Symplectic Geometry and Frame Theory

Clayton Shonkwiler

Colorado State University

http://shonkwiler.org

10.02.18

/mfo18

This talk!

Take Home Message

Symplectic geometry is a powerful set of tools which is useful for the study of frames

Frame Theory Basics

A frame in \(\mathbb{C}^k\) is a collection \(\{f_1,\dots ,f_N\}\subset \mathbb{C}^k\) so that

a\|v\|^2 \leq \sum_{i=1}^N \left| \langle v, f_i \rangle \right|^2 \leq b \|v\|^2

for some \(b\geq a > 0\) and for all \(v\in\mathbb{C}^k\).

The frame is tight if \(a=b\); equivalently, if \(F\) is the matrix with columns \(f_i\), then the frame operator \(FF^*=\frac{1}{a}I_k\).

If \(\|f_i\|=1\) for  all \(i\), the frame is unit norm. We abbreviate (Finite) Unit Norm Tight Frame as FUNTF.

Examples.

\frac{1}{\sqrt{2}}\begin{bmatrix} \sqrt{2} & 0 & 1 & 1 & 1 \\ 0 & \sqrt{2} & 1 & e^{2\pi i/3} & e^{4\pi i/3} \end{bmatrix}

Notation

\(\mathcal{F}^{N,k}_S\) is the collection of length-\(N\) frames in \(\mathbb{C}^k\) with frame operator \(S\); i.e., \(k \times N\) matrices \(F\) with \(FF^*=S\).

Note that \(S\) is necessarily invertible and positive-definite.

\(\mathcal{F}^{N,k}_S(\vec{r})\) is the subset of \(\mathcal{F}^{N,k}_S\) of frames with \(\|f_i\| = r_i\).

In particular, the length-\(N\) FUNTFs in \(\mathbb{C}^k\) are precisely \(\mathcal{F}^{N,k}_{\frac{N}{k}I_k}(\vec{1})\) since

\operatorname{tr}(FF^*) = \operatorname{tr}(F^*F) = \operatorname{tr}\begin{bmatrix}1 & * & \cdots & * \\ * & 1 & \cdots & * \\ \vdots & & \ddots & \vdots \\ * & * & \cdots & 1 \end{bmatrix} = N

Symplecti-what?

Definition. A symplectic manifold is a smooth manifold \(M\) together with a closed, non-degenerate 2-form \({\omega \in \Omega^2(M)}\).

Example: \((S^2,d\theta\wedge dz)\)

Example. \((\mathbb{R}^2,dx \wedge dy) = (\mathbb{C},\frac{i}{2}dz \wedge d\bar{z})\)

Example. \((T^*\mathbb{R}^n, \sum_{i=1}^n dq_i \wedge dp_i)\)

Example. \((S^2,\omega)\), where \(\omega_p(u,v) = (u \times v) \cdot p\)

Example. \((\mathbb{R}^2,\omega)\) where \(\omega(u,v) = \langle i u, v \rangle \)

Example. \((\mathbb{C}^n, \frac{i}{2} \sum dz_k \wedge d\overline{z}_k)\)

Complex Geometry

If \((M,\omega)\) is symplectic and \(g\) is a Riemannian metric on \(M\), then \(g(\cdot, \cdot) = \omega(\cdot , J \cdot)\) defines a compatible almost complex structure \(J\) on \(M\).

If this compatible almost complex structure is integrable, then \((M,J,g,\omega)\) is Kähler.

Since the Fubini–Study form \(\omega_{\text{FS}}\) on \(\mathbb{CP}^n\) is Kähler, all smooth projective varieties are Kähler, and in particular symplectic.

The name ‘complex group’ formerly proposed by me..has become more and more embarrassing through collision with the word ‘complex’ in the connotation of complex number. I therefore propose to replace it by the corresponding Greek adjective ‘symplectic’.

— Hermann Weyl

Consequences

A symplectic manifold must be even-dimensional (over \(\mathbb{R}\)).

\(\omega^{\wedge n} = \omega \wedge \dots \wedge \omega\) is a volume form on \(M\), and induces a measure

m(U) := \int_U \omega^{\wedge n}

called Liouville measure on \(M\).

If \(H: M \to \mathbb{R}\) is smooth, then there exists a unique vector field \(X_H\) (the Hamiltonian vector field for \(H\)) so that \({dH = \iota_{X_H}\omega}\), i.e.,

dH(\cdot) = \omega(X_H, \cdot)

(\(X_H\) is sometimes called the symplectic gradient of \(H\))

Example

\(H: (S^2, d\theta\wedge dz) \to \mathbb{R}\) given by \(H(\theta,z) = z\).

\(dH = dz = \iota_{\frac{\partial}{\partial \theta}}\omega\), so \(X_H = \frac{\partial}{\partial \theta}\).

Integrating \(X_H\) produces the one-parameter family of diffeomorphisms \(\phi_t(\theta, z) = (\theta+t,z)\).

Lie Group Actions

Let \(G\) be a Lie group, and let \(\mathfrak{g}\) be its Lie algebra. If \(G\) acts on \((M,\omega)\), then each \(V \in \mathfrak{g}\) determines a vector field \(X_V\) on \(M\) by

X_V(p) = \left.\frac{d}{dt}\right|_{t=0}\exp(t V) \cdot p

\(S^1=U(1)\) acts on \((S^2,d\theta \wedge dz)\) by

e^{it} \cdot(\theta, z) = (\theta + t, z)

For \(r \in \mathbb{R} \simeq \mathfrak{u}(1)\), \(X_r = r \frac{\partial}{\partial \theta}\).

Lie Group Actions

Let \(G\) be a Lie group, and let \(\mathfrak{g}\) be its Lie algebra. If \(G\) acts on \((M,\omega)\), then each \(V \in \mathfrak{g}\) determines a vector field \(X_V\) on \(M\) by

X_V(p) = \left.\frac{d}{dt}\right|_{t=0}\exp(t V) \cdot p

\(SO(3)\) acts on \(S^2\) by rotations. 

\(X_{V_{(a,b,c)}}((x,y,z)) = (bz-cy)\frac{\partial}{\partial x} + (cx - az) \frac{\partial}{\partial y} + (ay - bx) \frac{\partial}{\partial z}\)

For \(V_{(a,b,c)} = \begin{bmatrix} 0 & -c & b \\ c & 0 & -a \\ -b & a & 0 \end{bmatrix} \in \mathfrak{so}(3)\),

\(= (a,b,c) \times (x,y,z)\)

Moment(um) Maps

Definition. An action of \(U(1)\) on \((M,\omega)\) is Hamiltonian if there exists a map

\mu: M \to \mathfrak{u}(1)^* \simeq \mathbb{R}

so that \(d\mu = \iota_{X}\omega\), where \(X\) is the vector field generated by the circle action.

\(X = \frac{\partial}{\partial \theta}\)

\(\iota_X\omega = \iota_{\frac{\partial}{\partial \theta}} d\theta \wedge dz = dz \)

\(\mu(\theta,z) = z\)

Moment(um) Maps

Definition. An action of \(G\) on \((M,\omega)\) is Hamiltonian if each one-parameter subgroup action is Hamiltonian. Equivalently, there exists a map

\mu: M \to \mathfrak{g}^*

so that \(\omega_p(X_V, X) = D_p \mu(X)(V)\) for each \(p \in M\), \(X \in T_pM\), and \(V \in \mathfrak{g}\).

\(X_{V_{(a,b,c)}}(x,y,z) = (a,b,c) \times (x,y,z)\)

\((\iota_{X_{V_{(a,b,c)}}}\omega)_{(x,y,z)} = a dx + b dy + c dz \)

\(\mu(x,y,z)(V_{(a,b,c)}) = (x,y,z)\cdot(a,b,c)\)

A Circle Action on \(\mathbb{C}^k\)

Let \(U(1)\) act on \((\mathbb{C}^k,\omega_{\text{std}})\) by scalar multiplication.

The vector field generated by the circle action is

X(\vec{v}) = \left. \frac{d}{dt}\right|_{t=0} e^{it}\vec{v} = i\vec{v},

so \((\iota_X\omega_{\text{std}})_{\vec{v}}(Y) = (\omega_{\text{std}})_{\vec{v}}(X,Y) = \langle iX,Y\rangle_{\vec{v}} = \langle -\vec{v},Y\rangle\).

The moment map is

\mu: \mathbb{C}^k \to \mathfrak{u}(1)^* \simeq \mathbb{R}
\vec{v} \mapsto -\frac{1}{2} \left| \vec{v}\right|^2

since \(d\mu_{\vec{v}}(Y) = \left.\frac{d}{dt}\right|_{t=0} \left(-\frac{1}{2} \langle \vec{v} + tY, \vec{v} + tY\rangle\right) = \langle -\vec{v},Y\rangle\).

The Big Guns

Theorem (Atiyah, Guillemin–Sternberg). 

Let \((M^{2n},\omega)\) be a compact connected symplectic manifold with a Hamiltonian \(d\)-torus action with momentum map \(\mu: M \to \mathbb{R}^d\). Then

  • the nonempty level sets of \(\mu\) are connected;
  • the image of \(\mu\) is convex (called the moment polytope);
  • the image of \(\mu\) is the convex hull of the images of fixed points of the action.

Theorem (Duistermaat–Heckman)

The pushforward of Liouville measure on \((M^{2n},\omega)\) to the moment polytope is absolutely continuous w.r.t. Lebesgue measure, with continuous, piecewise-polynomial density of degree \(\leq n-d\).

Example

Theorem (Archimedes)

Let \(\mu: S^2 \to \mathbb{R}\) be given by \(\mu(x,y,z) = z\). Pushing forward the uniform measure on \(S^2\) to the image \([-1,1]\) gives Lebesgue measure.

Example 2

Proposition. If we rotate both factors of \(S^2 \times S^2\) simultaneously around the \(z\)-axis, the moment map is \(((x_1,y_1,z_1),(x_2,y_2,z_2))\mapsto z_1 + z_2\), and the pushforward measure on \([-2,2]\) has density given by the triangle function.

Complex Example

The \(\mathbb{T}^2 = U(1) \times U(1)\) action on \(\mathbb{CP}^2\) given by

(e^{i\theta_1},e^{i\theta_2})\cdot [z_0:z_1:z_2] = \left[z_0:e^{i\theta_1}z_1:e^{i\theta_2}z_2\right]

has moment map

\mu([z_0:z_1:z_2]) = -\frac{1}{2} \left(\frac{|z_1|^2}{|z_0|^2+|z_1|^2+|z_2|^2},\frac{|z_2|^2}{|z_0|^2+|z_1|^2+|z_2|^2}\right).

Symplectic Quotients

Theorem (Mayer, Marsden–Weinstein)

Let \(\mu: M \to \mathfrak{g}^*\) be the moment map for a Hamiltonian action of \(G\) on \((M,\omega)\). If \(\xi \in \mathfrak{g}^*\) is a regular value of \(\mu\) and \(\mathcal{O}_\xi\) is its coadjoint orbit, then

M /\!/\!_\xi\, G := \mu^{-1}(\mathcal{O}_\xi)/G

is a symplectic manifold with symplectic form \(\omega_{\text{red}}\) such that

\pi^*\omega_{\text{red}} = \iota^* \omega

\(\mu^{-1}(\mathcal{O}_\xi)\)

\(M /\!/\!_\xi\, G = \mu^{-1}(\mathcal{O}_\xi)/G\)

\(M\)

\(\pi\)

\(\iota\)

A Frame-Adjacent Example

\(U(1)^N\) acts on \(\mathbb{C}^{k \times N}\) by right-multiplication by diagonal unitary matrices.

Columnwise, this is the \(U(1)\) action on \(\mathbb{C}^k\) with moment map \(\vec{v} \mapsto -\frac{1}{2}|\vec{v}|^2\).

The moment map for the big torus action is

\widetilde{\Phi}: [\vec{v}_1 | \vec{v}_2 | \dots | \vec{v}_N] \mapsto \left(-\frac{1}{2}|\vec{v}_1|^2, -\frac{1}{2}|\vec{v}_2|^2, \dots , -\frac{1}{2}|\vec{v}_N|^2\right)

If \(-\vec{\frac{1}{2}} = \left(-\frac{1}{2},\dots , -\frac{1}{2}\right)\), then \(\widetilde{\Phi}^{-1}\!\left(-\vec{\frac{1}{2}}\right)\simeq \left(S^{2k-1}\right)^N\) consists of all matrices with unit-norm columns and

\mathbb{C}^{k \times N}/\!/\!_{ -\vec{\frac{1}{2}}} U(1)^N= \widetilde{\Phi}^{-1}\!\left(-\vec{\frac{1}{2}}\right)/U(1)^N = \left(S^{2k-1}/U(1)\right)^N = \left(\mathbb{CP}^{k-1}\right)^N

If \(\vec{w} \in \operatorname{image}(\widetilde{\Phi})\), then \(\widetilde{\Phi}^{-1}\!\left(\vec{w}\right)\simeq \prod_{i=1}^NS^{2k-1}(\sqrt{-2w_i})\) consists of matrices with column norms given by the \(\sqrt{-2w_i}\) and

\mathbb{C}^{k \times N}/\!/\!_{ \vec{w}} U(1)^N= \widetilde{\Phi}^{-1}\!\left(\vec{w}\right)/U(1)^N = \left(S^{2k-1}/U(1)\right)^N = \left(\mathbb{CP}^{k-1}\right)^N

A Frame-Related Example

\(U(k)\) acts on \(\mathbb{C}^{k \times N}\) by left multiplication with moment map

\mu: \mathbb{C}^{k \times N} \to \mathfrak{u}(k)^* \simeq \mathcal{H}(k)

given by

\mu(V) = V V^*

Hermitian \(k \times k\) matrices

\mathcal{H}(k) \to \mathfrak{u}(k)^*
A \mapsto \left( B \mapsto \frac{i}{2} \operatorname{tr}(AB)\right)
D_V \mu(X) = XV^* + VX^*
\text{For } B \in \mathfrak{u}(k), X_B(V) = BV
(\iota_{X_B} \omega)_V(X) = \omega_V(BV,X) = \dots = \frac{i}{2}\operatorname{tr}(VX^*B+XV^*B)

For invertible, positive-definite \(S \in \mathcal{H}(k)\), \(\mu^{-1}(S) = \mathcal{F}^{N,k}_S\), the set of frames with frame operator \(S\).

E.g., \(\mathcal{F}^{N,k}_{I_k}= \mu^{-1}(I_k)\) is (some of) the tight frames.

E.g., \(\mathcal{F}^{N,k}_{\frac{N}{k}I_k} = \mu^{-1}\left(\frac{N}{k}I_k\right)\) is (some of) the tight frames.

Grassmannians Enter the Story

Recall \(\mathcal{F}^{N,k}_{\frac{N}{k}I_k} =  \left\{F : FF^* = \frac{N}{k}I_k\right\}\) is the space of tight frames which contains the FUNTFs.

U(k)\backslash\mathcal{F}^{N,k}_{\frac{N}{k}I_k} = U(k)\backslash\mu^{-1}(\frac{N}{k}I_k) = U(k)_{\frac{N}{k}I_k}\!\backslash\!\backslash\mathbb{C}^{k \times N} = \operatorname{Gr}_k(\mathbb{C}^N)

is symplectic.

Observation.

...and Flag Manifolds

If \(S \in \mathcal{H}(k) \simeq \mathfrak{u}(k)^*\) is invertible and positive-definite, with eigenvalues \(\lambda_1 > \lambda_2 > \dots > \lambda_\ell > 0\) of multiplicities \(k_1, \dots , k_\ell\), then

\mathcal{O}_S \simeq \operatorname{Fl}_k(d_1, \dots , d_\ell)

where \(d_i = \sum_{j=1}^i k_j\).

U(k)\backslash \mu^{-1}(\mathcal{O}_S) = U(k)_S\!\backslash\!\backslash \mathbb{C}^{k \times N} \simeq \operatorname{Fl}_N(d_1, \dots , d_\ell, N)

is symplectic. \(\mu^{-1}(\mathcal{O}_S)\) is the space of all frames whose frame operator is conjugate to \(S\); call it \(\widetilde{\mathcal{F}^{N,k}_S}\).

FUNTFs

Notice that FUNTF space \(\mathcal{F}_{\frac{N}{k}I_k}(\vec{1})\) is exactly

\mu^{-1}\left(\frac{N}{k}I_k\right) \cap \widetilde{\Phi}^{-1}\!\left(-\vec{\frac{1}{2}}\right) \subset \mathbb{C}^{k \times N}

The \(U(1)^N\) action on \(\mathbb{C}^{k \times N}\) induces an action on \(\operatorname{Gr}_k(\mathbb{C}^N)\); since the diagonal \(U(1)\) acts trivially, this is an effective \(U(1)^{N-1}\) action with moment map \(\Phi: \operatorname{Gr}_k(\mathbb{C}^N) \to \mathbb{R}^{N-1}\)

[\vec{v}_1 | \vec{v}_2 | \dots | \vec{v}_N] \mapsto \left(-\frac{1}{2}|\vec{v}_1|^2, -\frac{1}{2}|\vec{v}_2|^2, \dots , -\frac{1}{2}|\vec{v}_{N-1}|^2\right)
U(k)\backslash\mathcal{F}^{N,k}_{\frac{N}{k}I_k}(\vec{1}) = U(k)\backslash\! \left(\mu^{-1}\left(\frac{N}{k}I_k\right) \cap \widetilde{\Phi}^{-1}\!\left(-\vec{\frac{1}{2}}\right)\!\right) = \Phi^{-1}\!\left(-\vec{\frac{1}{2}}\right)

By Atiyah’s connectedness theorem, this is connected!

The Frame Homotopy Conjecture

Theorem (Cahill–Mixon–Strawn ’17)

The space \(\mathcal{F}^{N,k}_{\frac{N}{k}I_k}(\vec{1})\) of length-\(N\) FUNTFs in \(\mathbb{C}^k\) is path-connected for all \(N \geq k\geq 2\).

Proof. \(U(k) \backslash\mathcal{F}^{N,k}_{\frac{N}{k}I_k}(\vec{1}) = \Phi^{-1}\!\left(-\vec{\frac{1}{2}}\right)\) is connected.

Since \(U(k)\) is connected, the total space \(\mathcal{F}_{\frac{N}{k}I_k}(\vec{1})\) is connected.

Since this is a real algebraic set in \(\mathbb{C}^{k \times N} \simeq \mathbb{R}^{2kN}\), it is locally path-connected and therefore path-connected.

A Generalization

Theorem (with Needham ’18)

For any frame operator \(S\) and any admissible vector \(\vec{r}\) of frame norms, the space \(\mathcal{F}^{N,k}_S (\vec{r})\) is path-connected.

Proof idea.

U(k) \backslash \widetilde{\mathcal{F}^{N,k}_S}(\vec{r}) = \Phi^{-1}\left(-\frac{1}{2}r_1^2,\dots , -\frac{1}{2} r_{N-1}^2\right) \subset \operatorname{Fl}_N(d_1, \dots , d_\ell, N)

is connected by Atiyah’s connectedness theorem, so \(\widetilde{\mathcal{F}^{N,k}_S}(\vec{r})\) is connected and path-connected.

If \(V_t\) is a path in \(\widetilde{\mathcal{F}^{N,k}_S}(\vec{r})\) connecting two points \({V_0,V_1 \in \mathcal{F}^{N,k}_S(\vec{r})}\), then \(V_tV_t^* = U_t^*SU_t\) and \(U_tV_t\) also connects \(V_0\) to \(V_1\) and stays in \(\mathcal{F}^{N,k}_S(\vec{r})\).

Questions

\operatorname{Gr}_k(\mathbb{C}^N)/\!/\!_{-\vec{\frac{1}{2}}}U(1)^{N-1} = U(k)\backslash \mathcal{F}^{N,k}_{\frac{N}{k}I_k}(\vec{1})/U(1)^{N-1}

is (almost) toric, and in particular is measure-theoretically isomorphic to the product of the moment polytope (eigenstep polytope) and a torus. This should lead to a sampling algorithm for FUNTFs...

1.

2.

3.

Is there a corresponding story for frames on an infinite-dimensional Hilbert space?

What’s the version of the generalized frame homotopy theorem that applies to real frames? For example, \(\mathcal{F}^{6,\mathbb{R}^2}_{\frac{15}{2}I_2}(2,2,2,1,1,1)\) is not connected.

Thank you!

References

Funding: Simons Foundation

The geometry of constrained random walks and an application to frame theory

Clayton Shonkwiler

2018 IEEE Statistical Signal Processing Workshop (SSP), 343–347