Sum of g-frames in Hilbert C^{\ast}-modules

Abdellatif Lfounoune, Hafida Massit, Abdelilah Karara and Mohamed Rossafi

arXiv: 2312.14319v2 · PDF · 18 February 2025
Categories: math.FA
Licence: CC BY 4.0 The site shows the text.

Abstract

In this article, we study g-frames in Hilbert C^*-modules and investigate conditions under which the sum of two g-frames (or a g-frame and a g-Bessel sequence) remains a g-frame. We also address the stability of g-frames under certain perturbations and provide illustrative examples in the context of C^*-algebras. Our results unify and extend many of the existing theorems on g-frames, focusing on the invertibility of associated operators as a key condition for guaranteeing that sums of g-frames preserve the g-frame property.

Items 40

Definition

Definition 1.1

H1RO

[Kal] Let \mathcal{A} be a unital C^{\ast}- algebra and \mathcal{H} be a left \mathcal{A}- module, such that the linear structures of \mathcal{A} and \mathcal{H} are compatible. \mathcal{H} is a pre-Hilbert \mathcal{A} module if \mathcal{H} is equipped with an \mathcal{A}-valued inner product \langle \cdot, \cdot\rangle_{\mathcal{A}} : \mathcal{H} \times \mathcal{H}\rightarrow \mathcal{A} such that is sesquilinear, positive definite and respects the module action. In the other words,

1 - \langle x, x\rangle_{\mathcal{A}} \geq 0, \forall x\in \mathcal{H} and \langle x, x\rangle_{\mathcal{A}} = 0 if and only if x=0.
2 - \langle ax+y, z\rangle_{\mathcal{A}} =a\langle x,z\rangle_{\mathcal{A}} +\langle y,z\rangle_{\mathcal{A}} for all a\in \mathcal{A} and x,y,z \in \mathcal{H}
3 - \langle x,y \rangle_{\mathcal{A}} = \langle y,x \rangle_{\mathcal{A}} ^{\ast} for all x,y \in\mathcal{H}.

For x\in\mathcal{H}, we define \Vert x\Vert = \Vert \langle x,x \rangle_{\mathcal{A}} \Vert^{\frac{1}{2}}. If \mathcal{H} is complete with \Vert .\Vert, it is called a Hilbert \mathcal{A}-module or a Hilbert C^{\ast} -module over \mathcal{A}. For every a in C^{\ast}-algebra \mathcal{A}, we have \vert a \vert = (a^{\ast}a)^{\frac{1}{2}} and the \mathcal{A}-valued norm on \mathcal{H} is defined by \vert x\vert= \langle x,x\rangle_{\mathcal{A}} ^{\frac{1}{2}} for x\in\mathcal{H}.

Lemma

Lemma 1.2

8OFD

[Pas] Let \mathcal{H} be a Hilbert \mathcal{A}-module. If \mathcal{T}\in End_{\mathcal{A}}^{\ast}(\mathcal{H}), then

\langle \mathcal{T}x,\mathcal{T}x\rangle_{\mathcal{A}}\leq\|\mathcal{T}\|^{2}\langle x,x\rangle_{\mathcal{A}}, \forall x\in\mathcal{H}.

Lemma

Lemma 1.3

7SZA

[Ara] Let \mathcal{H} and \mathcal{K} two Hilbert \mathcal{A}-modules and \mathcal{T} \in \operatorname{End}^*(\mathcal{H}, \mathcal{K}). Then the following statements are equivalent

\mathcal{T} is surjective.
\mathcal{T}^* is bounded below with respect to norm, i.e., there is m>0 such that m\|x\| \leq\left\|\mathcal{T}^* x\right\| for all x \in \mathcal{K}.
\mathcal{T}^* is bounded below with respect to the inner product, i.e., there is m^{\prime}>0 such that m^{\prime}\langle x, x\rangle \leq\left\langle \mathcal{T}^* x, \mathcal{T}^* f\right\rangle for all x \in\mathcal{K}.

Used by Proof

Definition

Definition 1.4

R6ZG

[Kal] A collection \{x_{\zeta}\}_{\zeta\in \Theta}\subset \mathcal{H} is called a frame for \mathcal{H}. If there exist two positive constants A and B such that for all x\in \mathcal{H},

A\langle x,x\rangle_{\mathcal{A}}\leq \sum_{\zeta\in \Theta} \langle x,x_{\zeta}\rangle_{\mathcal{A}} \langle x_{\zeta},x\rangle_{\mathcal{A}} \leq B \langle f,f\rangle_{\mathcal{A}},

Definition

Definition 1.5

BB3P

[Kh] A sequence \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} is said a g-frame for \mathcal{H} with respect to \{\mathcal{H}_{\zeta}: \zeta\in \Theta\}. If there exist two positive constants A and B such that for x\in\mathcal{H},

A\langle x,x\rangle_{\mathcal{A}}\leq \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta}x\rangle_{\mathcal{A}} \leq B \langle x,x\rangle_{\mathcal{A}}.

The numbers A and B are called lower and upper bounds of the g-frame, respectively. If A = B = \nu, the g-frame is \nu-tight. If A= B=1, it is called a g-Parseval frame. If olny the right-hand inequality (1.5) is satisfied, the family \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} is called a g-Bessel sequence for \mathcal{H} with g-Bessel bound B.

Proposition

Proposition 2.1

0UVY

Let \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-Bessel sequence for \mathcal{H} with the g-frame operator S. Then, \{\Psi_{\zeta}\}_{\zeta\in \Theta} is a g -frame for \mathcal{H} with respect to \{\mathcal{H}_{\zeta}\}_{\zeta\in \Theta} if and only if S\geq \alpha I for some \alpha >0.

Used by Proof, Proof

Proof of Proposition 2.1

Suppose \{\Psi_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H}, we have for some \alpha >0,

\alpha\langle x,x\rangle_{\mathcal{A}}\leq \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta}x\rangle_{\mathcal{A}},\: \forall x \in \mathcal{H}.

For any x\in\mathcal{H}, we have

\sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta}x\rangle_{\mathcal{A}}=\sum_{\zeta\in \Theta} \langle \Psi_{\zeta}^{\ast}\Psi_{\zeta}x,x\rangle_{\mathcal{A}}=\langle Sx,x\rangle_{\mathcal{A}}.

Therefore, \alpha I\leq S.

Suppose \{\Psi_{\zeta}\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H} and there exists some \alpha>0 such that S\geq \alpha I. Thus, we have

\langle \alpha I x,x\rangle_{\mathcal{A}}\leq \langle S x,x\rangle_{\mathcal{A}}

for all x\in \mathcal{H}, this implies

\langle \alpha I x,x\rangle_{\mathcal{A}}\leq \langle \sum_{\zeta\in \Theta}\Psi_{\zeta}^{\ast}\Psi_{\zeta} x,x\rangle_{\mathcal{A}}.

Then, the lower condition holds:

\alpha \langle x,x\rangle_{\mathcal{A}} \leq \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta}x\rangle_{\mathcal{A}},\; \forall x\in\mathcal{H}.

Corollary

Corollary 2.2

QUYI

The g-Bessel sequence \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} with the g-frame operator S is a g-frame for \mathcal{H} if and only if S>0.

Used by Proof

Lemma

Lemma 2.3

554Q

Let \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-Bessel sequence for \mathcal{H}, with the synthesis operator \mathcal{T} and the g-frame operator S. Then \mathcal{T} is surjective if only if S>0.

Used by Proof

Proof of Lemma 2.3

Let \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-Bessel sequence for \mathcal{H}. Assume that \mathcal{T} is surjective, by Lemma 1.3 there is \alpha>0 such that for any x\in \mathcal{H},

\left\langle \mathcal{T}^* x, \mathcal{T}^* x\right\rangle_{\mathcal{A}}\geq\alpha\langle x, x\rangle_{\mathcal{A}}.

Therefore

\left\langle \mathcal{T} \mathcal{T}^* f, f\right\rangle_{\mathcal{A}}\geq\alpha\langle f, f\rangle_{\mathcal{A}}.

Hence S>\alpha I. So by Proposition 2.1 and Corollary 2.2 we conclude that S>0.

Conversely suppose that the g-frame operator S>0, by Corollary 2.2 \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} is a g-frame for \mathcal{H}. According to Theorem 3.2 in [Xia], we conclude that \mathcal{T} is surjective.

Uses Lemma 1.3, Proposition 2.1, Corollary 2.2

Theorem

Theorem 2.4

QBCV

Let \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-frame for \mathcal{H} with bounds A,\; B, with the g-frame operator S. If \Lambda \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}) such that ( I+\Lambda)^{\ast}S(I+\Lambda)\geq S. Then \{\Psi_{\zeta}+\Psi_{\zeta}\Lambda\}_{\zeta\in \Theta} is a g-frame for \mathcal{H}.

Proof of Theorem 2.4

Assume that \{\Psi_{\zeta}\}_{\zeta\in \Theta} be a g-frame and \Lambda \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}) is bounded. We have for any x\in\mathcal{H},

A\langle x,x\rangle_{\mathcal{A}} \leq \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta}x\rangle_{\mathcal{A}}\leq B\langle x,x\rangle_{\mathcal{A}}

and

\begin{aligned} \langle( \Psi_{\zeta}+\Lambda \Psi_{\zeta})x,(\Psi_{\zeta}+\Lambda \Psi_{\zeta})x\rangle_{\mathcal{A}} &= \langle \Psi_{\zeta}x,\Psi_{\zeta} x\rangle_{\mathcal{A}}+ \langle \Psi_{\zeta}\Lambda x,\Psi_{\zeta}\Lambda x\rangle_{\mathcal{A}} + \langle \Psi_{\zeta}x,\Psi_{\zeta}\Lambda x\rangle_{\mathcal{A}} \\&+ \langle \Psi_{\zeta}\Lambda x,\Psi_{\zeta} x\rangle_{\mathcal{A}}\\ &\leq 2 \left(\langle \Psi_{\zeta}x,\Psi_{\zeta} x\rangle_{\mathcal{A}}+ \langle \Psi_{\zeta}\Lambda x,\Psi_{\zeta}\Lambda x\rangle_{\mathcal{A}} \right) \end{aligned}

Therefore

\begin{aligned} \sum_{\zeta\in \Theta} \langle( \Psi_{\zeta}+\Lambda \Psi_{\zeta})x,(\Psi_{\zeta}+\Lambda \Psi_{\zeta})x\rangle_{\mathcal{A}} &\leq 2\left( \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}\Lambda x,\Psi_{\zeta}\Lambda x\rangle_{\mathcal{A}}\right)\\ &\leq 2B(1+\Vert \Lambda \Vert^{2})\langle x,x\rangle_{\mathcal{A}}. \end{aligned}

Thus, \{\Psi_{\zeta}+\Psi_{\zeta}\Lambda\}_{\zeta\in \Theta} is a g-Bessel sequence. Since \{\Psi_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H}, by Proposition 2.1, for some \alpha >0, we have S\geq \alpha I. Also, for every x\in \mathcal{H}, we get

\begin{aligned} \sum_{\zeta\in \Theta} ( \Psi_{\zeta}+\Lambda \Psi_{\zeta})^{\ast}(\Psi_{\zeta}+\Lambda \Psi_{\zeta})&= \sum_{\zeta\in \Theta} \left( \Psi_{\zeta}^{\ast}\Psi_{\zeta}+ \Psi_{\zeta}^{\ast}\Psi_{\zeta}\Lambda \right. +\left. \Lambda^{\ast}\Psi_{\zeta}^{\ast}\Psi_{\zeta}+ \Lambda^{\ast}\Psi_{\zeta}^{\ast}\Psi_{\zeta}\Lambda \right) \\ & = S+S\Lambda+\Lambda^{\ast}S+\Lambda^{\ast}S\Lambda\\ &=( I+\Lambda)^{\ast}S(I+\Lambda) \end{aligned}

Then,

\mathcal{P}= S+S\Lambda+\Lambda^{\ast}S+\Lambda^{\ast}S\Lambda=( I+\Lambda)^{\ast}S(I+\Lambda)\geq S

Hence the operator \mathcal{P} is the g-frame operator of \{\Psi_{\zeta}+\Psi_{\zeta}\Lambda\}_{\zeta\in \Theta}. So, \mathcal{P}\geq \alpha I for some \alpha>0. Using proposition 2.1 we obtain \{\Psi_{\zeta}+\Psi_{\zeta}\Lambda\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{\mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Uses Proposition 2.1

Theorem

Theorem 2.5

FVFU

Let \Psi= \{\Psi_{\zeta}\}_{\zeta\in \Theta} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} are two g-Bessel sequences for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta} and the g-frame operators S_{\Psi}, S_{\Delta}, respectively. If M, N \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}), then the following conditions are equivalents:

(1) \{\Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta} is a g-frame for \mathcal{H}.
(2) M^{\ast}\mathcal{T}_{\Psi}+N\mathcal{T}_{\Delta} is surjective.
(3) S= M^{\ast}S_{\Psi}M+M^{\ast}\mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}N+N^{\ast}\mathcal{T}_{\Delta}\mathcal{T}^{\ast}_{\Psi}M+N^{\ast}S_{\Delta}N>0.

Then, S is the g-frame operator of \{\Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta}.

Used by Proof, Proof

Proof of Theorem 2.5

Let D_{1}, D_{2} be g-Bessel bounds of \Psi and \Delta, respectively. Since M, N are bounded, for any x\in \mathcal{H}, we have

\begin{aligned} \sum_{\zeta\in \Theta} \langle( \Psi_{\zeta}M+ \Delta_{\zeta}N)x,( \Psi_{\zeta}M+ \Delta_{\zeta}N)x\rangle_{\mathcal{A}}&\leq 2\left(\sum_{\zeta\in \Theta} \langle \Delta_{\zeta}Nx,\Delta_{\zeta}N x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}M x,\Psi_{\zeta}M x\rangle_{\mathcal{A}}\right)\\ &\leq 2(D_{1} \Vert M \Vert^{2}+D_{2} \Vert N \Vert^{2})\langle x,x \rangle_{\mathcal{A}}. \end{aligned}

Then, \{\Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H}.

If \mathcal{T} is the synthesis operator of \{\Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta}, then we have

\mathcal{T}^{\ast}x = \{(\Psi_{\zeta}M+\Delta_{\zeta}N)x\}_{\zeta\in \Theta} =\mathcal{T}_{\Psi}^{\ast} M x+\mathcal{T}_{\Delta}^{\ast} N x,\;\forall x\in \mathcal{H}.

i.e. \mathcal{T}= M^{\ast}\mathcal{T}_{\Psi} +N^{\ast}\mathcal{T}_{\Delta}.

By Lemma 2.3, the conditions (1) and (2) are equivalents. Moreover, we have

S=\mathcal{T} \mathcal{T}^{\ast}= M^{\ast}S_{\Psi}M+M\mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}N+ N^{\ast}\mathcal{T}_{\Delta}\mathcal{T}^{\ast}_{\Psi}M+N^{\ast}S_{\Delta}N>0.

Therefore, the conditions (2) and (3) are equivalent.

Uses Lemma 2.3

Corollary

Corollary 2.6

22XZ

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} are two g-Bessel sequences for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta} and the g-frame operators S_{\Psi}, S_{\Delta}, respectively. If \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta} is positive, then \{\Psi_{\zeta}+\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Proof of Corollary 2.6

If M=N=I in Theorem 2.5, then the g-frame operator of the g-Bessel sequence \{\Psi_{\zeta}+\Delta_{\zeta}\}_{\zeta\in \Theta} is as follows:

\begin{aligned} S&= \mathcal{T}_{\Psi}\mathcal{T}_{\Psi}^{\ast}+\mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}+\mathcal{T}_{\Delta}\mathcal{T}^{\ast}_{\Psi} \\ &= S_{\Psi}+\mathcal{T}_{\Psi}\mathcal{T}_{\Delta}^{\ast}+\mathcal{T}_{\Delta}\mathcal{T}_{\Psi}^{\ast}+S_{\Delta}>0. \end{aligned}

Uses Theorem 2.5

Theorem

Theorem 2.7

NEF9

Let \{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-frame for \mathcal{H} with bound D and D^{\prime}, let \{\Delta_{\zeta}\}_{\zeta\in \Theta} be a g-Bessel sequence for \mathcal{H} with Bessel bound D_{\Delta}. Suppose that \{\theta_{\zeta}\}_{\zeta\in \Theta} and \{\delta_{\zeta}\}_{\zeta\in \Theta} are two sequences from the algebra \mathcal{A} such that 0< A<\vert \theta_{\zeta}\vert^{2} , \vert \delta_{\zeta}\vert^{2}<B< \infty. If BD_{\Delta} < AD, then \{ \theta_{\zeta}\Psi_{\zeta}+\delta_{\zeta}\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Proof of Theorem 2.7

We have for any x\in\mathcal{H},

And for every x\in\mathcal{H}, we obtain

\begin{aligned} \Vert \sum_{\zeta\in \Theta} \langle( \theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta} \Delta_{\zeta})x,(\theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta}\Delta_{\zeta})x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}} &\geq \Vert \sum_{\zeta\in \Theta} \langle (\theta_{\zeta}\Psi_{\zeta})x,(\theta_{\zeta}\Psi_{\zeta}) x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}} \\ &-\Vert \sum_{\zeta\in \Theta} \langle( \delta_{\zeta} \Delta_{\zeta})x,( \delta_{\zeta}\Delta_{\zeta})x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}}\\ &\geq (\sqrt{AD}-\sqrt{BD_{\Delta}})\Vert x \Vert. \end{aligned}

Therefore \{ \theta_{\zeta}\Psi_{\zeta}+\delta_{\zeta}\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Theorem

Theorem 2.8

J5OV

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} be two g-frames for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta}, respectively. Suppose that \{\theta_{\zeta}\}_{\zeta\in \Theta} and \{\delta_{\zeta}\}_{\zeta\in \Theta} are two sequences from the algebra \mathcal{A} such that 0< A<\vert \theta_{\zeta}\vert^{2} , \vert \delta_{\zeta}\vert^{2}<B< \infty. If \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta} is positive, then \{\theta_{\zeta}\Psi_{\zeta}+\delta_{\zeta}\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Used by Proof

Proof of Theorem 2.8

As in the proof of Theorem 2.7 \{ \theta_{\zeta}\Psi_{\zeta}+\delta_{\zeta}\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}. For every x\in\mathcal{H}

\begin{aligned} \sum_{\zeta\in \Theta} \langle( \theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta} \Delta_{\zeta})x,(\theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta}\Delta_{\zeta})x\rangle_{\mathcal{A}}&= \sum_{\zeta\in \Theta} \langle \delta_{\zeta}\Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta} \Psi_{\zeta} x\rangle_{\mathcal{A}}\\&+ \sum_{\zeta\in \Theta} \langle \theta_{\zeta}\Psi_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \delta_{\zeta}\Delta_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\\ & \geq A\left( \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x, \Psi_{\zeta} x\rangle_{\mathcal{A}}\right)\\ &+ \min_{\zeta \in \Theta} \lbrace\lvert \delta_{\zeta}\theta_{\zeta}\rvert,\lvert \theta_{\zeta}\delta_{\zeta}\rvert \rbrace\left( \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Psi_{\zeta} x\rangle_{\mathcal{A}} \right)\\ &=A \left( \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x, \Psi_{\zeta} x\rangle_{\mathcal{A}} \right)\\ &+\min_{\zeta \in \Theta}\lbrace\lvert \delta_{\zeta}\theta_{\zeta}\rvert,\lvert \theta_{\zeta}\delta_{\zeta}\rvert \rbrace \left(\langle \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}x,x\rangle_{{\mathcal{A}}} +\langle \mathcal{T}_{\Delta}\mathcal{T}^{\ast}_{\Psi}x,x\rangle_{{\mathcal{A}}} \right). \end{aligned}

Then, we have

\sum_{\zeta\in \Theta} \langle( \theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta} \Delta_{\zeta})x,(\theta_{\zeta} \Psi_{\zeta}+ \delta_{\zeta}\Delta_{\zeta})x\rangle_{\mathcal{A}}\geq A \left( \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x, \Psi_{\zeta} x\rangle_{\mathcal{A}}+\sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\right)\geq A(\alpha+\beta)\langle x,x\rangle_{\mathcal{A}}.

Where \alpha and \beta is a lower g-frames bounds of \Psi and \Delta respectively.

Corollary

Corollary 2.9

9ZVQ

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} are two tight g-frame for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta} respectively. If \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}=0 , then \{\Psi_{\zeta}+\Delta_{\zeta}\}_{\zeta\in \Theta} is a tight g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Proof of Corollary 2.9

Using the proof of Theorem 2.8, for every x\in\mathcal{H} we have

\begin{aligned} \sum_{\zeta\in \Theta} \langle( \Psi_{\zeta}+ \Delta_{\zeta})x,( \Psi_{\zeta}+ \Delta_{\zeta})x\rangle_{\mathcal{A}} & = ( \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}x, \Psi_{\zeta} x\rangle_{\mathcal{A}})\\ &=( \alpha_{1}+\alpha_{2})\langle x,x\rangle_{\mathcal{A}}. \end{aligned}

Where \sum\limits_{\zeta\in \Theta} \langle \Psi_{\zeta}x,\Psi_{\zeta} x\rangle_{\mathcal{A}}= \alpha_{1}\langle x,x\rangle_{\mathcal{A}} and \sum\limits_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}= \alpha_{2}\langle x,x\rangle_{\mathcal{A}}.

Uses Theorem 2.8

Theorem

Theorem 2.10

K05Y

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a g-frame for \mathcal{H} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} be a g-Bessel sequence for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta}, respectively and \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta} is positive. If \Lambda\in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}) is an isometry, then \{\Psi_{\zeta}+\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta\in \Theta}\}_{\zeta\in \Theta}.

Proof of Theorem 2.10

Similar to the proof of Theorem 2.5, let A be a lower g-frame bound of \Psi and \{(\Psi_{\zeta}+\Delta_{\zeta})\Lambda\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}. For every x\in\mathcal{H}, we have

\begin{aligned} \sum_{\zeta\in \Theta} \langle (\Psi_{\zeta}+\Delta_{\zeta})\Lambda,(\Psi_{\zeta}+\Delta_{\zeta})\Lambda \rangle_{{\mathcal{A}}} &= \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}\Lambda x,\Delta_{\zeta} \Lambda x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}\Lambda x, \Psi_{\zeta} \Lambda x\rangle_{\mathcal{A}}\\&+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}\Lambda x,\Delta_{\zeta}\Lambda x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Delta_{\zeta} \Lambda x,\Psi_{\zeta} \Lambda x\rangle_{\mathcal{A}}\\ &= \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}\Lambda x,\Delta_{\zeta}\Lambda x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}\Lambda x, \Psi_{\zeta} \Lambda x\rangle_{\mathcal{A}}\\ &+\langle \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}\Lambda x,\Lambda x\rangle_{{\mathcal{A}}} +\langle \mathcal{T}_{\Delta}\mathcal{T}^{\ast}_{\Psi}\Lambda x,\Lambda x\rangle_{{\mathcal{A}}} . \end{aligned}

So,

\sum_{\zeta\in \Theta} \langle (\Psi_{\zeta}+\Delta_{\zeta})\Lambda,(\Psi_{\zeta}+\Delta_{\zeta})\Lambda \rangle_{{\mathcal{A}}} \geq \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}\Lambda x, \Psi_{\zeta} \Lambda x\rangle_{\mathcal{A}} \geq A \Vert \Lambda x \Vert^{2} =A\langle x,x\rangle_{\mathcal{A}},

Theorem

Theorem 2.11

5UUG

Let \Psi=\{\Psi_{\zeta} \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H},\mathcal{H}_{\zeta}): \zeta\in \Theta\} be a g-frame for \mathcal{H} with bounds D and D^{\prime}, let \{\Delta_{\zeta}\}_{\zeta\in \Theta} be a g-Bessel sequence for \mathcal{H} with Bessel bound D_{\Delta}. If M,N\in\operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}) such that for any x \in \mathcal{H}, \Vert Nx\Vert> \lambda \Vert x\Vert for some \lambda>0 and D_{\Delta}<D, then \{ \Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta} is a g- frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Proof of Theorem 2.11

Similar to the proof of Theorem 2.5, \{(\Psi_{\zeta}+\Delta_{\zeta})M\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}. For any x\in\mathcal{H}, we obtain

\begin{aligned} \Vert \sum_{\zeta\in \Theta} \langle (\Psi_{\zeta}+\Delta_{\zeta})M,(\Psi_{\zeta}+\Delta_{\zeta})M \rangle_{{\mathcal{A}}} \Vert^{\frac{1}{2}} &\geq \Vert\sum_{\zeta\in \Theta} \langle \Psi_{\zeta}M x, \Psi_{\zeta} Mx\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}} - \Vert\sum_{\zeta\in \Theta} \langle \Delta_{\zeta}Nx,\Delta_{\zeta} N x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}} \\&\geq \sqrt{D} \Vert M x\Vert -\sqrt{D_{\Delta}} \Vert N x\Vert \\ &\geq (\sqrt{D}-\sqrt{D_{\Delta}})\Vert Nx \Vert \\ &\geq \lambda(\sqrt{D}-\sqrt{D_{\Delta}})\Vert x \Vert . \end{aligned}

Therefore, \lambda^{2}(\sqrt{D}-\sqrt{D_{\Delta}})^{2} is a lower g-frame bound for \{(\Psi_{\zeta}+\Delta_{\zeta})M\}_{\zeta\in \Theta}.

Corollary

Corollary 2.12

5USG

Theorem

Theorem 2.13

1LHR

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a \alpha_{1}-tight g-frame for \mathcal{H} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} be a \alpha_{2}-tight g-frame for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta}, respectively, and \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}=0. If M,N \in \operatorname{End}_{\mathcal{A}}^{\ast}(\mathcal{H}), then \{\Psi_{\zeta}M+\Delta_{\zeta}N\}_{\zeta\in \Theta} is a \alpha-tight g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta\in \Theta}\}_{\zeta\in \Theta} if and only if \alpha_{1}M^{\ast}M+\alpha_{2}N^{\ast}N= \alpha I, for some \alpha>0.

Proof of Theorem 2.13

According to the proof of Theorem 2.5 \{(\Psi_{\zeta}+\Delta_{\zeta})M\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H}. Since \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}=0, for any x\in\mathcal{H}, we get \sum_{\zeta\in \Theta} \Psi_{\zeta}^{\ast}\Delta_{\zeta}x =0.

For every x\in \mathcal{H}, we have

\sum_{\zeta\in \Theta} \langle \Psi_{\zeta}M x, \Delta_{\zeta}Nx\rangle_{{\mathcal{A}}} = \sum_{\zeta\in \Theta} \langle M x,\Psi_{\zeta}^{\ast} \Delta_{\zeta}Nx\rangle_{{\mathcal{A}}} =\langle M x,\sum_{\zeta\in \Theta} \Psi_{\zeta}^{\ast}\Delta_{\zeta}Nx\rangle_{{\mathcal{A}}}= 0.

We obtain,

\begin{aligned} \sum_{\zeta\in \Theta} \langle (\Psi_{\zeta}M+\Delta_{\zeta}N)x,(\Psi_{\zeta}M+\Delta_{\zeta}N)x \rangle_{{\mathcal{A}}} &= \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}Nx,\Delta_{\zeta} N x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}M x, \Psi_{\zeta} M x\rangle_{{\mathcal{A}}} \end{aligned}

Then, we get

\begin{aligned} \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}Nx,\Delta_{\zeta} N x\rangle_{\mathcal{A}}+ \sum_{\zeta\in \Theta} \langle \Psi_{\zeta}M x, \Psi_{\zeta} M x\rangle_{{\mathcal{A}}}& = \alpha_{1}\Vert M x\Vert ^{2}+ \alpha_{2} \Vert Nx \Vert^{2} \\ &= \langle \alpha_{1}M^{*}M x,x\rangle_{{\mathcal{A}}}+ \langle \alpha_{2}N^{*}Nx,x\rangle_{{\mathcal{A}}}\\ &= \langle \alpha_{1}M^{*}M x+\alpha_{2} N^{*}Nx,x\rangle_{{\mathcal{A}}}\\ &= \langle \alpha Ix,x\rangle_{{\mathcal{A}}}\\ &= \alpha \langle x,x\rangle_{{\mathcal{A}}}. \end{aligned}

Uses Theorem 2.5

Corollary

Corollary 2.14

EZHP

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a \alpha_{1}-tight g-frame for \mathcal{H} and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} be a \alpha_{2}-tight g-frame for \mathcal{H} for \mathcal{H} with the synthesis operators \mathcal{T}_{\Psi}, \mathcal{T}_{\Delta}, respectively. If \mathcal{T}_{\Psi}\mathcal{T}^{\ast}_{\Delta}=0, then \{\Psi_{\zeta}+\Delta_{\zeta}\}_{\zeta\in \Theta} is a \alpha-tight g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta\in \Theta}\}_{\zeta\in \Theta} if and only if \alpha_{1}+\alpha_{2}= \alpha.

Proposition

Proposition 2.15

FQ7T

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a g-frame for \mathcal{H} with bounds C, D and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta}. Let \{\theta_{\zeta}\}_{\zeta\in \Theta} and \{\delta_{\zeta}\}_{\zeta\in \Theta} are two sequences from the algebra \mathcal{A} such that 0< A<\vert \theta_{\zeta}\vert^{2} , \vert \delta_{\zeta}\vert^{2}<B< \infty and 0< \alpha_{1},\alpha_{2}<1 such that for any x\in \mathcal{H}, we have

\begin{aligned} \left( \Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert - \Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}&\leq \alpha_{1} \left( \Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}\\ &+\alpha_{2}\left(\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}. \end{aligned}

Then \{ \Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H}.

Proof of Proposition 2.15

For any x\in\mathcal{H},

\begin{aligned} \left(\Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}} &\leq \left( \Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}} \Vert - \Vert \sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert \right)^{\frac{1}{2}}\\ &+\left(\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}\\ &\leq \alpha_{1} \left( \Vert\sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}} +\alpha_{2}\left(\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}\\ &+\left(\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\right)^{\frac{1}{2}}. \end{aligned}

Then, we obtain

\left(1-\alpha_{1}\right) \left( \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\right)^{\frac{1}{2}}\leq \left(1+\alpha_{2}\right)\left(\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\right)^{\frac{1}{2}}.

So, we have

CA\left(1-\alpha_{1}\right)^{2} \langle x, x\rangle_{\mathcal{A}}\leq B \left(1+\alpha_{2}\right)^{2}\left(\sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\right).

This implies that \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}} \geq \dfrac{CA\left(1-\alpha_{1}\right)^{2}}{ B \left(1+\alpha_{2}\right)^{2}}\langle x, x\rangle_{\mathcal{A}}, \forall x\in\mathcal{H}.

On the other hand, for any x\in\mathcal{H} we get

\left(1-\alpha_{2}\right) \left( \sum_{\zeta\in \Theta} \langle \delta_{\zeta}\Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\right)^{\frac{1}{2}}\leq \left(1+\alpha_{1}\right)\left(\sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\ \theta_{\zeta} \Psi_{\zeta}x\rangle_{\mathcal{A}}\right)^{\frac{1}{2}}.

So, we obtain

A\left(1-\alpha_{2}\right)^{2}\left(\sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\right)\leq \langle x, x\rangle_{\mathcal{A}} DB \left(1+\alpha_{2}\right)^{2}.

Therefore, \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\leq \dfrac{BD \left(1+\alpha_{2}\right)^{2}}{A\left(1-\alpha_{1}\right)^{2} } \langle x, x\rangle_{\mathcal{A}}, \forall x\in\mathcal{H}.

Theorem

Theorem 2.16

RZUU

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a g-frame for \mathcal{H} with bounds C, D and \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta}. Let \{\theta_{\zeta}\}_{\zeta\in \Theta} and \{\delta_{\zeta}\}_{\zeta\in \Theta} are two sequences from the algebra \mathcal{A} such that 0< A< \theta_{\zeta}^{2} , \delta_{\zeta}^{2}<B< \infty and 0< \alpha_{1},\alpha_{2}<1 such that for any x\in \mathcal{H},

\sum_{\zeta\in \Theta} \langle (\theta_{\zeta} \Psi_{\zeta}-\delta_{\zeta} \Delta_{\zeta})x,(\theta_{\zeta}\Psi_{\zeta}-\delta_{\zeta} \Delta_{\zeta}) x\rangle_{\mathcal{A}}\leq \alpha_{1} ( \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}) +\alpha_{2}( \sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}})

then \{ \Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta\in \Theta}\}_{\zeta\in \Theta}.

Proof of Theorem 2.16

For x \in \mathcal{H}, we have

\begin{aligned} ( \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}})^{\frac{1}{2}}&\leq \Vert \sum_{\zeta\in \Theta} \langle (\theta_{\zeta} \Psi_{\zeta}-\delta_{\zeta} \Delta_{\zeta})x,(\theta_{\zeta}\Psi_{\zeta}-\delta_{\zeta} \Delta_{\zeta}) x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}} + \Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert^{\frac{1}{2}}\\ &\leq( \alpha_{1} \Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert +\alpha_{2}\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}\\ &+( \Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}\\ &\leq (\alpha_{1}\Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}+ 2(\alpha_{2}\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}. \end{aligned}

Then, for any x\in\mathcal{H} we get

(1-\sqrt{\alpha_{1}})( \Vert \sum_{\zeta\in \Theta} \langle \theta_{\zeta} \Psi_{\zeta}x,\theta_{\zeta}\Psi_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}\leq (1-2\sqrt{\alpha_{2}})\Vert\sum_{\zeta\in \Theta} \langle \delta_{\zeta} \Delta_{\zeta}x,\delta_{\zeta}\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert)^{\frac{1}{2}}.

We obtain,

\Vert \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\geq \dfrac{CA (1+2\sqrt{\alpha_{2}})^{2}}{B(1-\sqrt{\alpha_{1}})^{2} } \langle x, x\rangle_{\mathcal{A}}, \forall x\in\mathcal{H}.

Similarly, for any x\in\mathcal{H},

\Vert \sum_{\zeta\in \Theta} \langle \Delta_{\zeta}x,\Delta_{\zeta} x\rangle_{\mathcal{A}}\Vert\geq \dfrac{DB (1+2\sqrt{\alpha_{2}})^{2}}{C(1-\sqrt{\alpha_{1}})^{2} } \langle x, x\rangle_{\mathcal{A}}, \forall x\in\mathcal{H}.

Theorem

Theorem 2.17

AU37

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a g-frame for \mathcal{H} with bounds C, D (D>1) and assume that \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} is a family so that for any J\subset \Theta with \vert J\vert< \infty, we have

\Vert \sum_{\zeta\in J}( \Psi_{\zeta}^{*}\Psi_{\zeta}-\Delta_{\zeta})x \Vert \leq \dfrac{C}{D}\Vert x\Vert, \;\forall x\in\mathcal{H}.

Then \{ \Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with bounds \dfrac{1}{C}(\dfrac{D+1}{D} ), \dfrac{C+D^{2}}{D}.

Used by Proof

Proof of Theorem 2.17

For any x\in\mathcal{H} we have

\begin{aligned} \Vert \sum_{\zeta\in J} \Psi_{\zeta}^{*}\Psi_{\zeta}x\Vert& = \sup_{y\in\mathcal{H}, \Vert y\Vert =1}\Vert \langle \sum_{\zeta\in J} \Psi_{\zeta}^{*}\Psi_{\zeta}x, y\rangle_{\mathcal{A}} \Vert\\ &= \sup_{y\in\mathcal{H}, \Vert y\Vert =1}\Vert \sum_{\zeta\in J}\langle \Psi_{\zeta}x, \Psi_{\zeta}y\rangle_{\mathcal{A}} \Vert\\ &\leq \sup_{y\in\mathcal{H}, \Vert y\Vert =1}\Vert \sum_{\zeta\in J}\langle \Psi_{\zeta}x, \Psi_{\zeta}x \rangle_{\mathcal{A}} \Vert^{\frac{1}{2}} \Vert \sum_{\zeta\in J}\langle \Psi_{\zeta}y, \Psi_{\zeta}y\rangle_{\mathcal{A}} \Vert^{\frac{1}{2}}\\ &\leq \sqrt{D} \Vert \sum_{\zeta\in J}\langle \Psi_{\zeta}x, \Psi_{\zeta}x\rangle_{\mathcal{A}} \Vert^{\frac{1}{2}}\\ &\leq D \Vert x\Vert . \end{aligned}

So we obtain

\Vert \sum_{\zeta\in J} \Delta_{\zeta}^{*}\Delta_{\zeta}x\Vert\leq (\dfrac{C}{D}+D) \Vert x\Vert.

Hence, the series \sum_{\zeta\in \Theta}\Delta_{\zeta}^{*}\Delta_{\zeta}x converge. We define

\mathcal{K}(x)= \sum_{\zeta\in \Theta} \Delta_{\zeta}^{*}\Delta_{\zeta}x ,\;\forall x\in\mathcal{H}.

This operator is well defined and bounded on \mathcal{H} and \Vert \mathcal{K} \Vert\leq \dfrac{C}{D}+D.

Also for x\in\mathcal{H}

\begin{aligned} \Vert \sum_{\zeta\in \Theta} \Delta_{\zeta}x\Vert^{2} &= \Vert \sum_{\zeta\in \Theta}\langle \Delta_{\zeta}x,\Delta_{\zeta}x\rangle_{\mathcal{A}}\Vert\\ &\leq \Vert \mathcal{K} \Vert \Vert x\Vert ^{2}. \end{aligned}

That is \{\Delta_{\zeta}\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H}. If S is the g-frame operator of \{\Psi_{\zeta}\}_{\zeta\in \Theta}, we get for all x\in\mathcal{H}

\Vert Sx-\mathcal{K} x\Vert \leq \dfrac{C}{D} \Vert x\Vert.

Then, \Vert x- \mathcal{K} S^{-1}x \Vert \leq \dfrac{1}{D}\Vert x\Vert \Rightarrow \Vert I- \mathcal{K} S^{-1} \Vert \leq \dfrac{1}{D}\leq 1. Hence, \mathcal{K} S^{-1} is invertible and so is \mathcal{K}. Therefore, \{\Delta_{\zeta}\}_{\zeta} is a g-frame for \mathcal{H}. Also, we have for x\in\mathcal{H}

\Vert \mathcal{K} ^{-1}\Vert \leq \Vert S^{-1}\Vert \Vert S\mathcal{K} ^{-1}\Vert \leq \dfrac{1}{C}(\dfrac{D+1}{D}).

Therefore, \dfrac{1}{C}(\dfrac{D+1}{D}), \dfrac{C}{A}+D are bounds for \{\Delta_{\zeta}\}_{\zeta}.

Corollary

Corollary 2.18

2M2N

Let \Psi= \{\Psi_{\zeta}\} _{\zeta\in \Theta} be a g-frame for \mathcal{H} with bounds C, D (0<\alpha<D). Suppose \Delta= \{\Delta_{\zeta}\}_{\zeta\in \Theta} such that for x\in\mathcal{H},

\Vert \sum_{\zeta\in \Theta}\left( \Psi_{\zeta}^{*}\Psi_{\zeta}-\Delta_{\zeta}^{*}\Delta_{\zeta}\right) x \Vert\leq \alpha \Vert x \Vert.

Then \{ \Delta_{\zeta}\}_{\zeta\in \Theta} is a g-frame for \mathcal{H} with respect to \{ \mathcal{H}_{\zeta}\}_{\zeta\in \Theta}.

Proof of Corollary 2.18

Similarly the proof of the Theorem 2.17, if we take

\mathcal{K} (x)= \sum_{\zeta\in \Theta} \Delta_{\zeta}^{*}\Delta_{\zeta}x ,\;\forall x\in\mathcal{H}.

Then \{ \Delta_{\zeta}\}_{\zeta\in \Theta} is a g-Bessel sequence for \mathcal{H}, we have

\Vert Sx-\mathcal{K} x\Vert \leq \alpha \Vert x\Vert .

Then

\Vert x-\mathcal{K} S^{-1} x \Vert \leq \alpha \Vert S^{-1}\Vert \Vert x\Vert \leq \alpha \dfrac{1}{C} \Vert x \Vert, \; \forall x\in\mathcal{H}.

This implies that

\Vert I-\mathcal{K} S^{-1}\Vert <1.

So, \mathcal{K} S^{-1} is invertible and \mathcal{K} is also invertible.

Hence, g-Bessel sequence \{\Delta_{\zeta}\}_{\zeta} is a g-frame for \mathcal{H} with respect to \{\mathcal{H}\}_{\zeta\in \Theta}.

Uses Theorem 2.17

Example

Example 3.1

X0A8

Let \mathcal{A} = M_2(\mathbb{C}) be the 2\times 2 complex matrices, which is a unital C^*-algebra with the usual operator norm and ^*-operation. Consider the Hilbert \mathcal{A}-module \mathcal{H} = \mathcal{A}^2 (i.e., column 2-vectors over \mathcal{A}) with the standard \mathcal{A}-valued inner product

\langle (a_1,a_2)^T,\;(b_1,b_2)^T\rangle_{\mathcal{A}} \;=\; a_1^* b_1 + a_2^* b_2,

where addition and multiplication are in M_2(\mathbb{C}).

Define two families \{\Lambda_1,\Lambda_2\} and \{\Theta_1,\Theta_2\} as follows. For x\in\mathcal{H} (which is a pair (x_1,x_2)^T with each x_i\in M_2(\mathbb{C})):

\Lambda_1(x) = \bigl(x_1,\,0\bigr)^T, \quad \Lambda_2(x) = \bigl(0,\,x_2\bigr)^T,

\Theta_1(x) = \bigl(x_2,\,0\bigr)^T, \quad \Theta_2(x) = \bigl(0,\,x_1\bigr)^T.

One checks that each \{\Lambda_1,\Lambda_2\} and \{\Theta_1,\Theta_2\} forms a g-frame for \mathcal{H} (they essentially “capture” each component of (x_1,x_2)^T in disjoint slots).

Now consider the sum \{\Lambda_1+\Theta_1,\;\Lambda_2+\Theta_2\}. One quickly sees that

(\Lambda_1+\Theta_1)(x) \;=\; \bigl(x_1+x_2,\;0\bigr)^T, \quad (\Lambda_2+\Theta_2)(x) \;=\;\bigl(0,\;x_2+x_1\bigr)^T.

Case 1: If x_1 and x_2 are such that x_1 = -x_2, then (\Lambda_1+\Theta_1)(x)=0 and (\Lambda_2+\Theta_2)(x)=0, so the sum has no lower bound and is not a g-frame in that arrangement.

Case 2: If we modify one family by a suitable invertible multiplier L\in \mathrm{End}_\mathcal{A}^*(\mathcal{H}) (for instance, letting L act diagonally on the \mathcal{A}^2 coordinate), then positivity or invertibility conditions can ensure no such complete cancellation occurs, restoring a uniform lower bound.

Hence, this example demonstrates how the sums of two g-frames may fail to be a g-frame unless additional constraints (like invertibility or positivity) are imposed, exactly as our theorems suggest.

Example

Example 3.2

Q0OE

Let \mathcal{A} = C([0,1]) be the C^*-algebra of continuous complex-valued functions on [0,1]. We consider the standard free module \mathcal{H} = \mathcal{A}^m for some m\in \mathbb{N}. An element x\in \mathcal{H} can be viewed as a vector (f_1,\ldots,f_m)^T with each f_j\in C([0,1]). The \mathcal{A}-valued inner product is

\langle x,y\rangle_{\mathcal{A}}(\alpha) \;=\; \sum_{j=1}^m \overline{f_j(\alpha)}\, g_j(\alpha),

for x=(f_1,\dots,f_m), y=(g_1,\dots,g_m) in \mathcal{H}, where (\langle x,y\rangle_{\mathcal{A}})(\alpha) is a continuous function of \alpha\in [0,1].

Define, for instance, \Lambda_1(x) = (f_1, 0,\dots,0), \Lambda_2(x) = (0,f_2,0,\dots,0), etc. so that \{\Lambda_i\}_{i=1}^m collectively “read off” the coordinates. This forms a g-frame (essentially a standard basis in \mathcal{A}^m). Next, define \Theta_1,\ldots,\Theta_m similarly but perhaps with a shift: \Theta_j(x)=(0,\dots,0,f_j,0,\dots). Each family alone is a g-frame. However, if we sum them coordinate-wise, \Lambda_j + \Theta_j, there may be points \alpha\in[0,1] at which f_j(\alpha) and g_j(\alpha) combine destructively (e.g. f_j(\alpha) + g_j(\alpha)=0), losing the lower bound in the entire module.

On the other hand, if we multiply one family by an invertible element in \mathrm{End}_\mathcal{A}^*(\mathcal{H}) (for example, a bounded diagonal operator with no vanishing entries over [0,1]), then we can often ensure positivity of the new g-frame operator. This aligns with Theorem 2.5, requiring that a certain operator be invertible in the Hilbert C^*-module sense.

Source and licence

The text on this page is from arXiv:2312.14319v2 by Abdellatif Lfounoune, Hafida Massit, Abdelilah Karara and Mohamed Rossafi, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2312/2312.14319.jsonl.