MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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

Recall a tranverse metric on a (regular) foliated manifold $(M,F)$ is a positive symmetric $C^\infty (M)$-bilinear form $g$ such that

1) $Ker(g_x)=T_x F$

2) It is invariant with respect to lie derivtives along vector fields tangent to the foliation.

I know that not every foliation $(M,F)$ admits such a tranverse metric, however, I would like to know some simple examples of when this fails. I do know that if the foliation arises as the fibers of a sumbersion, then it always admits a transverse metric, however I would also like to know some examples of foliations not of this form which DO admit a tranverse metric. Thank you!

share|cite|improve this question
You can take a foliation of a 2-torus by lines of an irrational slope. – Ian Agol Feb 7 '12 at 3:48

For your second kind of example, consider the foliation of the unit $3$-sphere that is the integral curves of the vector field $$ X = p\left(x^1\frac{\partial\ }{\partial x^0 } -x^0\frac{\partial\ }{\partial x^1 }\right) + q\left(x^2\frac{\partial\ }{\partial x^3 } -x^3\frac{\partial\ }{\partial x^2 }\right), $$ where $p$ and $q$ are relatively prime integers. This has a transverse metric, but it is not the fibers of any submersion from the $3$-sphere to a $2$-manifold.

As for things that don't admit transverse metrics at all, you want a foliation such that the holonomy of the leaves (actually, it's enough to have one such leaf) is not compact. A good example of this is the foliation of the unit circle bundle of a compact surface of negative curvature by the tangential lifts of geodesics of the metric.

share|cite|improve this answer
The holonomy group of any leaf is a discrete group, so what do you mean by compact? Did you mean finite? By the way, do you have a reference for this non-existence claim? Thanks! – David Carchedi Feb 7 '12 at 11:59
@David: Yes, the holonomy of each leaf is discrete in this case. If the leaf isn't closed, it's trivial and, if the leaf is closed, it's generated by a single element. However, that single element will not preserve any metric if it has a real eigenvalue of modulus greater than $1$, which is the case for closed geodesics on a surface of negative curvature. This is is thoroughly discussed under the topic of Anosov flows, which you should Google for a full explanation. The Wikipedia page on this is quite explicit. – Robert Bryant Feb 7 '12 at 12:58
@Robert: I am a little confused. The holonomy group of any leaf of any (regular) foliated manifold is discrete, by definition, since it is a quotient of the fundamental group. Are you speaking of something else? Are you claiming that if I have a foliated manifold $\left(M,F\right)$ and a leaf which is not compact (and I am working here without corners or boundaries) then its holonomy group has to have a single generator? If so, why is this true and where can I read this? – David Carchedi Feb 7 '12 at 14:14
Also, are you claiming that if there is such a non-compact leaf such that the generator of the holonomy group induces an endomorphism of the tangent space with a real eigenvalue with modulus greater than 1, then there cannot be a tranverse metric on the foliated manifold? Any references, or clarifications would be most helpful. Thank you! – David Carchedi Feb 7 '12 at 14:15
@David: You have misread what I wrote or aren't taking into account the fact that I'm talking about the particular example that I gave, in which the leaves are $1$-dimensional and thus are either circles or copies of $\mathbb{R}$. In the former case, the holonomy is generated by a single element (corresponding to the generator of $\pi_1$), in the later case, the holonomy is trivial, in this particular example. The claim about a generator with real eigenvalue greater than $1$ obstructing a transverse metric seems obvious to me, so I'm not sure what you are not getting. – Robert Bryant Feb 7 '12 at 15:55

Another (actually quite close) example of a foliation without a transverse metric: the stable foliation of the geodesic flow on the unit tangent bundle of a negatively curved manifold.

share|cite|improve this answer
Could you please give a reference for this claim? and also, why is that it does not have a transverse metric? – Camilo Angulo Feb 7 '12 at 15:51

By the corollary to theorem 4 in Kaimanovich's two page paper "Brownian motion on foliations: Entropy, Invariant measures, mixing" any foliation by leaves of sub-exponential volume growth must have a transverse invariant measure. An example of this is the foliation of the unit tangent bundle of $d$-dimensional hyperbolic space by the normal vectors to each horocycle (i.e. the strong stable foliation of the geodesic flow).

There is a $3$ dimensional manifold foliated by $2$-dimensional leaves without any transverse invariant measures which I think is due to Hirsch (I like it because because one can visualize it in Euclidean $3$ space easily).

To construct it take a solid torus $T = D \times S$ ($D$ a disk and $S$ the unit circle in the complex plane) and a solenoid map $f: T \to T$ which we take to be $s \mapsto s^2$ in the $S$ coordinate. Consider the foliation of $T \setminus f(T)$ by sets of the form $D \times \{s\}$ (each of these is a disk minus two smaller disks). Then identify the outer and inner boundaries of $T \setminus f(T)$ using $f$.

You get a compact foliated manifold (without boundary). Depending on whether the the $S$ coordinate of a point is eventually periodic under $s \mapsto s^2$ or not, the leaf the point belongs to can look like a $2$-sphere minus a Cantor set or a Torus minus a Cantor set. All leaves are dense.

The pasting you did makes is so that each time you want to go through a boundary torus you either square or take a square root of the $S$ coordinate.

In particular the holonomy along any curve which crosses the outer torus exactly once from the inside towards the outside is $s \mapsto s^2$ in the $S$ coordinate. Using this fact you can show there is no transverse invariant measure.

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.