Ergodic decomposition of vertical foliations under mass-loss degenerations in the moduli space of quadratic differentials

McMullen (2013) proved that if the Teichmüller geodesic flow orbit of a unit-area quadratic differential $(X_0,q_0)$ accumulates on a limit $(Y,q)$ that retains full area in the compactified moduli space of stable curves, then the vertical foliation $\mathcal{F}(q_0)$ has at most $k$ ergodic components, where $k$ is the number of connected components of the smooth locus of $Y$ on which $q$ is nonzero. The same paper shows the area-retention hypothesis is essential: inserting a positive-area vertical cylinder that collapses in the limit leaves the accumulation set unchanged while $\mathcal{F}(q_0)$ acquires infinitely many ergodic components, and recurrence in the unit-area locus ceases to be a monotone property of the vertical foliation. The open problem is to control ergodic fragmentation in this mass-loss regime: to classify, and to bound sharply in terms of the degeneration data, the ergodic decomposition of $\mathcal{F}(q_0)$ when limits in the compactification do not preserve total area; to determine which mass-escaping mechanisms produce which decomposition structures; and to find sharp criteria or invariants beyond the count of nonzero components of the limit. Katok's classical bound, which needs no degeneration hypothesis, and the recent work of Masur (2022), whose argument explicitly allows mass loss, control the ergodic measures of minimal translation flows, but the general case — non-minimal foliations with closed leaves and genuine quadratic differentials — remains open.

Background

A holomorphic quadratic differential $q$ on a compact Riemann surface $X$ of genus $g$, with at worst simple poles at $n$ marked points, determines a flat structure on $X$: away from the zeros of $q$ there are local coordinates with $q=dz^2$, giving a Euclidean metric with cone-type singularities and finite total area $m(X,q)=\int_X|q|$. The flat structure carries two transverse measured foliations, the horizontal and the vertical foliation $\mathcal{F}(q)$, whose leaves are the flat trajectories of a fixed direction and whose transverse measure records flat width; closed vertical leaves sweep out flat cylinders. A measurable set is saturated if it is a union of leaves of $\mathcal{F}(q)$. The foliation, with its area measure $|q|$, is ergodic if every saturated set has measure zero or full measure; in general the measure splits into ergodic components, the finest decomposition of $X$ into saturated pieces that cannot be further divided, and the number of components quantifies ergodic fragmentation. A uniquely ergodic foliation has one component; a foliation containing a cylinder of closed leaves has infinitely many, since each closed leaf is itself an ergodic component of zero area.

The space $Q\mathcal{M}_{g,n}$ of nonzero quadratic differentials of finite mass over the moduli space $\mathcal{M}_{g,n}$ of curves carries an action of $\mathrm{PSL}_2(\mathbb{R})$ (the projective group of real $2\times2$ matrices of determinant one, acting linearly on the flat charts) preserving $m(X,q)$, and the diagonal flow $g_t=\mathrm{diag}(e^t,e^{-t})$ — the Teichmüller geodesic flow — expands the horizontal foliation of $q$ and contracts the vertical one; $Q_1\mathcal{M}_{g,n}$ denotes the unit-area locus. Masur proved that if the geodesic ray generated by $(X_0,q_0)$ is recurrent in $\mathcal{M}_{g,n}$, i.e. returns infinitely often to a compact set, then $\mathcal{F}(q_0)$ is uniquely ergodic (this is Corollary 1.5 of the source paper, restating the theorem from Masur's 1992 Duke paper). For flows that are minimal (every leaf dense) but not uniquely ergodic, a classical theorem of Katok bounds the number of ergodic probability measures of a straight-line flow on a genus-$g$ translation surface (a holomorphic 1-form, the orientable special case) by $g$, and Veech constructed genus-two quadratic differentials whose vertical foliation is minimal but not ergodic, with two ergodic components; Theorem 1.6 of the source paper shows the bound $k=2$ is attained on these examples.

The problem lives on the boundary of moduli space. The Deligne–Mumford compactification $\overline{\mathcal{M}}_{g,n}$ adds stable curves, obtained by collapsing essential simple closed curves to nodes; for a stable curve $Y$, the smooth locus is denoted $Y^*$. The bundle extends over this compactification: a point of $Q\overline{\mathcal{M}}_{g,n}$ is a quadratic differential, holomorphic on $Y^*$, with at worst simple poles at punctures and at worst double poles with equal residues at nodes. The projectivization $PQ\overline{\mathcal{M}}_{g,n}$ is compact, but flat area can disappear in limits — the mass of the approximating differentials can concentrate on subsurfaces that collapse to nodes — so the unit-mass locus $Q_1\overline{\mathcal{M}}_{g,n}$ is not compact. This phenomenon is called mass loss.

The source paper's Theorem 1.4 controls ergodicity through such limits in the mass-retaining case:

$$(X_{t_i},q_{t_i})\to(Y,q)\in Q_1\overline{\mathcal{M}}_{g,n}\ \Longrightarrow\ \mathcal{F}(q_0)\text{ has at most }k\text{ ergodic components},$$

where $k$ is the number of connected components of $Y^*$ on which $q$ is not identically zero. The proof transports the area measure of $X_0$ to the limit through Teichmüller maps (the canonical quasi-conformal maps between surfaces carrying the corresponding flat structures) composed with area-preserving maps and chains together flat rectangles; combined with monotonicity of recurrence in $\mathcal{M}_{g,n}$ (Corollary 1.3 of the source paper: a property of measured foliations is monotone if it is inherited by every foliation whose transverse measure dominates it), it yields a new proof of Masur's theorem. The hypothesis that the limit retains full mass is essential. The paper demonstrates this by slitting two genus-one differentials along a unit vertical arc and gluing them: the resulting genus-two differential has a loop $\alpha_0$ that collapses to a node, no mass is lost, and $\mathcal{F}(q_0)$ has the expected two ergodic components; but if $\alpha_0$ is replaced by a positive-area cylinder $C_0$ foliated by closed leaves (Figure 2 of the source paper), the accumulation points of $(X_t,q_t)$ in $Q\overline{\mathcal{M}}_{g,n}$ are unchanged, the mass of $C_0$ disappears in the limit, and $\mathcal{F}(q_0)$ has infinitely many ergodic components. Consequently recurrence in $Q_1\overline{\mathcal{M}}_{g,n}$, unlike recurrence in $\mathcal{M}_{g,n}$, is not a monotone property of the vertical foliation.

After 2013 the mass-loss regime has been addressed only in a special case. Masur (2022) proved that a minimal, not uniquely ergodic straight-line flow on a translation surface of genus $g$ with $s$ zeros has at most $g+s-1$ invariant generic probability measures, and his argument explicitly allows loss of mass in the Deligne–Mumford limit: generic points of different measures land in different pieces of the thick-thin decomposition of the degenerating surfaces (the standard splitting of a flat surface into a thick part of bounded geometry and thin parts consisting of collar neighborhoods of short curves), some of which are collapsing cylinders of area bounded below. His Remark 5.5 contrasts this with Theorem 1.4 of the source paper, which assumed no loss of mass. The sharpness of Masur's bound is unknown; his hypotheses exclude non-minimal foliations, in which closed leaves and cylinders are present — precisely the mechanism of the source paper's counterexample — as well as genuine quadratic differentials, whose vertical foliation need not be orientable. This is the state of the literature against which the problem is posed.

Problem Statement

Let $(X_0,q_0)$ be a quadratic differential of unit mass on a compact Riemann surface of genus $g$ with $n$ punctures, let $(X_t,q_t)=g_t(X_0,q_0)\subset Q_1\mathcal{M}_{g,n}$ be its Teichmüller geodesic flow orbit, and suppose the orbit has accumulation points $(Y,q)\in Q\overline{\mathcal{M}}_{g,n}$ in the compactification by stable curves such that $m(Y,q)<1$: some flat area escapes in the limit (mass loss). The problem is to build the mass-loss analogue of Theorem 1.4 of McMullen (2013) — a quantitative theory, in terms of the degeneration data, of the ergodic decomposition of the vertical foliation $\mathcal{F}(q_0)$. Concretely:

(i) Establish, in terms of the degeneration of its orbit — the accumulation set in $Q\overline{\mathcal{M}}_{g,n}$, the structure of the escaping mass (which cylinders or subsurfaces collapse, and with what area and rate), and the topology and singularity data of $(X_0,q_0)$ — sharp criteria or invariants, beyond the number of components of $Y^*$ supporting the limiting differential, yielding a sharp upper bound on the number of positive-measure ergodic components of $\mathcal{F}(q_0)$ together with the structure of its zero-measure components (the closed leaves and cylinders, of which there may be infinitely many).

(ii) Classify the mechanisms by which mass-escaping degenerations shape the ergodic decomposition: for each collapsing phenomenon (disappearing cylinders, vanishing subsurfaces, degenerating nodes), determine which ergodic-decomposition structures it can produce, which of them are realizable and which are excluded, and explain the source paper's cylinder example as one instance of the general mechanism.

Both deliverables are jointly required — supplying the sharp bound (i) or the classification of mechanisms (ii) alone does not resolve the problem — and together serve a single objective: to replace the mass-retaining hypothesis of Theorem 1.4 with an effective theory of the regime where it fails, without restricting to minimal foliations or to the orientable (translation-surface) case.

References

1. Curtis T. McMullen, 'Diophantine and ergodic foliations on surfaces', Journal of Topology 6 (2013), no. 2, 349–360. DOI: 10.1112/jtopol/jts033. https://doi.org/10.1112/jtopol/jts033
2. Howard Masur, 'Generic measures for translation surface flows', Journal of Modern Dynamics 18 (2022), 495–521. DOI: 10.3934/jmd.2022014. https://doi.org/10.3934/jmd.2022014 (preprint: arXiv:2002.09416, https://arxiv.org/abs/2002.09416)
3. A. B. Katok, 'Invariant measures of flows on oriented surfaces', Soviet Mathematics. Doklady 14 (1973), 1104–1108; translation of Dokl. Akad. Nauk SSSR 211 (1973), 775–778. zbMATH: https://zbmath.org/3467479
4. Howard Masur, 'Hausdorff dimension of the set of nonergodic foliations of a quadratic differential', Duke Mathematical Journal 66 (1992), no. 3, 387–442. DOI: 10.1215/s0012-7094-92-06613-0. https://doi.org/10.1215/s0012-7094-92-06613-0
5. William A. Veech, 'Strict ergodicity in zero dimensional dynamical systems and the Kronecker-Weyl theorem mod 2', Transactions of the American Mathematical Society 140 (1969), 1–33. DOI: 10.1090/s0002-9947-1969-0240056-x. https://doi.org/10.1090/s0002-9947-1969-0240056-x
6. Hossein Namazi, Alexandra Pettet, Patrick Reynolds, 'Ergodic decompositions for folding and unfolding paths in Outer space', preprint (2014). arXiv:1410.8870. https://arxiv.org/abs/1410.8870
7. Jon Chaika, John Smillie, Barak Weiss, 'Tremors and horocycle dynamics on the moduli space of translation surfaces', Annals of Mathematics (2) 204 (2026), no. 1, 1–118. DOI: 10.4007/annals.2026.204.1.1. https://doi.org/10.4007/annals.2026.204.1.1 (preprint: arXiv:2004.04027, https://arxiv.org/abs/2004.04027)
8. Jonathan Chaika, Yitwah Cheung, Howard Masur, 'Winning games for bounded geodesics in moduli spaces of quadratic differentials', Journal of Modern Dynamics 7 (2013), no. 3, 395–427. DOI: 10.3934/jmd.2013.7.395. https://doi.org/10.3934/jmd.2013.7.395 (preprint: arXiv:1109.5976, https://arxiv.org/abs/1109.5976)
9. Curtis T. McMullen, 'Cascades in the dynamics of measured foliations', Annales scientifiques de l'École normale supérieure (4) 48 (2015), no. 1, 1–39. DOI: 10.24033/asens.2237. https://doi.org/10.24033/asens.2237
10. Curtis T. McMullen, 'Teichmüller dynamics and unique ergodicity via currents and Hodge theory', Journal für die reine und angewandte Mathematik 768 (2020), 39–54. DOI: 10.1515/crelle-2019-0037. https://doi.org/10.1515/crelle-2019-0037
11. Sébastien Ferenczi, Pascal Hubert, 'Minimality and unique ergodicity of Veech 1969 type interval exchange transformations', Geometriae Dedicata 218 (2024), no. 3, article 77. DOI: 10.1007/s10711-024-00888-1. https://doi.org/10.1007/s10711-024-00888-1 (preprint: arXiv:2103.09018, https://arxiv.org/abs/2103.09018)
12. Ian Frankel, 'Meromorphic L^2 functions on flat surfaces', Geometric and Functional Analysis 32 (2022), 832–860. DOI: 10.1007/s00039-022-00611-w. https://doi.org/10.1007/s00039-022-00611-w (preprint: arXiv:2005.13851, https://arxiv.org/abs/2005.13851)
13. Dawei Chen, 'Connectedness of the boundaries of the strata of differentials', preprint (2025). arXiv:2505.07190. https://arxiv.org/abs/2505.07190
14. Curtis T. McMullen, 'Billiards and Teichmüller curves', Bulletin of the American Mathematical Society 60 (2023), no. 2, 195–250 (published online 28 November 2022). DOI: 10.1090/bull/1782. https://doi.org/10.1090/bull/1782