Connectedness of fibers beyond semitoric systems II: ephemeral singular points

Daniele Sepe, Susan Tolman · 2025

Read Connectedness of fibers beyond semitoric systems II: ephemeral singular points by Daniele Sepe, Susan Tolman (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-2510-16976.txt.

Introduction

Daniele Sepe and Susan Tolman widen a connectedness result for a class of systems in a 2025 paper. These systems, called integrable, keep many quantities fixed as they evolve. Their fibers are the sets of states that share the same fixed values. Whether each fiber forms one connected piece matters for sorting such systems.

An earlier paper by the same authors proved connectedness under strict rules. Those rules barred the rougher singular points of a system. The new paper relaxes that demand in a real way. It allows a new class of brief, fading singular points. Examples come from resonance cases and special geometric settings. A new family of examples shows the wider reach is genuine. More real systems now fall under the connected-fibers result.

Catalog record

CREATORDaniele Sepe, Susan Tolman
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE86 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-16976

Browse the full library · About the Tower of Babel

Full text

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

Connectedness of fibers beyond semitoric systems II: ephemeral singular points Thanks: 2020 Mathematics Subject Classification . Primary 37J35, 53D20. Secondary 37J39, 53D35. Thanks: Keywords: Completely integrable Hamiltonian systems, complexity one actions, connected fibers. Daniele Sepe, Susan Tolman Abstract . In an earlier paper, we proved the connectedness of the fibers of every 2 ​ n 2n -dimensional integrable system satisfying both: (A) the action extends the action of an ( n − 1 ) (n-1) -dimensional torus which has a proper moment map, and (B) every tall singular point is non-degenerate and no such point has a hyperbolic block and connected T T -stabilizer. Unfortunately, these criteria are fairly restrictive. Our main goal in this paper is to find a larger class of integrable systems that has connected fibers by weakening the non-degeneracy assumption above. To achieve this, we introduce “ephemeral” degenerate singular points, examples of which have appeared in the literature in the context of p : − q p\!:\!-q resonances and special Lagrangian fibrations. Finally, we construct a family of examples that shows that our main theorem meaningfully extends previous results. 1. Introduction Let T be an (n-1) -dimensional torus with Lie algebra {\mathfrak{t}} . Building on [ At , GS2 , LMTW , VN , Wa ] , in [ ST1 , Theorem 1.6] we proved the connectedness of the fibers of every 2n -dimensional integrable system such that: (A) the action extends a complexity one T -action with proper moment map, and (B) every tall singular point is non-degenerate and no such point has a hyperbolic block and connected T -stabilizer. Unfortunately, these criteria are fairly restrictive; see Theorem 1.3 . Our main goal in this paper is to find a larger class of integrable systems that has connected fibers by weakening the non-degeneracy assumption above. To achieve this, we introduce “ephemeral” degenerate singular points (see Definition 1.8 below), examples of which have appeared in the literature in the context of p\!:\!-q resonances (see [ SD ] and references therein), and special Lagrangian fibrations (see [ HL , Chapter III.3.A] and [ J , CB ] ). To state our results, we need some definitions from integrable systems, following [ ST1 , Section 1] . Definition 1.1 . An integrable system is a quadruple \left(M,\omega,V,f\right) , where \left(M,\omega\right) is a connected 2n -dimensional symplectic manifold, V is an n -dimensional real vector space, and f\colon M\to V^{*} is a smooth map such that: (1) the Poisson bracket \big\{\langle f,v_{1}\rangle,\langle f,v_{2}\rangle\big\}=0 for all v_{1},v_{2}\in V , and (2) the function f is regular on a dense subset of M . Here \langle\cdot,\cdot\rangle\colon V^{*}\times V\to{\mathbb{R}} is the natural pairing. Throughout this paper, we assume that every integrable system is complete , i.e., it is a Hamiltonian V -space. Fix an integrable system \left(M,\omega,V,f\right) . A point p\in M is regular if D_{p}f has full rank and singular otherwise. Since the V -orbit \mathcal{O}:=V\cdot p through any point p\in M is isotropic, the vector space \left(T_{p}\mathcal{O}\right)^{\omega}/T_{p}\mathcal{O} inherits a natural symplectic structure. Since the stabilizer W\subseteq V of p acts by linear symplectomorphisms on T_{p}M and fixes T_{p}\mathcal{O}, it acts on (T_{p}\mathcal{O})^{\omega}/T_{p}\mathcal{O} by linear symplectomorphisms; we call this action the linearized action of W at p . Definition 1.2 . Let p be a point in an integrable system \left(M,\omega,V,f\right) . Let W_{0}\subseteq V be the Lie algebra of the stabilizer W of p . The point p is non-degenerate 1 1 1 This is consistent with the standard definition (see [ ST1 , Remark 1.3] ). if there exist identifications W_{0}\simeq{\mathbb{R}}^{\dim W} and (T_{p}\mathcal{O})^{\omega}/T_{p}\mathcal{O}\simeq{\mathbb{C}}^{\dim W} such that the homogeneous moment map for the linearized action of W at p is a product of the following “blocks”: • ( elliptic ) z\in{\mathbb{C}}\mapsto\frac{1}{2}|z|^{2} ; • ( hyperbolic ) z\in{\mathbb{C}}\mapsto{\mathfrak{I}}(z^{2}) ; • ( focus-focus ) (z_{1},z_{2})\in{\mathbb{C}}^{2}\mapsto\left(\frac{1}{2}(|z_{1}|^{2}-|z_{2}|^{2}),{\mathfrak{I}}(z_{1}z_{2})\right) . Here {\mathfrak{I}}(w) denotes the imaginary part of w\in{\mathbb{C}} and {\mathbb{C}}^{\dim W} has the standard symplectic structure. If p is non-degenerate and all the blocks are elliptic, then p has purely elliptic type . Next we recall some fundamental facts about complexity one T -spaces. Fix a complexity one \bm{T} -space , that is, a triple (M,\omega,\Phi) , where (M,\omega) is a connected 2n -dimensional symplectic manifold and \Phi\colon M\to{\mathfrak{t}}^{*} is a moment map for an effective Hamiltonian T -action. Since the T -orbit \mathfrak{O}:=T\cdot p through any point p\in M is isotropic, the vector space (T_{p}\mathfrak{O})^{\omega}/T_{p}\mathfrak{O} inherits a natural symplectic structure. Moreover, the natural action of the stabilizer H\subseteq T of p on T_{p}M induces a symplectic representation of H on (T_{p}\mathfrak{O})^{\omega}/T_{p}\mathfrak{O} , called the symplectic slice representation at p . It is isomorphic to the representation of H on {\mathbb{C}}^{h+1} associated to some injective homomorphism \rho\colon H\hookrightarrow(S^{1})^{h+1} , where h=\dim H . Let {\mathfrak{h}} be the Lie algebra of H and let \eta_{i}\in{\mathfrak{h}}^{*} be the differential of the i ’th component of \rho . The (not necessarily distinct) vectors \eta_{0},\dots,\eta_{h} are called the isotropy weights at p ; the multiset of isotropy weights at p is uniquely determined. The reduced space at \beta\in{\mathfrak{t}}^{*} is the quotient \Phi^{-1}(\beta)/T ; we set {M/\!/T}:=\Phi^{-1}(0)/T . A point p\in M is short if [p] is an isolated point in \Phi^{-1}(\Phi(p))/T and is tall otherwise. As we mentioned in the beginning of this introduction, in [ ST1 ] we proved that every 2n -dimensional integrable system that satisfies both condition (A) and (B) above has connected fibers. Unfortunately, as the following result shows, these conditions are fairly restrictive. Theorem 1.3 . Let (M,\omega,\Phi) be a complexity one T -space with a proper moment map. Assume that some fiber of \Phi contains (at least) three distinct orbits \mathfrak{O}_{1},\mathfrak{O}_{2},\mathfrak{O}_{3} such that, for all p\in\mathfrak{O}_{i} : (1) If one of the isotropy weights at p is zero, then the stabilizer of p has more than two components; and (2) If there is a pair \eta,\eta^{\prime} of isotropy weights at p such that \eta+\eta^{\prime}=0 , then the stabilizer of p is not connected. Then there is no integrable system of the form \left(M,\omega,{\mathfrak{t}}\times{\mathbb{R}},f:=(\Phi,g)\right) such that every tall singular point is non-degenerate and each fiber of f is connected. Remark 1.4 . If the fiber of \Phi in Theorem 1.3 lies over the interior of \Phi(M) , then the local model for \mathfrak{O}_{i} , as defined below, has a surjective moment map for all i (see [ Sja , Theorem 6.5] ). Hence, conditions ( 1 ) and ( 2 ) above can be replaced with the following (see Remark 5.4 ): (1) ′ The stabilizer of p is neither trivial nor isomorphic to {\mathbb{Z}}_{2} ; and (2) ′ If the stabilizer of p is isomorphic to S^{1} , then the set of isotropy weights is not \{-1,+1\} . Many complexity one T -actions have a fiber containing three distinct orbits that satisfy ( 1 ) and ( 2 ) above. Theorem 1.3 immediately implies that they cannot be extended to integrable systems satisfying the assumptions (A) and (B) above – if they did, then each fiber would be connected (cf. [ HSS , HSSS ] ). To overcome this limitation, we introduce some degenerate singular points to allow more integrable systems that extend complexity one T -spaces and have connected fibers. From now

Read the complete text