1 Statement of the main theorems
In low dimensional analytic dynamics, a fundamental question is wether far from periodic points,
the dynamics is rigid. In real or complex dimension 1, dynamics are there rigid [Yoc84,Sul85]: the
dynamics restricted to an invariant domain without periodic point is either a rotation or in the
basin of periodic points. In dimension 2, for real analytic volume preserving diffeomorphisms of
the cylinder or the real sphere, the Birkhoff rigidity conjecture for pseudo-rotations [Bir41] states
that such are also conjugate to rotations. We are going to disprove this conjecture. More precisely
we are going to give two examples of entire real symplectomorphisms of the annulus A=R/Z×R,
without periodic points and such that outside a region A0
0of Abounded by two disjoint analytic
curves, the dynamics is analytically conjugate to a rotation, while inside A0
0the dynamics is not
rigid. In the first example, the restriction f|A0
0will be ergodic – see Theorem A– while in the
second example it will be extremely far from being ergodic – see Theorem C. More precisely the
ergodic decomposition of the latter map will be infinite dimensional and even of maximal local
order, i.e. with local emergence of maximal order 2. This confirms a conjecture on typicality of
high emergence in many categories [Ber17].
From this we will deduce a corollary disproving the Birkhoff rigidity conjecture. The proof of
the main theorems are shown by developing the Anosov-Katok method [AK70], together with a
new approximation theorem for entire symplectomorphisms developing a recent work with Turaev
[BT22]. The corollary regarding the sphere is obtained by using a blow-down techniques introduced
in [Ber22].
Finally we will remark that these analytic constructions give examples of entires symplectomor-
phisms of C/Z×Cwithout periodic points and with a non-empty instability region J.
1.1 Ergodic analytic and symplectic pseudo-rotation
Let Sbe an orientable analytic surface. We recall a diffeomorphism of Sis symplectic if it
preserves the orientation and the volume.
Conjecture (Birkhoff [Bir41, Pb 14-15]).An analytic symplectomorphism of the sphere with only
two fixed points and no other periodic point is necessarily topologically conjugate to a rotation.
Analogously, an analytic symplectomorphism of a compact cylinder without periodic points is
topologically conjugate to a rotation.
While following Birkhoff “considerable evidence was adducted for this conjecture”, Anosov-
Katok [AK70] gave examples of smooth symplectomorphisms of the sphere or the annulus with
resp. 2 or 0 periodic points which are ergodic. Anosov and Katok proved their theorems by
introducing the approximation by conjugacy method, that we will recall in the sequel. Also,
in his famous list of open problems, Herman wrote that a positive answer to the following question
would disprove the Birkhoff rigidity conjecture:
Question 1.1 (Herman [Her98][Q.3.1]).Does there exist an analytic symplectomorphism of the
cylinder or the sphere with a finite number of periodic points and a dense orbit?
Corollary Bwill disprove both Birkhoff’s conjectures and answer positively to Herman’s question
in the cylinder case. It will also bring a new analytic case of application of the approximation by
conjugacy method, as wondered by Fayad-Katok in [FK04,§7]. To state the main theorems, we set:
T:= R/Z,I:= (−1,1),A:= T×R,and A0=A×I.
We recall that a map of Ais entire if it extends to an analytic map on C/Z×C. An entire
symplectomorphism is a symplectomorphism which is entire and whose inverse is entire.
2