Maximal Weinstein neighborhoods of symmetric R-spaces and their symplectic capacities

Johanna Bimmermann · 2025

Read Maximal Weinstein neighborhoods of symmetric R-spaces and their symplectic capacities by Johanna Bimmermann (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-02731.txt.

Introduction

Some curved spaces appear inside larger symplectic spaces as perfectly balanced Lagrangian surfaces, and the question is how fat a neighborhood they can claim before the geometry pinches. Johanna Bimmermann studies those maximal Weinstein neighborhoods in 2025 for the symmetric R-spaces, a family of real forms of Hermitian symmetric spaces.

She shows the maximal neighborhoods are dense in the ambient space, a roomy answer. That density then lets her compute two standard size measures, the Gromov width and the Hofer-Zehnder capacity, for the disc bundles of tangents and cotangents over these spaces. The results add exact numbers to a part of symplectic geometry where few families are fully priced.

Catalog record

CREATORJohanna Bimmermann
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-02731

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Maximal Weinstein neighborhoods of symmetric R-spaces and their symplectic capacities Johanna Bimmermann Abstract . Symmetric R R -spaces can be characterized as real forms of Hermitian symmetric spaces, and as such, they are all embedded as Lagrangian submanifolds. We show that their maximal Weinstein tubular neighborhoods are dense and use this property to compute both the Gromov width and the Hofer–Zehnder capacity of the corresponding disc (co)tangent bundles of the symmetric R R -spaces. 1. Introduction Symmetric R -spaces N are a special class of real flag manifolds that also possess the structure of compact-type symmetric spaces (N,g) . More precisely, fix a semisimple (non-compact) Lie group G and a parabolic subgroup P . Then, the (compact) coset space N=G/P is called a real flag manifold. If the action of the maximal compact subgroup K\subset G is transitive, we obtain a K -invariant metric on N . If the pair (K,H) , where H=K\cap P , forms a symmetric pair, then N is called a symmetric R -space. These spaces have been classified (see Appendix C ), and the list includes many notable examples: \mathrm{Gr}_{\mathbb{R}}(p,q),\mathrm{Gr}_{\mathbb{C}}(p,q),\mathrm{Gr}_{\mathbb{H}}(p,q),\mathrm{SO}(n),\mathrm{U}(n),\mathrm{Sp}(n),Q_{p,q}(\mathbb{R}),Q_{n}(\mathbb{C}),\mathbb{O}\mathrm{P}^{2},\ldots Complexification The dual description of N as a homogeneous space, N\cong K/H\cong G/P, gives rise to two natural complexifications: K^{\mathbb{C}}/H^{\mathbb{C}} and G^{\mathbb{C}}/P^{\mathbb{C}} . These two complexifications, however, are not the same. The space K^{\mathbb{C}}/H^{\mathbb{C}} is K -equivariantly biholomorphic to the tangent bundle TN , equipped with an adapted complex structure (see [ Tum23 , Thm. 2.1] 1 1 1 The identification in [ Tum23 , Thm. 2.1] is a diffeomorphism; we include a sketch of the proof in Appendix A explaining the biholomorphism part. ). In contrast, N_{\mathbb{C}}:=G^{\mathbb{C}}/P^{\mathbb{C}} is a Hermitian symmetric space of compact type (see [ Tak84 ] ). Holomorphic embedding The inclusion K^{\mathbb{C}}\subset G^{\mathbb{C}} induces a holomorphic embedding TN\hookrightarrow N_{\mathbb{C}} as an open dense K^{\mathbb{C}} -orbit. Both TN and N_{\mathbb{C}} admit invariant symplectic structures: \mathrm{d}\lambda (the pullback of the canonical symplectic form on T^{*}N via the metric g ) and \omega_{\mathrm{KKS}} (the Kirillov–Kostant–Souriau form, arising from the realization of N_{\mathbb{C}} as a coadjoint orbit). These symplectic forms, together with their respective complex structures, define Kähler structures. However, the embedding TN\hookrightarrow N_{\mathbb{C}} cannot be Kähler, as can already be seen from volume considerations: \mathrm{vol}(TN)=\infty>\mathrm{vol}(N_{\mathbb{C}}). This holomorphic embedding nevertheless shows that the tangent bundle of a symmetric R -space is uniruled; that is, there exists a (pseudo-)holomorphic curve through every point. Note that adapted complex structures are compatible with the symplectic form \mathrm{d}\lambda , but the holomorphic curves in TN obtained in this way will have infinite energy. In symplectic topology, (pseudo-)holomorphic curves play a prominent role, and finite energy curves are generally better behaved than infinite energy ones. This is one of the motivations for seeking an open-dense symplectic embedding of a fiberwise convex neighborhood of the zero section of (TN,\mathrm{d}\lambda) into (N_{\mathbb{C}},\omega_{\mathrm{KKS}}) . Stretching the neck along the boundary of this fiberwise convex neighborhood could then provide finite-energy foliations of the tangent bundle TN . In the cases N=\mathbb{S}^{n},\mathbb{RP}^{n},\mathbb{CP}^{n} , or for Hermitian symmetric spaces of compact type, such symplectic embeddings were constructed explicitly (case by case) in [ Ada25 , Bim24a , Bim24c ] , and used to compute symplectic capacities of disc tangent bundles (DTN,\mathrm{d}\lambda) . Symplectic embedding In this article, we provide a systematic construction of such symplectic embeddings for all symmetric R -spaces. To state our main result, we introduce some notation. Let \mathfrak{k}=\mathfrak{h}\oplus\mathfrak{l} be the Cartan decomposition associated with the symmetric pair (K,H) , and let \mathfrak{a}\subset\mathfrak{l} be a maximal abelian subalgebra. Note that \dim(\mathfrak{a})=:\mathrm{rk}(N) . These integrate into maximal flats, which can be thought of as immersed tori. Let \Sigma be the restricted root system of \mathfrak{k} with respect to \mathfrak{a}\subset\mathfrak{l} , and define \Box_{r}:=\big\{X\in\mathfrak{a}\mid|\alpha(X)|<r\text{ for all }\alpha\in\Sigma\big\}. We equivariantly associate a fiberwise convex neighborhood of the zero section by U_{r}N:=\big\{(k,X)\mid X\in\Box_{r}\big\}\subset TN\cong K\times_{H}\mathfrak{l}. The set U_{r}N\subset TN is open, as H acts transitively on maximal flats. In the non-compact setting, such neighborhoods of the zero section are often referred to as Grauert domains (see, for example, [ BHH03 , AG90 ] ), and for certain value of r , they yield the largest domain on which the adapted invariant complex structure is defined. Theorem 1 (Corollary 16 ) . For r\leq\frac{\mathrm{rk}(N_{\mathbb{C}})}{\mathrm{rk}(N)} , there exists a K -equivariant symplectic embedding (U_{r}N,\mathrm{d}\lambda)\hookrightarrow(N_{\mathbb{C}},\omega_{\mathrm{KKS}}), which is open-dense for r=\frac{\mathrm{rk}(N_{\mathbb{C}})}{\mathrm{rk}(N)} . A similar theorem was proved by Torres in [ MT23 , Thm. 1.1] in the broader setting of coadjoint orbits. However, in [ MT23 , Thm. 1.1] , the neighborhood of the zero section identified with the open-dense orbit is not explicit. The explicit characterization of the neighborhood U_{r}N is essential for computing capacities. Capacities We compute both the Gromov width and the Hofer–Zehnder capacity of the domains (U_{r}N,{{\mathrm{d}}}\lambda) . Notably, as explained in Section 3.1 , either \mathrm{rk}(N_{\mathbb{C}})=\mathrm{rk}(N) or \mathrm{rk}(N_{\mathbb{C}})=2\cdot\mathrm{rk}(N) and \mathrm{rk}(N_{\mathbb{C}})=2\cdot\mathrm{rk}(N) if and only if N is simply connected. Theorem 2 . Let \mathrm{sys} denote the length of the shortest closed geodesic of the symmetric R -space (N,g) . Then c_{G}(U_{1}N,\mathrm{d}\lambda)=c_{HZ}(U_{1}N,\mathrm{d}\lambda)=\begin{cases}\mathrm{sys},&\text{if }\mathrm{rk}(N_{\mathbb{C}})=2\cdot\mathrm{rk}(N),\\ 2\cdot\mathrm{sys},&\text{if }\mathrm{rk}(N_{\mathbb{C}})=\mathrm{rk}(N).\end{cases} Moreover, if \mathrm{rk}(N)=1 , that is, if N\in\{\mathbb{S}^{n},\mathbb{R}\mathrm{P}^{n},\mathbb{C}\mathrm{P}^{n},\mathbb{H}\mathrm{P}^{n},\mathbb{O}\mathrm{P}^{2}\} , then U_{r}N=D_{r}N . This theorem generalizes [ Bim24a , Thm. A] , where the Hofer–Zehnder capacity of (D_{1}N,\mathrm{d}\lambda) was computed when N\in\{\mathbb{RP}^{n},\mathbb{CP}^{n}\} . Note that both spaces \mathbb{RP}^{n} and \mathbb{CP}^{n} are symmetric R -spaces, and their complexifications are \mathbb{CP}^{n} and \mathbb{CP}^{n}\times\mathbb{CP}^{n} , respectively. Observe that \mathrm{rk}(\mathbb{CP}^{n})=1=\mathrm{rk}(\mathbb{RP}^{n})\quad\text{and}\quad\mathrm{rk}(\mathbb{CP}^{n}\times\mathbb{CP}^{n})=2=2\cdot\mathrm{rk}(\mathbb{CP}^{n}). Disc tangent bundles Disc bundles D_{r}N:=\big\{(x,v)\mid|v|_{x}<r\big\} are of particular interest, as they arise as sublevel sets of the kinetic Hamiltonian E(x,v)=\frac{1}{2}|v|_{x}^{2} , which generates the geodesic flow, revealing a deep connection to Riemannian geometry. For simply connected N , we obtain the following corollary: Corollary 3 . If N is simply connected, 2 2 2 N is simply connected if and only if \mathrm{rk}(N_{\mathbb{C}})=2\cdot\mathrm{rk}(N) ; see Appendix C . then c_{HZ}(D_{1}N,\mathrm{d}\lambda)=\mathrm{sys}. In the non-simply connected case, constructing a lower bound is more challenging and could potentially be achieved using billiards. However, the analogue of Corollary 3 does not hold in general. For example, if N=Q_{p,q}(\mathbb{R}):=\bigg\{[x]\in\mathbb{RP}^{p+q+1}\ \bigg|\ x_{1}^{2}+\cdots+x_{p+1}^{2}-x_{p+2}^{2}-\cdots-x_{p+q+2}^{2}=0\bigg\} is a real quadric, we find: Theorem 4 . If 1\leq p\leq q , then c_{HZ}(D_{1}Q_{p,q}(\mathbb{R}),\mathrm{d}\lambda)=\sqrt{2}\cdot\mathrm{sys}. On the other hand, for N=\mathbb{RP}^{n} , we have U_{r}N=D_{r}N , and therefore c_{HZ}(D_{1}\mathbb{RP}^{n},\mathrm{d}\lambda)=2\cdot\mathrm{sys}. In both cases the capacity is given by the length of the shortest closed contractible geodesic. It would be interesting to understand the Hofer–Zehnder capacity for the other non-simply connected symmetric R -spaces, but at present, we do not know how to approach this question systematically. Finsler Metrics and Symplectic Systolic Inequalities An interesting observation is that fiberwise convex subsets of the tangent bundle correspond to Finsler metrics. The Finsler metric associated with the neighborhood U_{r}N is a type of L^{\infty} -norm. More precisely, F^{p}_{x}\colon T_{x}N\xrightarrow{\ \ }\mathbb{R}_{\geq 0},\quad v=a^{\#}_{x}\mapsto\|\mathrm{ad}_{a}\|_{p}, defines a K -equivariant Finsler metric on N , where \|\cdot\|_{p} denotes the operator norm on \mathrm{End}(\mathfrak{k}) induced by the L^{p} -norm on \mathfrak{k} . A straightforward computation shows that U_{1}N is the unit disc tangent bundle with respect to F^{\infty} . When N is simply connected, modifying the function \mathrm{sys}\cdot F^{1} yields an admissible Hamiltonian on D_{1}^{F^{1}}N\subset D_{1}^{F^{\infty}}N=U_{1}N . Hence, the Hofer–Zehnder capacity of all F^{p} -unit disc bundles is the same. This implies that the invariant Riemannian metric g

Read the complete text