5
$\begingroup$

I am trying to make some order in the notions of cup-length, sum of Betti numbers, LS category, critical points of functions.

Let $M$ be smooth closed compact manifold. We denote by Crit($M$) the minimal number of critical points of a smooth function on $M$, $BS(M)$ the sum of Betti numbers, CL(M) its cup-length, and $LS(M)$ its LS category.

What is known about the relations between these notions ?

It seems that $BS(M) \geq CL(M)$, and it was proved that $Crit(M) \geq LS(M) \geq CL(M)$. Am I mistaken here ?

Is something kown about the relation between $LS(M)$ and $BS(M)$ ?

$\endgroup$
10
  • 1
    $\begingroup$ At the very least, you have $BS(M) \geq Crit(M)$ when $M$ is a simply connected manifold of dimension at least 5, by handle cancellation techniques a la Smale's proof of the h-cobordism theorem. $\endgroup$
    – mme
    Oct 17, 2018 at 14:52
  • 4
    $\begingroup$ The comment describes how to construct a Morse function on $M$ with $BS(M)$ critical points. $\endgroup$
    – mme
    Oct 17, 2018 at 15:03
  • 2
    $\begingroup$ Combining that with the inequality $Crit(M) \geq LS(M)$ which is already in your post gives $BS(M) \geq LS(M)$. Observe that this inequality is not true for arbitrary manifolds: On page 354 of this article, it is explained that $LS(P) = 3$, where $P$ is the Poincare homology sphere, which has $BS(P) = 2$. $\endgroup$
    – mme
    Oct 17, 2018 at 15:19
  • 2
    $\begingroup$ Yes, many. See the book "Lusternik-Schnirelmann category" by Cornea-Lupton-Oprea-Tanre for a good discussion of this. It is a good reference for many related topics. In particular, Proposition 7.26 states that a closed connected manifold of dimension $n$ has $Crit(M) \leq n + 1$. So, for instance, every surface of genus $g$ admits a smooth function with three critical points. $\endgroup$
    – mme
    Oct 17, 2018 at 15:51
  • 2
    $\begingroup$ See also the answers to this question mathoverflow.net/questions/29434 $\endgroup$
    – j.c.
    Oct 17, 2018 at 17:32

0

Your Answer

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