Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Definition of the problem

Let $\mathcal{H}$ be a separable Hilbert space and $J\subset\mathbb{N}$ an index set. Let $\Phi:=\left(\varphi_{j}\right)_{j\in J}\subset\mathcal{H}$ be a frame for $\mathcal{H}$, i.e. $\exists A,B>0\,\forall x\in\mathcal{H},\, A\left\Vert x\right\Vert ^{2}\leq\sum_{j\in J}\left|\left\langle x,\varphi_{j}\right\rangle \right|^{2}\leq B\left\Vert x\right\Vert ^{2}$.

Show that the closed linear span $cls\left\{ \varphi_{j}:j\in J\right\} :=\overline{span\left\{ \varphi_{j}:j\in J\right\} }$ coincides with $\mathcal{H}$. Furthermore, show that $\left(\left\Vert \varphi_{j}\right\Vert \right)_{j\in J}$ is always bounded.

My efforts

For the first part, I have tried to prove that the synthesis operator $T_{\Phi}^{\star}:\ell_{2}\left(J\right)\rightarrow\mathcal{H}$ is surjective. We know that the analysis operator $T_{\Phi}:\mathcal{H}\rightarrow\ell_{2}\left(J\right)$ of $\Phi$ is injective. Since the synthesis operator is the adjoint operator of the analysis operator, we have that $$ (\star)\ \forall x\in\mathcal{H},\, y\in\ell_{2}\left(J\right):\quad\left\langle T_{\Phi}x,y\right\rangle _{\ell_{2}\left(J\right)}=\left\langle x,T_{\Phi}^{\star}y\right\rangle _{\mathcal{H}}. $$

From this, I derive that $T_{\Phi}^{\star}$ has the form $T_{\Phi}^{\star}\left(c_{j}\right)=\sum_{j\in J}c_{j}\varphi_{j},\,\left(c_{j}\right)\in\ell_{2}\left(J\right)$.

Since $\left(\star\right)$ must hold, we have that $\left(\star\star\right)\ \ker T_{\Phi}=\left(ran\, T_{\Phi}^{\star}\right)^{\perp}.$ And since $T_{\Phi}$ is injective, we have that its kernel is the trivial kernel, i.e. $\ker T_{\Phi}=\left\{ 0\right\} $. From this point, I do not see how to use this to prove $\left(\star\star\right)$.

After having shown that the synthesis operator $T_{\Phi}^{\star}:\ell_{2}\left(J\right)\rightarrow\mathcal{H}$ is surjective, and has the form $T_{\Phi}^{\star}\left(c_{j}\right)=\sum_{j\in J}c_{j}\varphi_{j},\,\left(c_{j}\right)\in\ell_{2}\left(J\right)$, we know that all points in codomain $\mathcal{H}$ are covered by $T_{\Phi}^{\star}$. It seems to me that the synthesis operator has kind of the form of a linear combination of the frame-vectors of $\Phi$.

My question

How can I complete the proof of the surjectivity of the synthesis operator, and how can I properly interpert that to show that the closed linear span of $\Phi$ coincide with $\mathcal{H}$.

Thank you, Franck!

share|cite|improve this question
up vote 1 down vote accepted

If a closed subspace $V\subset \mathcal H$ does not coincide with $\mathcal H$, then there exists a nonzero vector $x$ that is orthogonal to $V$. What happens with the lower frame bound when you plug in such $x$?

To show that $\|\varphi_j\|$ are uniformly bounded, plug $x=\varphi_j$ into the upper frame bound.

share|cite|improve this answer
Thanks for the hint! $x\in\mathcal{H}$ such that $x\in\left(span\left\{ \varphi_{j}:j\in J\right\} \right)^{\perp} \Rightarrow\forall j\in J:\,\left\langle x,\varphi_{j}\right\rangle =0$. Thus, the lower frame-bound must be zero. Hence, $\left\{ \varphi_{j}:j\in J\right\} $ cannot be a frame. Here is the contradiction. For the uniformely boundedness: I plug $x=\varphi_j$ in, and I obtain $$\sum_{j\in J}\left|\left\langle \varphi_{n},\varphi_{j}\right\rangle \right|^{2}\leq B\left\Vert \varphi_{n}\right\Vert ^{2}.$$ I am stuck again. Do we know anything about $<\varphi_n,\varphi_j>$? – franckysnow Jul 2 '12 at 22:21
@FranckStudiesCommEng We know something when $n=j$. – user31373 Jul 2 '12 at 22:24
Since $<\cdot,\cdot>$ is an inner-product, it is positive-definite, hence $<\varphi_n,\varphi_n>\geq 0$. Besides that, I tried using Cauchy-Schwarz, but it did not help. What are you precisely thinking about? – franckysnow Jul 2 '12 at 23:00
All terms of the sum are nonnegative. Ignoring all of them except the one with $j=n$, we get $|\langle \varphi_n,\varphi_n\rangle|^2\le B\|\varphi_n\|^2$. Do you know the relation of the norm to the inner product of a vector with itself? – user31373 Jul 2 '12 at 23:21
$$\left\Vert \varphi_{n}\right\Vert ^{4}=\left|\left\langle \varphi_{n},\varphi_{n}\right\rangle \right|^{2}\leq\sum_{j\in J}\left|\left\langle \varphi_{n},\varphi_{j}\right\rangle \right|^{2}\leq B\left\Vert \varphi_{n}\right\Vert ^{2} $$ $$ \Rightarrow\left\Vert \varphi_{n}\right\Vert ^{2}\leq B. $$ Thank you again for your great help, I really appreciate that you do not give me directly the answer but that you lead me on the way to the answer! Best regards, Franck! – franckysnow Jul 2 '12 at 23:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.