A fundamental result in topology is that the $n$-sphere is not a retract of the $n+1$-ball. It implies that the $n$-sphere is not an absolute retract.

Is there a generalization from the sphere to closed manifolds (compact manifolds without boundary)? It would be the statement that no closed manifold is an absolute retract.

  • $\begingroup$ Perhaps a bit more properties need to be assumed about the manifold? Do you know any sufficient conditions for your theorems to work? $\endgroup$ Mar 15, 2021 at 11:38

1 Answer 1


A metrizable space is an absolute retract (AR) if and only if it is an absolute neighbourhood retract (ANR) and it is contractible.

Closed manifolds are not contractible (if $\dim M=n$ look at $H_n(M)$) hence they are not ARs. Note however that contractibility is the only obstruction for manifolds, since every topological manifold is an ANR.

  • $\begingroup$ Thank you! I see here that $H_n(M; \mathbb{Z}_2)=\mathbb{Z}_2$. $\endgroup$
    – mathieu
    Mar 15, 2021 at 12:07
  • 1
    $\begingroup$ @mathieu: Yes, every closed topological manifold is $\mathbb{Z}/2\mathbb{Z}$-orientable. For example, you can find the result and its proof in Hatcher's book, p.235 in 2nd ed. $\endgroup$
    – M.G.
    Mar 15, 2021 at 17:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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