MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose that $G$ is a group and $d$ is a left-invaraint metric on $G$, e.g., the word metric (provided that $G$ is finitely-generated) or distance function determined by a left-invariant Riemannian metric on $G$ (provided that $G$ is a connected Lie group). Let $\Gamma$ be another group.

A map $f: \Gamma\to G$ is called a quasi-homomorphism if there exists a constant $A$ so that for all $\gamma_1, \gamma_2\in \Gamma$ we have $$ d(f(\gamma_1 \gamma_2), f(\gamma_1) f(\gamma_2))\le A. $$

There are two sources of unbounded quasi-homomorphisms I know:

  1. Group homomorphisms (and their bounded perturbations).

  2. Quasi-morphisms, i.e., quasi-homomorphisms to ${\mathbb R}$ (with the standard metric) or, equivalently, to ${\mathbb Z}$ (again, with the standard metric). (One can modify this construction by taking quasi-morphisms with values in other abelian groups, but I will not regard this as a separate source of examples.)

There is, by now, a considerable literature on constructing quasi-morphisms, both in geometric group theory and symplectic geometry.

One can combine 1 and 2, by taking, say, compositions and direct sums/products: I am considering this as, again a trivial modification.

Question 1: Are there are other sources of quasi-morphisms which do not reduce to 1 and 2?

If this is a bit vague, here is a sub-question:

Question 2: Suppose that $\Gamma$ is higher-rank (real rank $\ge 3$) irreducible lattice (thus, it has no unbounded quasimorphisms) and $G$ is a nonelementary hyperbolic group (to exclude homomorphisms $\Gamma\to G$ with infinite image). Are there examples of unbounded quasi-homomorphisms $\Gamma\to G$?

Is there any literature on this? I could not find anything, but maybe I was looking at wrong places.

Note that the setting of quasi-homomorphisms (in the case of nonabelian targets) is very different from the one of quasi-actions, in which case there is a substantial literature and I know most of it.

share|cite|improve this question
For question 2, note that a hyperbolic group $G$ has lots of quasi-morphisms, so any quasi-homomorphism $\Gamma\to G$ would have to lie in the "kernel" of all of these, which gives evidence to me that its image should be finite. – Ian Agol Jul 17 '13 at 5:05
Are you aware of Ozawa's work – Mikael de la Salle Jul 17 '13 at 5:27
See also the construction of Rolli – Andreas Thom Jul 17 '13 at 6:38
@MikaeldelaSalle: Thank you, I did not know about it. In particular, Ozawa proves that any quasi-homomorphism $SL(n,Z)\to G$ with $G$ amenable or hyperbolic, has bounded image, which supports the guess that answer to Question 2 is negative. – Misha Jul 17 '13 at 11:59
@AndreasThom: Thank you, I did not know of the construction. Sadly, however, Rolli needs a bi-invariant metric on the target group. The only interesting (nonabelian) examples I know are groups of Hamiltonian symplectomorphisms with Hofer metric. – Misha Jul 17 '13 at 12:02

As it turned out (contrary to my expectations), all quasihomomorphisms with noncommutative target groups (which are discrete with proper left invariant metric) are obtained via some natural constructions starting from homomorphisms to groups and quasihomomorphisms to abelian groups:

This is the main result of our paper with Koji Fujiwara here.

To the list of constructions I had in my question one would have to add extension from a finite index subgroup in $\Gamma$ and a lifting construction using a bounded cohomology class in $H^2(\Gamma, A)$, where $A$ is an abelian group. In particular, the answer to Question 2 is negative.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.