betaFeedback

WorksMathematics

Quaternionic Torus Actions on Manifolds

Abstract

We study smooth actions of quaternionic tori SU⁡(2)r\operatorname{SU}(2)^r on closed, connected smooth manifolds, establishing sharp bounds for rr in terms of the dimension of the manifold when the action is either effective, effective almost-free, or almost-effective. In particular, we prove that an almost-effective SU⁡(2)r\operatorname{SU}(2)^r-action on an nn-dimensional manifold satisfies r≤⌊2n/3⌋r\leq \lfloor2n/3\rfloor, and we classify up to diffeomorphism all simply connected manifolds attaining the maximal possible rank. This gives a quaternionic counterpart of the maximal symmetry rank conjecture without any curvature assumption. We also establish the bounds r≤n/3r\leq n/3 for effective and effective almost-free quaternionic torus actions, and we compute the exact value for spheres, quoric manifolds, or manifolds of the form (S3)a×(S1)b(S^3)^a\times (S^1)^b.

Introduction

The study of torus actions on manifolds is a classical subject in the theory of compact transformation groups, with applications across algebraic topology, differential geometry, and algebraic geometry. A common theme is determining how the topology and geometry of a manifold constrain the rank of tori that can act effectively on it. For example, a classical theorem states that if TrT^r acts effectively on a closed, connected, smooth nn-dimensional manifold, then r≤nr \le n and the bound is attained if and only if MM is diffeomorphic to TnT^n. In differential geometry, Grove’s symmetry program studies isometric group actions of large compact Lie groups on manifolds of non-negative sectional curvature. In particular, the maximal symmetry rank conjecture predicts that if TrT^r acts effectively and isometrically on a closed, simply connected, non-negatively curved nn-manifold, then r≤⌊2n/3⌋r \le\lfloor2n/3 \rfloor, with strong restrictions on the manifolds for which the equality is attained; see Conjecture 1 for a precise formulation and [19, 35] for surveys of the symmetry program. Toric topology provides another source of examples. Davis and Januszkiewicz [13] introduced quasitoric manifolds as topological analogues of smooth projective toric varieties. They are closed, smooth 2n2n-manifolds with an effective TnT^n-action that is locally the standard TnT^n action on Cn\mathbb{C}^n and such that the orbit space is a simple polytope. These and many other classes of manifolds connecting torus actions and combinatorics are presented in [8].

Actions of compact connected non-abelian Lie groups have also been extensively studied. For example, [10, 37, 38, 39 Xu07] obtained bounds for the dimension of semisimple compact Lie groups acting effectively on some classes of manifolds, while Lawson and Yau [LY74] proved that a closed, connected, smooth manifold admitting an effective action of such a group admits a metric of positive scalar curvature. In toric topology, Hopkinson introduced quoric manifolds as quaternionic analogues of quasitoric manifolds in [23]. They have dimension 4n4n and carry locally regular actions of SU(2)n\mathrm{SU}(2)^n whose orbit spaces are nn-dimensional simple polytopes. Related constructions connecting actions of SU(2)r\mathrm{SU}(2)^r with combinatorial objects can be found in [34, 17, 6, 5]. We also note that the orbit space of actions of SU(2)r\mathrm{SU}(2)^r on products of spheres is studied in [1].

These developments motivate our study of actions of SU(2)r\mathrm{SU}(2)^r on manifolds, with the goal of examining how they resemble their abelian counterparts and which features of semisimple group actions they retain. We write

Q:=Sp(1)≅SU(2).Q := \mathrm{Sp}(1) \cong\mathrm{SU}(2).

Since QQ is diffeomorphic to S3S^3, we call QrQ^r a quaternionic torus of rank rr.

Throughout the paper, all manifolds are closed, connected and smooth, and all actions are smooth. We denote a smooth action of a Lie group GG on a manifold MM, Φ:G⟶Diff⁡(M)\Phi:G\longrightarrow\operatorname{Diff}(M), by G↷MG\curvearrowright M. The kernel of the action is the subgroup Ker⁡(Φ)⊆G\operatorname{Ker}(\Phi)\subseteq G. Recall that an action is effective if its kernel is trivial, almost-effective if its kernel is finite, and almost-free if all isotropy subgroups are finite.

We introduce three invariants that determine the maximal rr such that QrQ^r acts on a manifold MM under different assumptions on the actions.

Definition 1.1. Let MM be a closed, connected smooth manifold. Then:

  1. The quaternionic degree of symmetry of MM is

q-sym⁡(M)=max⁡{r:Qr↷M effectively}.\operatorname{q\text{-}sym}(M)=\max\{r:Q^r\curvearrowright M\text{ effectively}\}.
  1. The quaternionic rank of MM is

q-rank⁡(M)=max⁡{r:Qr↷M effectively and almost-freely}.\operatorname{q\text{-}rank}(M)=\max\{r:Q^r\curvearrowright M\text{ effectively and almost-freely}\}.
  1. The almost-effective quaternionic degree of symmetry of MM is

q-sym⁡ae(M)=max⁡{r:Qr↷M almost-effectively}.\operatorname{q\text{-}sym}_{ae}(M)=\max\{r:Q^r\curvearrowright M\text{ almost-effectively}\}.

The first two invariants are quaternionic counterparts of the toral degree of symmetry tor-sym⁡(M)\operatorname{tor\text{-}sym}(M) and the toral rank rank⁡(M)\operatorname{rank}(M), defined using effective and almost-free actions of TrT^r, respectively (see [24, 26, 32] and [3, 16] for background on the toral degree of symmetry and the toral rank, respectively). If GG is a compact, connected Lie group acting almost-effectively on MM, then G/ΓG/\Gamma acts effectively on MM, where Γ\Gamma denotes the kernel of the GG-action on MM. Thus, if TrT^r acts almost-effectively on MM with kernel a finite group Γ⊆Tr\Gamma\subseteq T^r, then Tr/Γ≅TrT^r/\Gamma\cong T^r acts effectively on $M. Note that if QrQ^r acts almost-effectively on a manifold MM, then the restriction of the action to a maximal torus Tr≤QrT^r\leq Q^r is also almost-effective, and therefore q-sym⁡ae(M)≤tor-sym⁡(M)\operatorname{q\text{-}sym}_{ae}(M)\leq\operatorname{tor\text{-}sym}(M). Hence, for a closed, connected smooth manifold MM, we have the string of inequalities

q-rank⁡(M)≤q-sym⁡(M)≤q-sym⁡ae(M)≤tor-sym⁡(M).\operatorname{q\text{-}rank}(M)\leq\operatorname{q\text{-}sym}(M)\leq\operatorname{q\text{-}sym}_{ae}(M)\leq\operatorname{tor\text{-}sym}(M).

Our main result gives a sharp upper bound for the almost-effective quaternionic degree of symmetry and classifies, up to diffeomorphism, the simply connected manifolds for which the bound is attained.

Theorem 1.2. Let MM be a closed, connected smooth nn-dimensional manifold. Then q-sym⁡ae(M)≤2n3\operatorname{q\text{-}sym}_{ae}(M)\leq\frac{2n}{3}. Moreover, if MM is simply connected and q-sym⁡ae(M)=⌊2n3⌋\operatorname{q\text{-}sym}_{ae}(M)=\left\lfloor\frac{2n}{3}\right\rfloor, then:

  1. If n=3kn=3k, then MM is diffeomorphic to (S3)k(S^3)^k.

  1. If n=3k+1n=3k+1, then MM is diffeomorphic to either (a) (S3)k−1×S2×S2(S^3)^{k-1}\times S^2\times S^2, (b) (S3)k−1×S4(S^3)^{k-1}\times S^4, or (c) (S3)k−2×S7(S^3)^{k-2}\times S^7.

  1. If n=3k+2n=3k+2, then MM is diffeomorphic to (S3)k×S2(S^3)^k\times S^2.

Every manifold in this classification admits an almost-effective quaternionic torus action that attains the bound and preserves the standard product metric, which has non-negative sectional curvature. This relates our theorem to the maximal symmetry rank conjecture, a central problem in the study of isometric torus actions on non-negatively curved manifolds.

Conjecture 1. (Maximal symmetry rank conjecture, [20, 15, 12]) Let TrT^r act isometrically and effectively on a closed, simply connected, non-negatively curved nn-dimensional Riemannian manifold MM. Then r≤2n3r \le\frac{2n}{3}. Moreover, if r=⌊2n3⌋r = \lfloor\frac{2n}{3} \rfloor, then:

  1. If n=3kn = 3k, then MM is diffeomorphic to (S3)k(S^3)^k.

  2. If n=3k+1n = 3k + 1, then MM is diffeomorphic to either (a) (S3)k−1×S4(S^3)^{k-1} \times S^4, (b) (S3)k−2×S7(S^3)^{k-2} \times S^7, (c) (S3)k−3×S5×S5(S^3)^{k-3} \times S^5 \times S^5, (d) (S3)k+1/T2(S^3)^{k+1}/T^2, or (e) N/T1N/T^1, where NN is a manifold belonging to item 3.

  3. If n=3k+2n = 3k + 2, then MM is diffeomorphic to either (a) (S3)k−1×S5(S^3)^{k-1} \times S^5, or (b) (S3)k+1/T1(S^3)^{k+1}/T^1.

The torus actions defining the quotients are free and linear.

Note that since the quotient of a free linear action of T1T^1 on S3S^3 is S3/T1≅S2S^3/T^1 \cong S^2, the cases (S3)k−1×S2×S2(S^3)^{k-1} \times S^2 \times S^2 and (S3)k×S2(S^3)^k \times S^2 correspond to the cases 2.(d) and 3.(b) respectively. Theorem 1.2 proves a quaternionic version of the maximal symmetry rank conjecture without any curvature assumption. Note that if QrQ^r acts almost-effectively and isometrically on MM, then we can construct an effective isometric action of TrT^r on MM. Thus, Theorem 1.2 provides further evidence of the maximal symmetry rank conjecture.

We also point out that Galaz-García, Kerin and Radeschi proved that tor-sym⁡(M)≤2n3\operatorname{tor\text{-}sym}(M) \le\frac{2n}{3} if MM is a closed, simply connected rationally elliptic manifold, and they classify the manifolds for which the bound is attained up to equivariant rational homotopy equivalence (see [GKR21, Theorem A, Theorem B]). They further conjectured that these manifolds admit an equivariant diffeomorphism classification (see [GKR21, Rigidity conjecture]). Theorem 1.2 assumes the stronger symmetry of an almost-effective quaternionic torus action, but requires no rational ellipticity hypothesis and yields a classification up to diffeomorphism. Thus, Theorem 1.2 also provides a quaternionic counterpart to the diffeomorphism conclusion of their rigidity conjecture.

We also prove bounds and partial classification results for q-sym⁡(M)\operatorname{q\text{-}sym}(M) and q-rank⁡(M)\operatorname{q\text{-}rank}(M), which are analogous to the results for tor-sym⁡(M)\operatorname{tor\text{-}sym}(M) and rank⁡(M)\operatorname{rank}(M).

Theorem 1.3. Let MM be a closed, connected smooth nn-dimensional manifold. Then q-sym⁡(M)≤n3\operatorname{q\text{-}sym}(M) \le\frac{n}{3}. Moreover, if MM is simply connected, n=3kn = 3k and q-sym⁡(M)=n3\operatorname{q\text{-}sym}(M) = \frac{n}{3}, then MM is diffeomorphic to QkQ^k.

Theorem 1.4. Let MM be a closed, connected smooth nn-dimensional manifold. Then q-rank⁡(M)≤n3\operatorname{q\text{-}rank}(M) \le\frac{n}{3}. Moreover:

  1. If n=3kn = 3k and q-rank⁡(M)=n3\operatorname{q\text{-}rank}(M) = \frac{n}{3}, then MM is finitely covered by QkQ^k.

  2. If n=3k+1n = 3k + 1 and q-rank⁡(M)=n−13\operatorname{q\text{-}rank}(M) = \frac{n-1}{3}, then MM is finitely covered by Qk×S1Q^k \times S^1.

A key feature of the almost-effective bound is that any action attaining it has cohomogeneity at most one. This permits the complete diffeomorphism classification in Theorem 1.2. By contrast, actions attaining the effective or almost-free dimension bounds can have cohomogeneity two, where classification is more difficult. Our classification results in these cases are therefore partial.

We also compute these invariants for some classes of manifolds.

Theorem 1.5. Let MM be a closed, connected smooth nn-dimensional manifold. Then:

  1. If H∗(M,Z/2)≅H∗(Sn,Z/2)H^*(M,\mathbb{Z}/2) \cong H^*(S^n,\mathbb{Z}/2), then q-sym⁡(M)≤n+14\operatorname{q\text{-}sym}(M) \le\frac{n+1}{4}.

  2. If χ(M)≠0\chi(M) \ne0, then q-sym⁡(M)≤n4\operatorname{q\text{-}sym}(M) \le\frac{n}{4}.

  3. If M≅Qa×TbM \cong Q^a \times T^b, then q-sym⁡(M)=a\operatorname{q\text{-}sym}(M) = a.

The first two bounds parallel the classical bounds tor-sym⁡(Sn)≤n+12\operatorname{tor\text{-}sym}(S^n) \le\frac{n+1}{2} and tor-sym⁡(M)≤n2\operatorname{tor\text{-}sym}(M) \le\frac{n}{2} when χ(M)≠0\chi(M) \ne0. The third complements the computation tor-sym⁡(Qa×Tb)=2a+b\operatorname{tor\text{-}sym}(Q^a \times T^b) = 2a+b in [37 Theorem C].

References

  1. [1]Anton A. Ayzenberg and Dmitry V. Gugnin. “On actions of tori and quaternionic tori on products of spheres”. English. In: Proc. Steklov Inst. Math. 326 (2024), pp. 1–10. ISSN: 0081-5438. DOI: 10.1134/S0081543824040011.DOI
  2. [2]Alejandro Adem, Johann Leida, and Yongbin Ruan. Orbifolds and stringy topology. Vol. 171. Cambridge University Press, 2007.DOI
  3. [3]Christopher Allday and Volker Puppe. Cohomological methods in transformation groups. Vol. 32. Cambridge University Press, 1993.DOI
  4. [4]Theodor Bröcker and Tammo tom Dieck. Representations of Compact Lie Groups. Vol. 98. Graduate Texts in Mathematics. New York: Springer-Verlag, 1985. DOI: 10.1007/978-3-662-12918-0.DOI
  5. [5]Panagiotis Batakidis and Ioannis Gkeneralis. “Tetraplectic structures compatible with local quaternionic toric actions”. In: Annals of Global Analysis and Geometry 69.1 (2026), p. 1.DOI
  6. [6]Panagiotis Batakidis and Ioannis Gkeneralis. The topology of local quaternionic toric actions. 2026. arXiv: 2512.07707 [math.GT]. URL: https://arxiv.org/abs/2512.07707.
  7. [7]Armand Borel. Seminar on transformation groups. 46. Princeton University Press, 1960.DOI
  8. [8]Victor M. Buchstaber and Taras E. Panov. Toric Topology. Vol. 204. Mathematical Surveys and Monographs. Providence, RI: American Mathematical Society, 2015.
  9. [9]Glen E Bredon. Introduction to compact transformation groups. Vol. 46. Academic press, 1972.
  10. [10]Dan Burghelea and Reinhard Schultz. “On the semisimple degree of symmetry”. In: Bulletin de la Société mathématique de France 103 (1975), pp. 433–440.
  11. [11]Graciela Chichilnisky. “Group actions on spin manifolds”. In: Transactions of the American Mathematical Society 172 (1972), pp. 307–315.DOI
  12. [12]Zheting Dong, Christine Escher, and Catherine Searle. “Almost isotropy-maximal manifolds of non-negative curvature”. In: Transactions of the American Mathematical Society 377.07 (2024), pp. 4621–4645.arxiv.org/abs/1811.01493
  13. [13]Michael W. Davis and Tadeusz Januszkiewicz. “Convex polytopes, Coxeter orbifolds and torus actions”. English. In: Duke Math. J. 62.2 (1991), pp. 417–451. ISSN: 0012-7094. DOI: 10.1215/S0012-7094-91-06217-4.DOI
  14. [14]Jordi Daura Serrano, Michael Kohn, and Marie-Amélie Lawn. “G-invariant spin structures on spheres”. In: Annals of Global Analysis and Geometry 62.2 (2022), pp. 437–455.
  15. [15]Christine Escher and Catherine Searle. “Torus actions, maximality, and non-negative curvature”. In: Journal für die reine und angewandte Mathematik (Crelles Journal) 2021.780 (2021), pp. 221–264.DOI
  16. [16]Yves Félix, John Oprea, and Daniel Tanré. Algebraic models in geometry. OUP Oxford, 2008.
  17. [17]Graziano Gentili, Anna Gori, and Giulia Sarfatti. “Quaternionic toric manifolds”. English. In: J. Symplectic Geom. 17.1 (2019), pp. 267–300. ISSN: 1527-5256. DOI: 10.4310/JSG.2019.v17.n1.a7. URL: hdl.handle.net/2434/611682.DOI
  18. [19]Karsten Grove. “Geometry of, and via, symmetries”. In: University Lecture Series-American Mathematical Society 27 (2002), pp. 31–51.DOI
  19. [20]Fernando Galaz-Garcia and Catherine Searle. “Low-dimensional manifolds with non-negative curvature and maximal symmetry rank”. In: Proceedings of the American Mathematical Society 139.7 (2011), pp. 2559–2564.
  20. [21]Corey A Hoelscher. “Classification of cohomogeneity one manifolds in low dimensions”. In: Pacific journal of mathematics 246.1 (2010), pp. 129–185.arxiv.org/abs/0712.1327
  21. [23]Jeremy Hopkinson. “Quoric manifolds”. PhD thesis. University of Manchester, School of Mathematics, 2008.
  22. [24]Wu Yi Hsiang. Cohomology theory of topological transformation groups. Vol. 85. Springer Science & Business Media, 2012.
  23. [25]H Blaine Lawson and Marie-Louise Michelsohn. Spin geometry. Vol. 38. Princeton university press, 2016.
  24. [26]Kyung Bai Lee and Frank Raymond. Seifert fiberings. 166. American Mathematical Soc., 2010.DOI
  25. [27]Kyung Bai Lee and Frank Raymond. “Manifolds on Which Only Tori Can Act”. In: Transactions of the American Mathematical Society 304.2 (1987), pp. 487–499. ISSN: 00029947. URL: http://www.jstor.org/stable/2000727 (visited on 03/21/2026).DOI
  26. [29]Philippe Mazaud. “Spin(4) actions on 8-dimensional manifolds. I”. English. In: Ill. J. Math. 44.1 (2000), pp. 183–211. ISSN: 0019-2082.DOI
  27. [30]Philippe Mazaud. “8-dimensional manifolds with S³ × S³ actions”. In: Topology and its Applications 115.1 (2001), pp. 63–95.DOI
  28. [31]Paul S Mostert. “On a compact Lie group acting on a manifold”. In: Annals of Mathematics 65.3 (1957), pp. 447–455.DOI
  29. [32]Ignasi Mundet i Riera. “Discrete degree of symmetry of manifolds”. In: Transformation Groups (2024), pp. 1–38.
  30. [33]Jeff Parker. “4-dimensional G-manifolds with 3-dimensional orbits”. In: Pacific Journal of Mathematics 125.1 (1986), pp. 187–204.
  31. [34]Richard Scott. “Quaternionic toric varieties”. In: Duke Math. J. 80.1 (1995), pp. 373–397.DOI
  32. [35]Catherine Searle. “Symmetries of spaces with lower curvature bounds”. In: Notices of the American Mathematical Society 70.4 (2023).arxiv.org/abs/2303.10479
  33. [36]Fabio Simas. “Nonnegatively curved five-manifolds with non-abelian symmetry”. English. In: Geom. Dedicata 181 (2016), pp. 61–82. ISSN: 0046-5755. DOI: 10.1007/s10711-015-0112-6.DOI
  34. [37]Kazuo Saito and Tsuyoshi Watabe. “Semisimple degree of symmetry of manifold with the homotopy type of product (S¹)ʳ × (S²)ˢ × (S³)ᵗ”. In: Osaka Journal of Mathematics 21.3 (1984), pp. 493–506.
  35. [38]Tsuyoshi Watabe. “Semisimple degree of symmetry and maps of non-zero degree into a product of 1-spheres and 2-spheres”. In: Tohoku Mathematical Journal, Second Series 37.1 (1985), pp. 33–42.DOI
  36. [39]Bin Xu. “The degree of symmetry of certain compact smooth manifolds”. In: Journal of the Mathematical Society of Japan 55.3 (2003), pp. 727–737.DOI

Paper details

vlnc:W_buxdq5k3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents