---
title: Franks' Theorem in Surface and Symplectic Dynamics
url: https://www.emergentmind.com/topics/franks-theorem
type: topic
---

# Franks' Theorem in Surface and Symplectic Dynamics

to=arxiv_search.search ,一本道json
{"query":"all:Franks theorem annulus periodic points sphere geodesic flows", "max_results": 10, "sort_by":"submittedDate", "sort_order":"descending"}】【：】【“】【assistant to=arxiv_search.search in commentary  天天送json
{"query":"all:Franks theorem annulus periodic points sphere geodesic flows", "max_results": 10, "sort_by":"submittedDate", "sort_order":"descending"}planning to use search tool per instruction.
to=arxiv_search.search  天天中彩票如何json
{"query":"ti:\"Franks' dichotomy\" OR ti:\"Franks' lemma\" OR ti:\"Franks-Misiurewicz\" OR ti:\"A refinement of a theorem by Franks\"", "max_results": 20, "sort_by":"submittedDate", "sort_order":"descending"}
The contemporary literature suggests that the expression **Franks' theorem** is used for several closely related results associated with John Franks rather than for a single statement. Its classical core is a family of periodic-point dichotomies for area-preserving surface maps, especially on the annulus and the two-sphere. Around that core lies a broader Franks program: rational rotation data forcing periodic orbits, perturbative realizability results now called **Franks' lemma**, and higher-dimensional or symplectic descendants in Hamiltonian dynamics and geodesic flows [1410.3997][1107.1282][1307.6573].

## 1. Classical periodic-point theorems on the annulus and the sphere

One standard formulation concerns the annulus
\[
A:=\{z\in \mathbb{R}^2 \mid 1\le |z| \le 2\},\qquad \mathring A:=\{z\in \mathbb{R}^2 \mid 1<|z|<2\}.
\]
In this form, Franks’ theorem states that every area-preserving homeomorphism on \(A\) or \(\mathring A\) with a periodic point has infinitely many interior periodic points [1410.3997]. A closely related formulation, used in work on area-preserving annulus homeomorphisms isotopic to the identity, says that if such a map has at least one fixed or periodic point, then it must have infinitely many interior periodic points [1511.06803].

A second canonical form is the sphere dichotomy:
\[
\text{every area-preserving homeomorphism of }S^2\text{ has either }2\text{ or infinitely many periodic points.}
\]
This statement is the version emphasized in symplectic reinterpretations of Franks’ theorem and is described there as Franks’ celebrated “two-or-infinitely-many” theorem [1107.1282][2309.07991]. In the smooth category, a Hamiltonian reformulation on \((S^2,\omega)\) yields the same dichotomy for Hamiltonian diffeomorphisms [1107.1282].

These statements already display the characteristic Franks phenomenon: minimal periodic behavior is rigid, and any deviation from the minimal regime forces infinite periodic proliferation. Much of the later literature treats this phenomenon as the model case for stronger forcing results, symplectic generalizations, and arithmetic refinements.

## 2. Rotation forcing and arithmetic refinements on the annulus

A deeper annulus statement, used repeatedly as a forcing mechanism, is the rotation-interval theorem. For a homeomorphism \(f\) of the open or closed annulus isotopic to the identity, with lift \(\widetilde f\), if there exist two recurrent points \(z_1,z_2\) such that
\[
-\infty \le \rho(\widetilde f,z_1)<\rho(\widetilde f,z_2)\le +\infty,
\]
then for any rational number
\[
\frac pq\in \bigl(\rho(\widetilde f,z_1),\rho(\widetilde f,z_2)\bigr)
\]
written in irreducible form, there exists a \(q\)-prime-periodic point with rotation number \(p/q\) [2202.11517]. In the closed-annulus case this goes back to Franks, and for the open annulus it is extended in work of Le Calvez and Wang as cited there [2202.11517].

This forcing theorem underlies refined versions of Franks’ classical infinitude result. A first refinement shows that if \(k,n\in\mathbb N\) satisfy \((k,n)=1\) and an area-preserving annulus homeomorphism isotopic to the identity has a \(k\)-prime-periodic point, then it has infinitely many prime-periodic points whose prime periods are also coprime to \(n\) [1511.06803]. A later formulation states the same phenomenon for finite-area annulus homeomorphisms in the form
\[
(k,n_0)=1,\ \mathrm{Per}_k(f)\neq\emptyset
\quad\Longrightarrow\quad
\#\bigl\{\mathrm{Per}_{k'}(f)\mid (k',n_0)=1\bigr\}=\infty,
\]
with the special case \(n_0=2\) implying that one odd periodic point forces infinitely many odd periodic points [2202.11517].

The significance of these refinements is arithmetic rather than merely existential. Classical Franks theory gives infinitely many periodic points. The refined theorems show that the least periods can be forced inside prescribed coprimality classes, which is precisely the type of control needed in later applications to reversible dynamics, Reeb flows, and celestial mechanics [2202.11517].

## 3. Reversible dynamics and symmetric periodic points

In reversible surface dynamics, Franks’ theorem acquires a symmetric analogue. Let
\[
I:\mathbb R^2\to\mathbb R^2,\qquad (x,y)\mapsto (-x,y),
\]
and let \(\Omega\) be an \(I\)-invariant domain. A homeomorphism \(f\) on \(\Omega\) is called reversible if
\[
f\circ I_\Omega = I_\Omega\circ f^{-1}.
\]
A point \(z\in\Omega\) is a symmetric periodic point if
\[
f^k(z)=z,\qquad f^\ell(z)=I_\Omega(z)\quad \text{for some } k,\ell\in \mathbb N
\]
[1410.3997].

In this setting, every area-preserving reversible map on \(A\) or \(\mathring A\) is either periodic-point free or has infinitely many interior symmetric periodic points [1410.3997]. The reversible conclusion is stronger in flavor than the classical one: even a non-symmetric periodic point guarantees infinitely many symmetric periodic points [1410.3997]. Under isotopy to the identity, an odd-periodic point likewise forces infinitely many interior symmetric odd-periodic points [1410.3997].

This picture also admits an arithmetic refinement. For reversible finite-area annulus homeomorphisms isotopic to the identity, if there exist recurrent points with different rotation numbers, then every intermediate rational rotation number is realized by a **symmetric** \(q\)-prime-periodic orbit [2202.11517]. Consequently, if such a reversible map has a periodic point and \((k,n_0)=1\), then it has infinitely many symmetric periodic orbits with prime periods also coprime to \(n_0\) [2202.11517]. This extends Franks-type forcing from periodic points to symmetric periodic points and from qualitative infinitude to arithmetic control.

## 4. Rotation sets on the torus and the Franks–Misiurewicz program

In torus dynamics, the phrase **Franks’ theorem** often points not to a single periodic-point theorem but to a broader program in which the geometry of the rotation set constrains periodic behavior, semiconjugacy, and rigidity. For \(f\in \mathrm{Homeo}_0(\mathbb T^2)\) and a lift \(F:\mathbb R^2\to\mathbb R^2\), the rotation set is
\[
\rho(F)=\left\{\lim_i\frac{F^{n_i}(x_i)-x_i}{n_i}:\ x_i\in\mathbb R^2,\ n_i\nearrow +\infty\right\},
\]
and by Misiurewicz–Ziemian it is always a compact convex subset of \(\mathbb R^2\) [1803.03294]. The Franks–Misiurewicz conjecture proposes a classification of rotation sets with empty interior: they should be either singletons or nontrivial line segments satisfying precise rationality restrictions [1611.05498].

Part of this conjecture is known to fail: Avila produced a counterexample in the irrational-slope case, and Le Calvez–Tal proved that if a nontrivial segment has irrational slope and contains a rational point, then that rational point must be an endpoint [1611.05498]. The rational-slope case remains central. In one important class, extensions of irrational circle rotations, the rotation set is always a singleton; equivalently, every toral homeomorphism homotopic to the identity that is topologically semiconjugate to an irrational circle rotation is a pseudo-rotation [1611.05498]. This resolves the rational-slope problem in the minimal category described there.

A further restriction comes from deviation theory. If a lift satisfies
\[
\rho(F)=\{\alpha\}\times[\rho^-,\rho^+],\qquad \rho^-<\rho^+,\qquad \alpha\in\mathbb Q^c,
\]
then any hypothetical counterexample to the rational case of the Franks–Misiurewicz conjecture must have unbounded horizontal deviation [1803.03294]. More geometrically, any counterexample for the rational case must have infinite perpendicular deviation [1803.03294]. The relation to classical Franks theorems is conceptual rather than a direct generalization of a single periodic-point statement: rational data in the rotation set are expected to force dynamically realized structure, and avoiding that conclusion requires highly non-classical transverse behavior [1803.03294].

## 5. Franks’ lemma and perturbative realizability

A different but equally standard meaning of the name is **Franks’ lemma**: a perturbative realization theorem for linearized dynamics along a chosen orbit segment. In geodesic-flow form, the statement is that the derivative of the geodesic Poincaré map along a chosen geodesic segment can be perturbed freely within a neighborhood in \(Sp(n)\) by a \(C^2\)-small perturbation of the Riemannian metric that keeps the geodesic itself unchanged [1307.6573]. On surfaces this holds for every \(C^4\) metric, whereas in higher dimension the theorem is proved on a \(C^2\)-open and \(C^\infty\)-dense subset \(G_1\) of metrics [1307.6573].

A geometric-control proof rewrites the Jacobi equation as a bilinear control system on the symplectic group. For \(g_0\in R^k(M,G_1)\), \(2\le k\le\infty\), there exist \(\bar r,K>0\) such that for every geodesic arc \(\gamma\) of length \(1\),
\[
B\bigl(S(g_0),Kr\bigr)\cap \mathrm{Sp}(n-1)\subset S\bigl(B_{C^k}(g_0,r)\bigr),
\]
where \(S(\bar g)=P_{\bar g}(\gamma)(1)\) is the linearized Poincaré map [1401.8159]. The support of the perturbation can moreover be confined to a tubular neighborhood of the chosen geodesic and made disjoint from finitely many transverse geodesics [1401.8159].

A second-order uniform version, phrased for conformal or Mané perturbations, takes the form \(h=(1+\sigma)g\). For every \(T>0\) there exist \(\delta_T,\tau_T,K_T>0\) such that along any geodesic segment \(\gamma_\theta:[0,T]\to M\), every symplectic map in a sufficiently small ball around \(P_g(\gamma)(T)\) is realized by a \(C^\infty\) conformal perturbation preserving the geodesic, supported in a small geodesic cylinder, and satisfying
\[
\|\sigma\|_{C^2}<K_T\sqrt{\delta}
\]
[1502.01145]. The same perturbative philosophy extends to magnetic flows: by perturbing the magnetic field within a fixed cohomology class through exact \(2\)-forms \(d\eta\), one can realize any sufficiently small perturbation of the linearized Poincaré map along a magnetic orbit segment [1601.00935].

## 6. Symplectic generalizations and recent applications

Franks’ theorem on \(S^2\) admits a fully symplectic proof in the smooth category. Every Hamiltonian diffeomorphism of \((S^2,\omega)\) has either two or infinitely many periodic points, and if it has exactly two periodic points \(P\) and \(Q\), then both are nondegenerate elliptic fixed points with irrational mean indices satisfying
\[
\Delta(P)+\Delta(Q)=0\pmod 4
\]
[1107.1282]. The proof uses mean index theory, resonance relations of Ginzburg and Kerman, and Floer homology on the torus rather than classical low-dimensional topological dynamics [1107.1282].

This sphere dichotomy has become a model for higher-dimensional Hamiltonian analogues. One such descendant proves that on a closed strictly monotone symplectic manifold, if a strongly non-degenerate perfect Hamiltonian diffeomorphism has bounded barcode norm along the iterates \(\varphi^{2^k}\), then it must be a pseudo-rotation; consequently, if the number of fixed points exceeds the sum of Betti numbers, the number of periodic points along the \(2^k\)-tower tends to infinity [2009.13052]. A toric version goes further: for a compact symplectic toric manifold, if
\[
N(\phi):= \sum_{x \in {\rm Fix}(\phi)} {\rm dim}\, HF^{\rm loc}(\phi, x)
\]
is greater than the total rank of homology, then \(\phi\) has infinitely many simple periodic points [2309.07991]. This is presented there as a vast generalization of Franks’ famous two-or-infinity dichotomy and as a proof of the Hofer–Zehnder conjecture in the toric case [2309.07991].

Recent applications also return to the annulus theorem itself. For any reversible Finsler metric on \(S^2\), the number of prime closed geodesics grows quadratically with respect to length, and one of the two main tools is an improvement on Franks’ theorem about the number of periodic points of area-preserving annulus maps [2508.00147]. This suggests that Franks’ theorem is no longer only a periodic-point dichotomy for surface homeomorphisms: it has become a structural template for forcing arguments across symplectic dynamics, Reeb dynamics, and geometric flows [2508.00147].

Source: https://www.emergentmind.com/topics/franks-theorem