3
$\begingroup$

It is well known that compact manifolds of negative sectional curvature don't admit self-maps of degree $>1$. At the same time positively curved manifolds such as $S^n$ and $\mathbb CP^n$ clearly admit self-maps of degree $>1$. Moreover I guess if we take a compact Lie group and consider map $x\to x^2$ the degree of such map is $>1$ (in fact I did not check it).

Question. Let $G$ be a compact Lie group and $H$ its proper connected Lie subgroup. Is it true that $G/H$ admits self-maps of degree $>1$? If not, is there some classification of such quotients and that admit / don't admit a self-map with $deg>1$?

$\endgroup$

1 Answer 1

4
$\begingroup$

If $X$ is a simply connected compact manifold, then one sufficient condition for the existence of maps $X\to X$ of any sufficiently divisible degree is formality: there is a commutative differential graded algebra (cdga) $A$ and cdga maps $A\to H^*(X,\mathbb{R}),A\to \Omega^*(X)$ (the de Rham complex) that both induce isomorphisms in cohomology, see Sullivan, Infinitesimal computations in topology, \S 12, in particular, Theorem 12.2.

Lie groups are formal, and so are their classifying spaces and also many homogeneous spaces, in particular symmetric spaces, see e.g. F\'elix, Halperin, Thomas, Rational homotopy theory, \SS 12 and 15.

$\endgroup$
3
  • 2
    $\begingroup$ Combining this with weak Lefschetz and formality of compact Kahler manifolds (Deligne-Griffiths-Morgan-Sullivan), this answers the other related question mathoverflow.net/questions/112578/… by the same user. $\endgroup$ Commented Nov 16, 2012 at 16:21
  • 1
    $\begingroup$ By the way, is there any homogeneous space that is not formal? $\endgroup$
    – aglearner
    Commented Nov 17, 2012 at 0:31
  • 1
    $\begingroup$ aglearner -- yes, there are, see e.g. a recent paper by M. Amann arxiv.org/pdf/1206.0786v1.pdf and references therein. $\endgroup$
    – algori
    Commented Nov 17, 2012 at 1:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .