A LENGTH COMPARISON THEOREM FOR GEODESIC CURRENTS JENYA SAPIR

2025-04-27 0 0 785.03KB 50 页 10玖币
侵权投诉
A LENGTH COMPARISON THEOREM FOR GEODESIC
CURRENTS
JENYA SAPIR
Abstract. We work with the space C(S) of geodesic currents on a closed sur-
face Sof negative Euler characteristic. By prior work of the author with Se-
bastian Hensel, each filling geodesic current µhas a unique length-minimizing
metric Xin Teichm¨uller space. In this paper, we show that, on so-called
thick components of X, the geometries of µand Xare comparable, up to a
scalar depending only on µand the topology of S. We also characterize thick
components of the projection using only the length function of µ.
1. Introduction
Let Sbe a closed, oriented, finite type surface with negative Euler characteristic.
The space of geodesic currents, C(S), contains many of the structures one might
wish to study on S. For example, it contains the set of closed curves up to homotopy,
as well as an embedded copy of Teichm¨uller space, T(S). These sets are united by
an intersection pairing i(·,·) on C(S). If µ, ν ∈ C(S) represent two closed curves,
then i(µ, ν) is just their geometric intersection number. And if µrepresents a
metric, while νrepresents a closed curve, then i(µ, ν) is the length of the geodesic
representative of νin the metric µ.
We let Cf ill(S) denote the set of filling currents, that is, those currents that have
positive intersection with all other geodesic currents. Then all currents representing
metrics are examples of filling currents. We extend the notion of length function
from currents representing metrics to all of Cfill(S). If µ∈ Cfill(S), we define its
length function `µ:C(S)Rso that
`µ(ν) := i(µ, ν)
for all ν∈ C(S).
In [HS21], Hensel and the author show that there is a continuous projection
π:Cfill(S)→ T (S)
that minimizes the length of µin the sense that `π(µ)(µ)< `X(µ) for all X6=
π(µ) T (S). We call π(µ) the length minimizer of µ. The goal of this paper is
to compare the length function of a filling current µto the length function of its
length minimizer π(µ).
To state our result precisely, let cbbe the Bers constant. Then any two curves
of length less than cbwith respect to π(µ) are disjoint. Cut π(µ) along all such
curves. Then the connected components of the result are the thick components
of π(µ). We describe how to identify these components in Theorems 1.2 and 1.3
below.
In [BIPP19], they show that if µ∈ Cfill(S), then its systolic length sys(µ) =
infαSi(µ, α) is positive, where the infimum is taken over all simple closed curves
1
arXiv:2210.00925v2 [math.GT] 18 Apr 2023
2 JENYA SAPIR
on S. For any subsurface Yof S, they also define the Y-systolic length to be
sysY(µ) = inf
αYi(µ, α)
where the infimum is taken over all essential, non-peripheral simple closed curves
in Y. Then we show the following:
Then we show that the geometries of µand π(µ) are comparable on thick com-
ponents:
Theorem 1.1. Let µ∈ Cfill(S), and let π(µ)be its length minimizer. Let Ybe a
thick component of π(µ). Then for all essential, non-peripheral simple closed curves
αin Y,
`µ(α)
`π(µ)(α)sysY(µ)
where sysY(µ)is the Y-systolic length of µ, and the constants depend only on the
Euler characteristic χ(S).
1.1. Notation. Given quantities A, B, we use say ABwith constants depending
only on Cif there is a constant cdepending only on Cso that AcB. Likewise,
we say ABif Ac0Bfor some c0>0 depending only on C, and ABif
ABand BA.
1.2. Identifying thick components of π(µ).We also characterize when curves
in π(µ) are short, and when subsurfaces are thick, in terms of the length function
of µ.
First, it turns out that a simple closed curve αis short in π(µ) if all simple closed
curves βcrossing αare relatively long with respect to µ.
Theorem 1.2. For every  > 0, there are constants N1, N2so that for any µ
P Cfill(S), any simple closed curve α, and any simple closed curve βwith i(α, β)
1,
(1) If `π(µ)(α)< , then
i(µ, β)> N1i(µ, α)
(2) If
i(µ, β)> N2i(µ, α)
then `π(µ)(α)< .
Note that this theorem is a coarse biconditional. The constant N1grows coarsely
like the width of the collar about α, while N2comes from the constant in Theorem
1.1, and is more mysterious.
Once we know which curves are short, the following theorem gives a more prac-
tical characterization of thick components of π(µ).
Theorem 1.3. Let Yπ(µ)be a subsurface so that `π(µ)(β)< cbfor each bound-
ary component βof Y, where cbis the Bers constant. Then for each essential
simple closed curve αin Y,
`π(µ)(α)1
if and only if there exists a marking Γof Yso that
i(µ, γ)sysY(µ)
for all γΓ, where all constants depend only on S.
A LENGTH COMPARISON THEOREM FOR GEODESIC CURRENTS 3
1.3. Outline and idea of proof. The paper is organized as follows.
(1) In Section 3, we prove Proposition 3.1. This is a preliminary version of
Theorem 1.1. It shows that, if µ∈ Cfill(S) and Yis a thick component of
π(µ), then
sysY(µ)2
i(µ, Γ) i(µ, α)
i(π(µ), α)i(µ, Γ)
where Γ is a shortest marking for Ywith respect to π(µ) (see Section 3 for a
definition). We spend the rest of the paper to show that i(µ, Γ) sysY(µ).
This result is completely independent from the rest of the paper. In fact,
it relies only on analogues of results in [Raf07], which we prove for geodesic
currents rather than flat structures.
(2) In Sections 4 - 7 we prove Theorem 7.3, giving a mixed collar lemma for
geodesic currents and their length minimizers. Roughly, this theorem says
the following. Let µ∈ Cfill(S), and let αand βbe two simple closed curves
with i(α, β)1. If αis short and βis long in π(µ), then µmust intersect
βmuch more than α. An idea of the proof is given in Section 4.
(3) In Sections 8 and 9, we prove a few more mixed collar theorems that follow
from Theorem 7.3.
(4) In Section 10, we prove Proposition 10.1, which says that the hyperbolically
short marking Γ of a thick subsurface Yis actually µ-short:
i(µ, Γ) sysY(µ)
The theorem then follows directly from Propositions 3.1 and 10.1.
(5) In Section 11, we prove Theorems 1.2 and 1.3 that show how to identify
short curves and thick subsurfaces of π(µ).
(6) Section 12 is an appendix proving a few identities about the collars of
geodesics in hyperbolic surfaces.
1.4. Acknowledgements. The author would like to thank D´ıdac Mart´ınez-Granado,
Giuseppe Martone and Sebastian Hensel for many helpful discussions.
2. Background and connection to related results
2.1. Geodesic currents. Geodesic currents on Swere first defined by Bonahon in
[Bon85] as follows. Fix a complete hyperbolic metric Xfor S. Identify its universal
cover with H2. Let Gbe the set of all unparameterized, unoriented geodesics in
H2. Each geodesic in H2is uniquely determined by its endpoints on the circle S1
at infinity. Thus Gcan be identified with S1×S1\/, where ∆ is the diagonal
in S1×S1, and is the relation so that (a, b)(b, a). Since π1(S) acts on H2
by isometries, it also acts on G. A geodesic current is a π1(S)-invariant, Borel
measure on G. We let C(S) denote the space of currents and endow it with the
weak* topology. It turns out that C(S) is independent of the choice of metric X.
In fact, the space of currents defined using any other metric Yis H¨older equivalent
to C(S) [Bon85].
The set of closed geodesics on Xembeds into the space of currents as follows.
Given any closed geodesic γon X, we can lift it to a subset ˜γ∈ G. Then the
current corresponding to γwill be the Dirac measure on ˜γ. Abusing notation, we
will still let γrefer to this current. There is a natural action of R+on C(S) by
scaling. It turns out that the R+orbit of the set of closed geodesics is dense in C(S)
[Bon85]. Moreover, Bonahon also shows that the geometric intersection function
4 JENYA SAPIR
i(·,·) extends continuously from pairs of closed geodesics to a bilinear, symmetric
intersection form on pairs of currents.
By work of Bonahon, Teichm¨uller space also embeds in C(S). Moreover, suppose
γis a closed geodesic and Y∈ T (S). We abuse notation slightly, and let γand Y
still denote the corresponding currents. Then,
i(γ, Y ) = `Y(γ)
In fact, many spaces of metrics on Sembed into C(S): spaces of singular flat
metrics [DLR10], metrics of variable negative curvature [Ota90], and others. The
embeddings are all characterized by the fact that intersecting a metric with a closed
curve γgives the length of the geodesic representative of γwith respect to that
metric.
2.2. Relationship to the thick-thin decomposition for quadratic differen-
tials. This theorem is inspired by an analogous result of Rafi in [Raf07]. Rafi shows
the following. Each holomorphic quadratic differential on Sinduces a singular flat
metric q. By [DLR10], this set of singular flat metrics can be viewed as filling
currents in the sense that if qis a flat structure and γis a curve, then i(q, γ) is
the length of the q-geodesic representative of γin q. Moreover, qhas a unique
hyperbolic metric Xin its conformal class. So, we get a projection from singular
flat structures (which we will think of as a subset of Cfill(S)) to T(S).
Given a hyperbolic metric Xconformally equivalent to a flat structure q, Rafi
considers the thick components of X, defined the same way as above. He shows
that if Yis a thick component of X, and αis any essential, non-peripheral simple
closed curve in Y, then, in the language of this paper,
`q(α)
`X(α)sysY(q)
where the constants depend only on χ(S) [Raf05, Theorem 1.3]. Thus, the state-
ment of the theorem of Rafi is analogous to ours, although the projections to Te-
ichm¨uller space are different. It would be interesting to see how different the two
projections are.
Question 1. Given a singular flat metric qcoming from a holomorphic quadratic
differential, let Xbe the hyperbolic metric in its conformal class, and let π(q)be its
length minimizing metric. Are Xand π(q)at a uniformly bounded distance, with
respect to some natural metric on T(S)?
The method of proof of Theorem 1.1 is also inspired by the proof in [Raf07].
Rafi’s proof relies crucially on a mixed collar lemma for quadratic differentials,
which he proves in [Raf05]. His theorem says that if Xis the hyperbolic metric
in the conformal class of a holomorphic quadratic differential q, then for any two
intersecting simple closed curves αand β,
`q(α)d·`q(β)
where d=d(`X(β)) depends only on the length of βwith respect to X, and on
χ(S) [Raf05, Theorem 1.3].
The first half of this paper involves proving an analogous result (Theorem 7.3).
If αand βare intersecting simple closed curves, and π(µ) is the length minimizer
A LENGTH COMPARISON THEOREM FOR GEODESIC CURRENTS 5
of some µ∈ Cfill(S), we show that
`µ(α)D
i(α, β)`µ(β)
where D=D(`π(µ)(β)) depends only on the length of βwith respect to π(µ). The
factor of i(α, β) in the denominator is due to the differences in the methods of proof.
We use a careful geometric argument to show how the length of µwill change if
we pinch a metric Xalong α. If µintersects αmuch less than β, then pinching α
increases the length of µcoming from crossing α, while decreasing the contribution
to length coming from intersecting β. Balancing these two effects gives our mixed
collar lemma.
We can prove a version of the length comparison theorem (Proposition 3.1) using
techniques similar to those of [Raf07, Theorem 1]. We need this result to prove the
main theorem, however its statement is rather unsatisfying on its own. To get the
full version of Theorem 1.1, we have to work specifically with length minimizing
metrics, and the techniques are rather different.
2.3. Connection to Higher Teichm¨uller Spaces. Giuseppe Martone recently
made us aware of a similar length comparison theorem in Higher Teichm¨uller theory.
Take the space of representations ρ:π1(S)P SL(3,R). Consider its Hitchin
component TH, which is a connected component of discrete, faithful, orientation-
preserving representations. There is a map from THto Cfill(S). This map is
natural in the sense that, if ρ∈ Cfill(S) represents an element of TH, then for
any closed curve γ,i(ρ, γ) gives the so-called Hilbert length of γwith respect to ρ
[BCLS18, MZ19].
Labourie [Lab07] and Loftin [Lof01] showed independently that there is a map-
ping class group-invariant projection from THto Teichm¨uller space. This is part
of a much larger, quite active research program: see [Wie19, Conjecture 14] for an
overview of the broader context.
As noted in [DM20, Lemma 5.1], work of Tholozan [Tho17, Theorem 3.9, Corol-
lary 3.10] implies a length comparison result for all but a bounded set of represen-
tations ρ∈ THwith projection X T (S). For all (not necessarily simple) closed
curves γon S,
`ρ(γ)
`X(γ)1
h(ρ)
where h(ρ) is the topological entropy of the Hilbert length of ρ, and the length
functions are the ones defined above for the associated currents. In fact, it follows
from [MZ19, Theorem 1.4, Corollary 1.5] that sys(ρ)1/h(ρ)sysY(ρ) for a thick
subsurface Yof π(ρ). The upper bound is not stated this way in their paper, but it
can be deduced using Theorem 7.3 that the function Kρin [MZ19] is the Y-systole
of ρon a certain subsurface Y. Thus, we get the inequality
sys(ρ)`ρ(γ)
`X(γ)sysY(ρ)
where the lower bound is the systole of ρon all of S, and the upper bound is the
Y-systole for a certain YS.
It should be noted that the constants in this inequality depend on the projection
Xof ρ. It would be interesting to see if one could state this theorem with constants
that depend only on the topology of S, and simultaneously tighten the two bounds
to get a statement similar to Theorem 1.1. Moreover, it would again be interesting
摘要:

ALENGTHCOMPARISONTHEOREMFORGEODESICCURRENTSJENYASAPIRAbstract.WeworkwiththespaceC(S)ofgeodesiccurrentsonaclosedsur-faceSofnegativeEulercharacteristic.BypriorworkoftheauthorwithSe-bastianHensel,each llinggeodesiccurrenthasauniquelength-minimizingmetricXinTeichmullerspace.Inthispaper,weshowthat,onso...

展开>> 收起<<
A LENGTH COMPARISON THEOREM FOR GEODESIC CURRENTS JENYA SAPIR.pdf

共50页,预览5页

还剩页未读, 继续阅读

声明:本站为文档C2C交易模式,即用户上传的文档直接被用户下载,本站只是中间服务平台,本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私,请立即通知玖贝云文库,我们立即给予删除!
分类:图书资源 价格:10玖币 属性:50 页 大小:785.03KB 格式:PDF 时间:2025-04-27

开通VIP享超值会员特权

  • 多端同步记录
  • 高速下载文档
  • 免费文档工具
  • 分享文档赚钱
  • 每日登录抽奖
  • 优质衍生服务
/ 50
客服
关注