On quasimorphisms and distortion in homeomorphism groups

Michael Brandenbursky, Jarek Kędra, Michał Marcinkowski, Egor Shelukhin · 2025

Read On quasimorphisms and distortion in homeomorphism groups by Michael Brandenbursky, Jarek Kędra, Michał Marcinkowski, Egor Shelukhin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-15375.txt.

Introduction

Michael Brandenbursky and three coauthors study which measures of twisting survive when smooth maps give way to merely continuous ones. A quasimorphism is a map from a group to the real numbers that is almost a homomorphism, off by a bounded error. The known examples of Gambaudo-Ghys and Polterovich live on groups of volume keeping smooth maps of a manifold.

The authors pin down which of those measures extend in a continuous way to the group of measure keeping homeomorphisms that can be deformed to the identity. The extensions stay valid there, and act as cochains with semi bounded gaps on the full group. One payoff: some bi invariant metrics on the commutator subgroup are unbounded. They also give tests for when a homeomorphism is undistorted, so its powers grow at the slowest rate.

Catalog record

CREATORMichael Brandenbursky, Jarek Kędra, Michał Marcinkowski, Egor Shelukhin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-15375

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On quasimorphisms and distortion in homeomorphism groups Thanks: M.B. was partially supported by the Israel Science Foundation grant 823/23 Michael Brandenbursky, Jarek Kędra, Michał Marcinkowski, Egor Shelukhin Abstract . Let M M be a smooth compact oriented manifold, and Homeo 0 ​ ( M , μ ) {\rm Homeo}_{0}(M,\mu) the group of homeomorphisms of M M supported away from ∂ M , \partial M, which preserve a Borel probability measure μ \mu induced by a volume form on M M , and are isotopic to the identity. In this paper, we identify those Gambaudo-Ghys and Polterovich quasimorphisms Ψ : Diff 0 ​ ( M , μ ) → 𝐑 \Psi\colon{\rm Diff}_{0}(M,\mu)\to{\mathbf{R}} which extend C 0 C^{0} -continuously to Homeo 0 ​ ( M , μ ) {\rm Homeo}_{0}(M,\mu) as quasimorphisms, and to Homeo 0 ​ ( M ) {\rm Homeo}_{0}(M) as group cochains whose differentials are semi-bounded cocycles. We present several applications of this result which include unboundedness of certain bi-invariant metric on the commutator subgroup of Homeo 0 ​ ( M , μ ) {\rm Homeo}_{0}(M,\mu) , and conditions under which a homeomorphism in Homeo 0 ​ ( M ) {\rm Homeo}_{0}(M) is undistorted. 1. Introduction Let M be a smooth compact oriented manifold equipped with a Borel probability measure \mu defined by a volume form on M . Let {\rm Homeo}_{0}(M,\mu) denote the group of measure preserving homeomorphisms of M which act by the identity homeomorphism on a neighborhood of \partial M and are isotopic to the identity. Let P_{n}(M,z)=\pi_{1}({\rm C}_{n}(M),z) be the pure braid group on M . Here {\rm C}_{n}(M)\subset M^{n} denotes the space of ordered configurations of n points in M , equipped with the product {\rm Homeo}_{0}(M) action. If \dim(M)=2 then we let n\in{\mathbf{N}} be an arbitrary number, and when \dim(M)\geq 3, we assume n=1 throughout the paper. That is, for higher dimensional manifolds we consider only their fundamental groups P_{1}(M,z)=\pi_{1}(M,z) . Let \gamma\,\colon{\rm Homeo}_{0}(M)\times M\to P_{n}(M) be a measurable cocycle defined by \gamma(f,x)=\left[\ell_{z,f(x)}*\{f_{t}(x)\}*\ell_{x,z}\right], where \{f_{t}\} is an isotopy from the identity to f and \ell_{x,y} is a certain path in {\rm C}_{n}(M) from x to y which is defined in Section 2 . By convention, the concatenation of paths is read from right to left. The main object of study in the present paper is a family of maps \Psi\colon{\rm Homeo}_{0}(M)\to{\mathbf{R}} defined by (1.1) \Psi(f)=\int_{{\rm C}_{n}(M)}\varphi(\gamma(f,x))dx, where \varphi\colon P_{n}(M)\to{\mathbf{R}} is a homogeneous quasimorphism vanishing on the centre of P_{n}(M) . On {\rm C}_{n}(M) we consider the product measure \mu^{n} . The formula ( 1.1 ) was used by Gambaudo-Ghys to define non-trivial quasimorphisms on groups of area preserving diffeomorphisms of compact surfaces [ 19 ] and by Polterovich for volume preserving diffeomorphisms of compact manifolds of higher dimension [ 26 ] . It was further exploited by several authors in various configurations [ 12 , 6 , 5 , 8 , 27 ] . In general, it is not well defined for homeomorphisms and we identify quasimorphisms \varphi\colon P_{n}(M)\to{\mathbf{R}} for which it is. Here is our main result. Theorem 1.1 . Let \varphi\colon P_{n}(M)\to{\mathbf{R}} be a non-trivial homogeneous quasimorphism vanishing on the centre of P_{n}(M) . If M is a surface, we assume that M is closed of positive genus and \varphi vanishes on the subgroup P_{n}(\Delta)\leq P_{n}(M) of braids supported in an interior of a full measure two-cell \Delta , see Section 2 for definition. Then the map \Psi\colon{\rm Homeo}_{0}(M)\to{\mathbf{R}} given by ( 1.1 ) is well defined. Moreover: (1) The differential \delta\Psi is a semi-bounded 2 -cocycle, that is D(f):=\sup_{g}|\delta\Psi(f,g)|=\sup_{g}|\Psi(g)-\Psi(fg)+\Psi(f)|<\infty. (2) The restriction of \Psi to the subgroup of homeomorphisms preserving the measure \mu is a quasimorphism whose homogenization is C^{0} -continuous. (3) If \varphi extends to an unbounded quasimorphism on the full braid group B_{n}(M) then the homogeneous quasimorphism from the previous item is non-trivial. Remark 1.2 . Let \Sigma be a closed surface of a positive genus and n>1 . Recall that B_{n}(\Sigma)/Z(B_{n}(\Sigma)) , where Z(B_{n}(\Sigma)) is the center of B_{n}(\Sigma) , is a non-reducible subgroup of the n punctured mapping class group \operatorname{MCG}_{n} [ 22 , Corollary 7.13] and obviously P_{n}(\Delta)\leq\operatorname{MCG}_{n} is a reducible subgroup. It follows from Bestvina-Fujiwara construction [ 1 , Theorem 12] that the space of quasimorphisms on B_{n}(\Sigma) which vanish on P_{n}(\Delta) is infinite-dimensional. Hence there are infinitely many linearly independent quasimorphisms \varphi\colon B_{n}(\Sigma)\to{\mathbf{R}} which satisfy conditions of Theorem 1.1 . Let q:\pi_{1}(\Sigma)\to{\mathbf{R}} be a quasimorphism. It defines a quasimorphism \varphi:P_{n}(\Sigma)\to{\mathbf{R}} as follows. Let i_{*}:P_{n}(\Sigma)\to(\pi_{1}(\Sigma))^{n} be the homomorphism induced by the inclusion i:{\rm C}_{n}(\Sigma)\to(\Sigma)^{n} . We define \varphi=q\circ p_{i}\circ i_{*} , where p_{i}:(\pi_{1}(\Sigma))^{n}\to\pi_{1}(\Sigma) is the projection on the i -th factor. It is a well-known fact that the kernel of homomorphism i_{*} equals to the normal closure H_{n} of P_{n}(\Delta) in the group P_{n}(\Sigma) [ 20 , Theorem 1] , see also [ 2 ] . Thus every such \varphi vanishes on H_{n} . Let us consider the case when \Sigma=T is a torus. In this case the group H_{n} equals to the commutator subgroup [P_{n}(T),P_{n}(T)] , see [ 2 ] . Hence there are infinitely many linearly independent quasimorphisms on P_{n}(T) , that vanish on P_{n}(\Delta) , which are obviously different from each such \varphi , since in this case \varphi is a homomorphism. Let us discuss in more details the case n=2 . The commutator subgroup [P_{2}(T),P_{2}(T)]\cong{\mathbf{F}}_{2}=\langle a,b\rangle and P_{2}(\Delta)\cong{\mathbf{Z}} is generated by the commutator [a,b] . Let \operatorname{Q}({\mathbf{F}}_{2};{\mathbf{Z}}/2\times{\mathbf{Z}}/2) be the subspace of quasimorphisms on {\mathbf{F}}_{2} invariant under the action generated by inverting generators. Then every homogeneous quasimorphism \phi\in\operatorname{Q}({\mathbf{F}}_{2};{\mathbf{Z}}/2\times{\mathbf{Z}}/2) vanishes on [a,b] and hence on P_{2}(\Delta) . Moreover, each such \phi extends to B_{2}(T) , see [ 5 , Proposition 2.8] , hence by (3) in Theorem 1.1 \Psi is non-trivial. In particular, this gives an elementary construction of unbounded quasimorphisms on \operatorname{Homeo}_{0}(T,\mu) . Note that in this case, one needs to take n>1 , since for n=1 we have B_{1}(T)=Z(B_{1}(T))\cong{\mathbf{Z}}^{2} . Let us discuss the remaining case when \Sigma is a closed hyperbolic surface. We would like to point out that there are infinitely many linearly independent quasimorphisms on P_{n}(\Sigma) , that vanish on P_{n}(\Delta) , which are different from each \varphi described above. Note that, there are infinitely many linearly independent quasimorphisms on P_{n}(\Sigma) that vanish on P_{n}(\Delta) , but do not vanish on H_{n} . Indeed, by [ 1 , Theorem 12] it is enough to show that the group H_{n} is a non-reducible subgroup of \operatorname{MCG}_{n} . This can be shown as follows: Let C be a multicurve (a simplex in the curve complex). If it intersects non-trivially \Delta\setminus z , there is an element in P_{n}(\Delta) which does no preserve C . Otherwise, if C lies outside of \Delta , then since P_{n}(\Sigma) is non-reducible, there exists \alpha\in P_{n}(\Sigma) so that \alpha(C)\neq C . Note that one can choose \alpha such that \alpha(C) intersects \Delta\setminus z non-trivially. Hence there exists \beta\in P_{n}(\Delta) such that \beta\alpha(C)\neq\alpha(C) . It follows that \alpha^{-1}\beta\alpha\in H_{n} and \alpha^{-1}\beta\alpha(C)\neq C . Thus H_{n} is a non-reducible subgroup of \operatorname{MCG}_{n} . Remark 1.3 . Semi-boundedness of 2 -cocycles was introduced by Gal-Kędra [ 16 ] where it was used to prove undistortedness of symplectic diffeomorphisms for certain symplectic manifolds. It was further generalized to

Read the complete text