Realizing finite subgroups of Out(π₁(M)) by isometries of a nonpositively curved metric: the Schoen–Yau problem

Whether the finite symmetries of a compact nonpositively curved manifold are exactly the finite subgroups of the outer automorphism group of its fundamental group is the higher-dimensional Nielsen realization problem posed by Schoen and Yau in 1979: given a compact manifold $M$ that admits some Riemannian metric of nonpositive sectional curvature and a finite subgroup $G$ of $\operatorname{Out}(\pi_1(M))$, is there a nonpositively curved metric on $M$ whose isometry group contains a finite subgroup inducing $G$? It is affirmative for closed surfaces, for closed hyperbolic manifolds of dimension at least 3, for compact flat manifolds, and for compact locally symmetric spaces without one- or two-dimensional local factors; it is false in general from dimension 7 onward by Farrell–Jones' exotic smooth structures, and it fails for manifolds whose fundamental group has nontrivial center whenever the required group extension does not exist. The surviving open core is precise: in dimension 3, whether every finite group action by orientation-preserving diffeomorphisms on a closed nonpositively curved 3-manifold preserves some nonpositively curved metric (the case of free actions was settled by Zou in 2026, actions with fixed points remain open); and in dimensions 4, 5, and 6, the question for manifolds outside the settled geometric classes.

Background

A compact Riemannian manifold has nonpositive curvature (NPC) if every sectional curvature is at most $0$; by the Cartan–Hadamard theorem its universal cover is diffeomorphic to Euclidean space, so such a manifold is aspherical: its higher homotopy groups vanish and its homotopy type is recorded entirely by its fundamental group $\pi_1(M)$.

For an aspherical $M$, a self-diffeomorphism acts on $\pi_1(M)$ by an automorphism that is well defined only up to conjugation, that is, by an element of the outer automorphism group $\operatorname{Out}(\pi_1(M))=\operatorname{Aut}(\pi_1(M))/\operatorname{Inn}(\pi_1(M))$. Diffeomorphisms homotopic to the identity act trivially, so there is a natural homomorphism $\operatorname{Diff}(M)\to\operatorname{Out}(\pi_1(M))$, and for each Riemannian metric $g$ likewise $\operatorname{Isom}(M,g)\to\operatorname{Out}(\pi_1(M))$. The classical Nielsen realization problem asked this for closed surfaces $S$ of genus at least $2$, where $\operatorname{Out}(\pi_1(S))$ is the extended mapping class group: is every finite subgroup realized by isometries of some hyperbolic metric? Kerckhoff proved in 1983 that the answer is yes, which is exactly the present question in dimension $2$.

The 1979 source paper of Schoen and Yau studied compact group actions on manifolds that admit a map of nonzero degree (or a cohomologically nontrivial pairing) into a compact NPC manifold, using harmonic maps — critical points of the Dirichlet energy within a homotopy class, which into nonpositively curved targets enjoy strong uniqueness and rigidity properties. Their results bound compact transformation groups (for instance, the identity component of an effective compact group action is a torus of controlled dimension) and realize finite actions by isometries in two special cases: when the target is a compact locally symmetric space of nonpositive curvature — a quotient of a symmetric space by a lattice, such as a hyperbolic or complex-hyperbolic manifold, or a product of such — having no one- or two-dimensional geodesic local direct factor; and when the target is a compact flat manifold (Bieberbach manifold), that is, a compact quotient of $\mathbb{R}^n$ by a discrete cocompact group of Euclidean isometries. They posed, as a natural question for arbitrary compact NPC manifolds, the direct analog of Nielsen realization that is the subject of this question.

Three bodies of later work frame the current status.

Three-manifold structure. A closed orientable 3-manifold admitting an NPC metric is irreducible, and its JSJ decomposition — the canonical splitting along embedded incompressible tori (tori whose inclusion injects on the fundamental group) — has pieces that are either Seifert fibered (fibrated by circles over a 2-dimensional orbifold) or atoroidal (containing no essential embedded torus, that is, no incompressible torus beyond boundary-parallel ones), hence hyperbolic by 3-manifold geometrization as used in Dinkelbach–Leeb (2009). A graph manifold is one whose JSJ pieces are all Seifert fibered. Meeks–Scott (1986) and Dinkelbach–Leeb (2009) proved that a finite group acting on a closed 3-manifold modeled on one of the eight standard 3-dimensional model geometries is realized by isometries of a suitable geometric metric, and Leeb (1995) gave a criterion — restated by Zou (2026) in the language of the charges of framed Seifert fibered spaces developed by Buyalo–Svetlov (2005), numerical gluing invariants of the Seifert pieces and their boundary tori — for when flat metrics on the JSJ tori extend across the Seifert fibered pieces to a metric of nonpositive curvature on the whole manifold.

The algebraic obstruction. If a finite group $G$ acts on an aspherical $M$ realizing a given outer action $\psi:G\to\operatorname{Out}(\pi_1(M))$, then lifting the action to the universal cover produces a group extension $1\to\pi_1(M)\to\Pi\to G\to 1$ inducing $\psi$. Conversely, when the center $Z(\pi_1(M))$ is trivial such an extension always exists and is unique, but when the center is nontrivial (Seifert fibered spaces, manifolds with a circle factor) the extension need not exist: the obstruction is a class in the cohomology group $H^3(G;Z(\pi_1(M)))$. Zimmermann (1985) characterized which finite groups are realizable by homeomorphisms of Seifert fiber spaces and flat manifolds in arbitrary dimensions.

High-dimensional smoothing theory. A homotopy $n$-sphere is a smooth manifold homotopy equivalent to $S^n$; these form a finite abelian group $\Theta_n$ under connected sum, which is trivial in dimensions $5$ and $6$ and nontrivial from dimension $7$ on. For a closed manifold $M$, the connected sum $M\#\Sigma$ with a homotopy sphere $\Sigma$ is homeomorphic to $M$ but generally not diffeomorphic — an exotic smooth structure — with the same fundamental group. Farrell and Jones used this to produce, in dimensions at least $7$, smooth manifolds homeomorphic to hyperbolic manifolds on which specific outer automorphisms of $\pi_1$ are not induced by any diffeomorphism; no metric, nonpositively curved or otherwise, can then realize those automorphisms isometrically.

Problem Statement

The Schoen–Yau question (1979, quoted as Problem 1.1 in Zou 2026): let $M$ be a compact manifold that admits a Riemannian metric of nonpositive sectional curvature, and let $G$ be a finite subgroup of $\operatorname{Out}(\pi_1(M))$. Does there exist a nonpositively curved metric $g$ on $M$ such that $G$ is realized by a finite subgroup of $\operatorname{Isom}(M,g)$, that is, such that some homomorphism of $G$ into $\operatorname{Isom}(M,g)$ composes with the natural map $\operatorname{Isom}(M,g)\to\operatorname{Out}(\pi_1(M))$ to the given inclusion $G\hookrightarrow\operatorname{Out}(\pi_1(M))$?

The later literature partially answers this — affirmatively for closed surfaces, for closed hyperbolic manifolds of dimension at least 3, for compact flat manifolds, and for compact locally symmetric spaces without one- or two-dimensional local direct factors; negatively in dimensions at least 7 and whenever the group-extension obstruction in $H^3(G;Z(\pi_1(M)))$ is nonzero — so the open target is the derived, still-open core of the same question:

(a) Dimension 3 (Zou's Conjecture 1.5). Let $M$ be a closed orientable 3-manifold admitting an NPC metric, and let $G$ be a finite group acting on $M$ by orientation-preserving diffeomorphisms. Must $M$ admit a $G$-invariant NPC metric? For $M$ that is Haken (a compact irreducible 3-manifold containing a properly embedded two-sided incompressible surface, that is, a surface whose inclusion injects on the fundamental group) and not Seifert fibered, Zimmermann's 1982 lifting theorem motivates this problem from the original Schoen–Yau question: finite subgroups of $\operatorname{Out}(\pi_1(M))$ lift to finite groups of diffeomorphisms, so realizing the lifted action by an invariant NPC metric realizes the given outer action. For free actions the answer is affirmative in full generality (Zou 2026, Corollary 1.4; the Seifert fibered, atoroidal, and mixed-JSJ cases with at least one atoroidal piece were previously known under the freeness assumption), so the remaining open case is actions with fixed points, including graph manifolds, where the quotient $M/G$ is an orbifold rather than a manifold. As stated, the conjecture covers orientation-preserving actions; the orientation-reversing case is the directly adjacent variant.

(b) Dimensions 4, 5, and 6. The Schoen–Yau question for compact NPC manifolds outside the settled classes — in particular those that are neither hyperbolic, nor flat, nor locally symmetric without low-dimensional local factors. Note that locally symmetric spaces with one- or two-dimensional factors, such as products of hyperbolic surfaces or of a hyperbolic manifold with a circle, lie outside the settled hypotheses, and no known source addresses them.

Together (a) and (b) constitute the still-open core of the Schoen–Yau question, and each is an independent research target.

References

1. Richard Schoen and Shing Tung Yau, Compact group actions and the topology of manifolds with non-positive curvature, Topology 18 (1979), no. 4, pp. 361-380. DOI: 10.1016/0040-9383(79)90026-0, https://doi.org/10.1016/0040-9383(79)90026-0
2. Zhengyu Zou, Isometric free finite group actions on non-positively curved 3-manifolds, preprint v2, July 2026. arXiv:2606.22088, https://arxiv.org/abs/2606.22088
3. Steven P. Kerckhoff, The Nielsen Realization Problem, Annals of Mathematics, Second Series 117 (1983), no. 2, pp. 235-265. DOI: 10.2307/2007076, https://doi.org/10.2307/2007076
4. Bruno Zimmermann, Das Nielsensche Realisierungsproblem für hinreichend große 3-Mannigfaltigkeiten, Mathematische Zeitschrift 180 (1982), no. 2, pp. 349-359. DOI: 10.1007/BF01214175, https://doi.org/10.1007/BF01214175
5. Wolfgang Heil and Jeffrey L. Tollefson, Deforming homotopy involutions of 3-manifolds to involutions, Topology 17 (1978), no. 4, pp. 349-365. DOI: 10.1016/0040-9383(78)90003-4, https://doi.org/10.1016/0040-9383(78)90003-4
6. Bruno Zimmermann, Nielsensche Realisierungssätze für Seifertfaserungen, Manuscripta Mathematica 51 (1985), no. 1-3, pp. 225-242. DOI: 10.1007/BF01168354, https://doi.org/10.1007/BF01168354
7. F. T. Farrell and L. E. Jones, Smooth Non-Representability of Out π1 M, Bulletin of the London Mathematical Society 22 (1990), no. 5, pp. 485-488. DOI: 10.1112/blms/22.5.485, https://doi.org/10.1112/blms/22.5.485
8. F. T. Farrell and L. E. Jones, Negatively curved manifolds with exotic smooth structures, Journal of the American Mathematical Society 2 (1989), no. 4, pp. 899-908. DOI: 10.1090/s0894-0347-1989-1002632-2, https://doi.org/10.1090/s0894-0347-1989-1002632-2
9. F. T. Farrell and P. Ontaneda, Piecewise Linear Non-Representability of Out π1 M, Bulletin of the London Mathematical Society 34 (2002), no. 5, pp. 582-586. DOI: 10.1112/S0024609302001224, https://doi.org/10.1112/S0024609302001224
10. Mauricio Bustamante and Bena Tshishiku, Symmetries of exotic aspherical space forms, Mathematische Zeitschrift 309 (2025), no. 3, article 47. DOI: 10.1007/s00209-024-03673-8, https://doi.org/10.1007/s00209-024-03673-8 (preprint: arXiv:2109.09196, https://arxiv.org/abs/2109.09196)
11. Mauricio Bustamante and Bena Tshishiku, Symmetries of exotic negatively curved manifolds, Journal of Differential Geometry 120 (2022), no. 2, pp. 231-250. DOI: 10.4310/jdg/1645207478, https://doi.org/10.4310/jdg/1645207478 (preprint: arXiv:1901.00815, https://arxiv.org/abs/1901.00815)
12. Jonathan Block and Shmuel Weinberger, On the generalized Nielsen realization problem, Commentarii Mathematici Helvetici 83 (2008), no. 1, pp. 21-33. DOI: 10.4171/cmh/116, https://doi.org/10.4171/cmh/116 (preprint: arXiv:math/0508135, https://arxiv.org/abs/math/0508135)
13. William H. Meeks and Peter Scott, Finite group actions on 3-manifolds, Inventiones Mathematicae 86 (1986), no. 2, pp. 287-346. DOI: 10.1007/BF01389073, https://doi.org/10.1007/BF01389073
14. Jonathan Dinkelbach and Bernhard Leeb, Equivariant Ricci flow with surgery and applications to finite group actions on geometric 3-manifolds, Geometry & Topology 13 (2009), no. 2, pp. 1129-1173. DOI: 10.2140/gt.2009.13.1129, https://doi.org/10.2140/gt.2009.13.1129
15. Bernhard Leeb, 3-Manifolds with(out) Metrics of Nonpositive Curvature, Inventiones Mathematicae 122 (1995), no. 1, pp. 277-289. DOI: 10.1007/BF01231445, https://doi.org/10.1007/BF01231445
16. S. V. Buyalo and P. V. Svetlov, Topological and geometric properties of graph-manifolds, St. Petersburg Mathematical Journal 16 (2005), no. 2, pp. 297-340. DOI: 10.1090/S1061-0022-05-00852-6, https://doi.org/10.1090/S1061-0022-05-00852-6
17. Michel Boileau, Sylvain Maillot, and Joan Porti, Three-Dimensional Orbifolds and their Geometric Structures, Panoramas et Synthèses 15, Société Mathématique de France, Paris, 2003. Series volume: Panoramas et Synthèses 15, https://projecteuclid.org/ebooks/Panoramas-et-Synth%C3%A8ses/Panoramas-et-Synth%C3%A8ses-15
18. James F. Davis and Wolfgang Lueck, On Nielsen realization and manifold models for classifying spaces, Transactions of the American Mathematical Society 377 (2024), no. 11, pp. 7557-7600. DOI: 10.1090/tran/9155, https://doi.org/10.1090/tran/9155 (preprint: arXiv:2303.15765, https://arxiv.org/abs/2303.15765)
19. Jordi Daura Serrano, Large and iterated finite group actions on manifolds admitting non-zero degree maps to nilmanifolds, preprint (2025). arXiv:2506.11174, https://arxiv.org/abs/2506.11174