambda$ with eigenvalues $\seqf{\lambda_j} \subset \complex$. For each $1 \le j \le n$, let $P_j \in M_n(\complex)$ be the rank one projection onto the span of $\xi^{(j)}$, then
|
||||
\[
|
||||
S_A(\Lambda) = \sum_{j = 1}^n \lambda_j S_A(P_{j}) \ge 0
|
||||
\]
|
||||
@@ -45,7 +45,7 @@
|
||||
|
||||
(2) $\Leftrightarrow$ (3): Let $q = \sum_{-N \le j \le N}\mu_j z^j$ and $\xi = (\mu_{-N}, \cdots, \mu_{N}) \in \complex^{2N+1}$, then
|
||||
\begin{align*}
|
||||
\dpn{Pq, q}{L^2(\mathbb{T}; \complex)} &= \sum_{-N \le j, k, l \le N}\dpn{\lambda_j z^j \cdot \mu_k z^k, \mu_l z^l}{L^2(\mathbb{T}; \complex)} \\
|
||||
\dpn{Pq, q}{L^2(\mathbb{T}; \complex)} &= \sum_{-n \le j \le n}\sum_{-N \le k, l \le N}\dpn{\lambda_j z^j \cdot \mu_k z^k, \mu_l z^l}{L^2(\mathbb{T}; \complex)} \\
|
||||
&= \sum_{-N \le k, l \le N}\dpn{\lambda_{l-k}\mu_k z^l, \mu_l z^l}{L^2(\mathbb{T}; \complex)} \\
|
||||
&= \sum_{-N \le k, l \le N} \lambda_{l - k}\mu_k\ol{\mu_l}= \sum_{k, l = 1}^{2N+1} \lambda_{l - k}\xi_{k}\ol{\xi_{l}} \\
|
||||
&= \angles{\paren{\sum_{i, j = 1}^{2N+1} E_{ij}\lambda_{i - j}}\xi, \xi} \\
|
||||
|
||||
Reference in New Issue
Block a user