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

In Analysis Situs, Poincaré studies the following question: which sequences of integers $b_0,\ldots,b_n$ are the Betti numbers of an orientable compact manifold of dimension $n$?.

He knows that necessary conditions are $b_k=b_{n-k}$ and if $n=4k+2$, $b_{2k+1}$ is even. Then he computes the homology of a product of spheres, reducing the problem to finding a manifold of dimension $4k$ with $b_0=b_{2k}=b_{4k}=1$ and the other are 0. He proposes the symetric product of two spheres $S^{2k}$, missing that they are singular for $k>1$.

For $n=1$ and $2$, the projective plane and the quaternionic plane answer the question and I was very surprised to learn that the Hirzebruch signature theorem implies that a manifold of dimension 12 with $b_4=0$ has signature divisible by 62 (see p.8). Hence, the smallest odd value of $b_6$ of a 12-dimensional manifold with vanishing other Betti numbers is at least 63.

Here is the question: do we know the smallest odd value of $b_6$? More generally, do we know other obstructions for realizing an arbitrary sequence of Betti numbers?

share|cite|improve this question
There are similar obstructions in dimension 20. Check my answer to this question… – Liviu Nicolaescu May 31 '13 at 20:41
The minimal $b_6$ you seek is $2$. The manifold $S^6\times S^6$ has Poincare polynomial $1+2t^6+t^{12}$. – Liviu Nicolaescu May 31 '13 at 20:45
Regarding the question when the only non-zero Betti numbers are $0$, $2k$ and $4k$: If we want the stronger statement that $H_i(M) = \mathbb{Z}$ if $i = 0,2k,4k$ and $0$ else, then the only possibilites are indeed the dimensions of the complex, quaternionic and octonionic plane, i.e. $4k = 4,8$ or $16$. For rational coefficients there are more possibilities, e.g. $4k = 32$, but a general restriction is that $k$ has to be even (and not $24$) if $M$ is simply-connected. You might have a look at this article by Su: – Lennart Meier Jun 3 '13 at 12:50
up vote 2 down vote accepted

For $M^{4k}$, the intersection form on middle-dimensional homology $H_{2k}$ is non-degenerate and symmetric, so the signature equals $b_{2k} \pmod 2$. Therefore in your case the signature theorem implies $b_6$ must be even if $b_4=0$.

See also Torsion in cohomology of smooth manifolds for some related information concerning other restrictions.

share|cite|improve this answer
This is a nice observation, thanks! Now I have to think about the realization problem in dimension 12 with $b_4>0$. Is there a 12-manifold with $b_4=b_6=b_8=1$... – Julien Marché May 31 '13 at 21:05
this is realized by CP^6, so you might want to ask more precisely whether there is a 12-manifold with b_0=b_4=b_6=b_8=b_12=1 and all other Betti numbers zero. – nsrt May 31 '13 at 22:24
I think I'm missing something here. Are we still talking about 12-manifolds? If so, what's the relation between $b_{6}$ and $b_4$? – Greg Friedman Jun 8 '13 at 2:23

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.