14

It is known that $\mathbb{R}^4$ has exotic smooth structures, and there are many such examples in higher dimensions, such as the famous 7-sphere. My (probably very naive) question is, for every $n\geq4$, does there exist an $n$-manifold with exotic smooth structures?

In other words, for every $n\geq4$, does there exist topological $n$-manifolds which admit more than one diffeomorphism class of smooth structures?

timur
  • 3,292
  • 1
  • 35
  • 41
  • 2
    What is an "exotic smooth structure"? – Mariano Suárez-Álvarez May 18 '16 at 04:32
  • It is the phenomenon that one underlying topological manifold to have more than one smooth structures. From nLab: An exotic smooth structure is, roughly speaking, a smooth structure on a topological manifold X which makes the resulting smooth manifold be non-diffeomorphic to the smooth manifold given by some evident ‘standard’ smooth structure on X. – timur May 18 '16 at 04:36
  • 4
    Are you asking if for every n there exist topological n-manifolds which admit more than one diffeomorphism class of smooth structures? – Mariano Suárez-Álvarez May 18 '16 at 04:48
  • 1
    I then suggest you edit the question and rephrase it. – Mariano Suárez-Álvarez May 18 '16 at 04:58
  • 3
    Done! Although 'exotic smooth structure', 'exotic manifold' etc. seem to be standard terms. – timur May 18 '16 at 05:09
  • 7
    Exotic only makes sense when there is a standard smooth structure, and then asking for exotic structures on random manifolds does not make much sense. In the case of spheres, for example, the term distinctly refers to smooth structures different from in usual one, but there is no "usual one" on a general manifold. – Mariano Suárez-Álvarez May 18 '16 at 05:13
  • 3
    Perhaps, but it not so difficult to see what is being asked either, if the person who is reading is a bit forgiving and an expert who is actually willing to answer the question. – timur May 18 '16 at 05:16
  • 6
    A precise question is always better. – Mariano Suárez-Álvarez May 18 '16 at 05:28
  • 2
    I would say that, in dimension 4, the expression "exotic manifold" is quite common (though, admittedly, imprecise). I like the expression "exotic pair" to denote two non-diffeomorphic smooth structures on a topological manifold a lot more, though. – Marco Golla May 18 '16 at 06:59
  • 7
    From a recently arXiv'd paper: "Following results of Moise [35], Kervaire-Milnor [25], Browder [10] and Hill-Hopkins-Ravenel [19], we show that the only odd dimensional spheres with a unique smooth structure are $S^1$, $S^3$, $S^5$ and $S^{61}$." http://arxiv.org/abs/1601.02184 This post gives a lucid summary of the results: http://math.stackexchange.com/a/1609522/155629 – Travis Willse May 18 '16 at 07:27
  • As for even-dimensional spheres, I believe S^12 is the only S^n, n ≥ 8 currently known to have only one smooth structure – Daniel Asimov Mar 27 '23 at 01:30

1 Answers1

21

Yes. For every $n\ge 5$ there are exotic tori.

In fact, the PL-structures on $T^n$ are in one-to-one correspondence with $H^3(T^n;\mathbb{Z}/2)$, and every one of these is smoothable (Reference: "Surgery on Compact Manifolds" by C. T. C. Wall, Chapter 15A). Since any smooth manifold admits a unique PL-structure up to PL-isomorphism, it follows that there are many manifolds homeomorphic but not diffeomorphic to the standard torus.

Mark Grant
  • 35,004
  • Thanks a lot! I am sure this builds on contributions from many people, but if I want to attribute this result to a few individuals, who would be most appropriate? – timur May 18 '16 at 16:14
  • 2
    I'm not an authority on this myself, but search for papers on "homotopy tori" and "fake tori" around 1969-70. I think the main names are Hsiang and Shaneson, Wall and Casson (all of whom apparently proved this result independently using the surgery machine). – Mark Grant May 18 '16 at 16:25
  • 1
    Is the dimension range for the sentence beginning with "In fact" n ≥ 5 ? – Daniel Asimov Mar 27 '23 at 02:31
  • 1
    There are not really as many PL structures on $T^n$ as that: typically one mods out $H^3(T^n;\mathbb{Z}/2)$ by the action of $GL(n,\mathbb{Z}/2)$. This is discussed by Wall in the reference given. Nonetheless, $T^n$ does have multiple PL structures up to PL-homeomorphism, for all $n \geq 5$, so the thrust of the answer is correct. – Dave Davidson Mar 27 '23 at 00:50