Dear Colleagues and Friends,

Please let me know if you are aware of any references to the following question.

The classical result of Atiyah, Patodi and Singer tells us that if $W$ is a compact oriented Riemannian 4-manifold with boundary $M$ and, moreover, if we assume that near M the metric is isometric to a product, then $$ sign(W)= \frac{1}{3} \int_W p_1 - \eta(M),$$ where $p_1$ is the differential form representing the first Pontryagin class of $W$, and $\eta$ is the eta-invariant of $M$.

What about the case when both $W$ and $M$ are hyperbolic manifolds and are allowed to have cusps? Or, say, $W$ and $M$ are Riemannian as above, with infinite ends of finite volume, on which the metric is isometric to a product?

Any information will be appreciated. Please excuse my ignorance as differential geometer.

Correction: for hyperbolic manifolds with boundary and/or cusps the metric near the boundary is not a product (otherwise we won't have finite volume, I'd suppose). There is a correction term (already in the paper by Long and Reid), which is, however, vanishing due to various reasons (e.g. for totally geodesic boundary its second fundamental form vanishes, and it annihilates the correction term, and for the cusp case we can deduce it from the fact that the volume of cusp section by a small horoball is 0 in the limit, and we integrate over that horoball).

  • 2
    $\begingroup$ This is a nice question, but probably very hard. A starting point might be a version of the eta-invariant for the Dirac operator (rather than signature operator) on an odd-dimensional hyperbolic manifold with a cusp, due to J. Park (Amer. J. Math. 127 (2005), no. 3, 493–534 or arxiv.org/abs/math/0111175.) $\endgroup$ Commented Aug 5, 2018 at 18:28
  • 1
    $\begingroup$ Perhaps, the following will help "Index theorems on manifolds with straight ends" by Werner Ballmann, Jochen Bruning and Gilles Carron, mathematik.hu-berlin.de/~geomanal/bruening_publications/…. This is not on my fingertips so I cannot claim anything definite. $\endgroup$ Commented Aug 5, 2018 at 19:53
  • $\begingroup$ @DannyRuberman: Thanks, Danny. Ok, I see: it might be hard, and I'm no real differential geometer. However, this question appears extremely interesting in the context of geometric cobordisms of hyperbolic 3-manifolds. If you're interested, I can write more by e-mail. $\endgroup$ Commented Aug 5, 2018 at 21:18
  • $\begingroup$ @IgorBelegradek: Thanks, Igor! Just in case, my question is related to this paper (arxiv.org/pdf/math/0007197.pdf) of Long and Reid about geometric cobordisms of hyperbolic 3-manifolds. $\endgroup$ Commented Aug 5, 2018 at 21:19

1 Answer 1


This type of questions has been investigated systematically by Melrose in the framework of 'c-calculus', where $c$ stands for the cusp. The basic idea, if I recall correctly is to blow up the boundary with the cusp three times (two blow ups for the boundary, an extra blow-up for the cusp on the boundary) and work with the heat operator associated with the Dirac operator on the blow-up space. Normally this does not help you "gain" anything since the metric is arbitrary. In your case, however the metric is product type. So your can explicitly recover the $\eta$-invariant and there will be an extra contribution from the third blow up. The exact meaning of this contribution may be difficult to interpret from the view of spectral flow.

I am not entirely sure what is a good reference for this, though. There is no good reference for $c$-calculus (Melrose has a book on $b$-calculus, which is dense). This paper may be the closest thing I can find at the moment (there should be expository articles on this written by Rafe, here is an article by Paul). I am no longer working in the area, so there may be mistakes from my poor memory.

  • $\begingroup$ Thanks. As I corrected myself above, the metric in the case of finite-volume hyperbolic manifolds is not a product near the boundary, but there is a vanishing correction term. However, this is the presence of cusps that's real concern. Also, I'm not aiming at computing the invariant or getting a very good formula for it: I need to know that some non-trivial invariant of the cusped boundary manifold has to be integral. Just to show that being "bounding" is, in principle, a non-trivial property. $\endgroup$ Commented Aug 6, 2018 at 9:09
  • $\begingroup$ @SashaKolpakov: Yes, I was confused with the presence of finite volume and product type metric. I suppose some explicit formula for your boundary condition may be helpful. $\endgroup$ Commented Aug 6, 2018 at 14:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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