Skip to main content
17 events
when toggle format what by license comment
Mar 9, 2011 at 5:54 comment added Ryan Budney @Kelly, I think your meta-discussion might be more appropriate for meta.mathoverflow.
Mar 8, 2011 at 18:18 comment added Kelly Davis @Ryan Maybe I am just missing something, which may be the case, but I don't understand how you're answering the posed question. The question you answer is: "Does there exist an $m < 2n$ and a connected smooth $m$-manifold into which any connected closed smooth $n$-manifold smoothly embeds?". While, as far as I can tell, the posed question is: "Does there exist an $m < 2n$ and a connected closed smooth $m$-manifold into which any connected closed smooth $n$-manifold smoothly embeds?" Notice, another "closed" was introduced into the second question. Are you answering the second question too?
Mar 7, 2011 at 22:36 comment added Ryan Budney I'm not certain where that result first appears, but the technology is called "smoothing theory". This is a tool that allows you to enumerate all the smooth structures that are compatible with a given PL manifold structure. Hirsch and Mazur "Smoothings of piecewise linear manifolds" and Kirby and Siebenmann's "Foundational essays on topological manifolds, smoothings and triangulations" are standard references.
Mar 7, 2011 at 22:23 comment added Alessandro Sisto Thanks, I didn't know that result on triangulations.
Mar 7, 2011 at 15:38 comment added Ryan Budney Hi Alessandro, there's many ways to argue this. One is that all smooth manifolds have triangulations, and there's only countably-many triangulations. Moreover, there's only finitely many smooth structures supported by a given triangulation.
Mar 7, 2011 at 13:08 comment added Alessandro Sisto The connected sum you describe is second countable only if there are countably many closed n-manifolds up to diffeomorphism. Please forgive me for my ignorance: is this the case? Why?
Mar 6, 2011 at 23:46 comment added Kelly Davis @Ryan In Edit 2 the OP's requires all manifolds, the "source" manifolds $M_i$ and the "universal" target manifold $U$, are closed, compact without boundary. So, my original objection rules out $(M_1 \times S^1) \# (M_2 \times S^1) \# \cdots$ and as $U$ must be closed $\mathbb R^{n+1} \# (M_1 \times S^1) \# (M_2 \times S^1) \# \cdots$ is ruled out.
Mar 6, 2011 at 23:22 comment added Ben McMillan That is a very interesting solution, thanks! I wasn't even aware that Whitney embedding could be improved for Euclidean embeddings.
Mar 6, 2011 at 23:19 vote accept Ben McMillan
Mar 6, 2011 at 23:19
Mar 6, 2011 at 22:05 comment added Kelly Davis @Ryan I see. So, if we limit ourselves to compact $n$-manifolds $M_i$, then it becomes "interesting" once again as $M_i \times S^1$ is always compact.
Mar 6, 2011 at 22:02 comment added Ryan Budney @Gil, as Andy mentions this is Massey's Immersion Conjecture. en.wikipedia.org/wiki/Whitney_immersion_theorem The analogous conjecture for embeddings hasn't been formulated let alone proven, as far as I know.
Mar 6, 2011 at 22:01 comment added Ryan Budney @Kelly: the reason why I'm choosing this non-compact setting is to ensure the infinite connect-sum makes formal sense. Mazur's swindle construction doesn't live entirely in the realm of smooth manifolds.
Mar 6, 2011 at 21:54 comment added Andy Putman @Gil : That's for immersions, not embeddings. It's the Immersion Conjecture, which was proven by Ralph Cohen.
Mar 6, 2011 at 21:54 comment added Ryan Budney Kelly, I think there is a mis-communication over what is meant by "infinite connect-sum". Such arguments aren't a problem in the realm I'm working in because the manifolds are non-compact. If you want to make what I said more precise, the idea would be to take $\mathbb R^{n+1}$, and pick a countable discrete collection of disjoint points in $\mathbb R^{n+1}$, and use them as the "attaching points" for the connect sum operation. This doesn't cause any Mazur-swindle problems since the resulting manifold is non-compact.
Mar 6, 2011 at 21:50 comment added Kelly Davis @Ryan If I understand correctly, your answer is $(M_1 \times S^1) \# (M_2 \times S^1) \# \cdots$. This won't work. An infinite connected sum is not defined. For example, we know there exists an exotic $S^7$, notated as $S_+$ say, and it has an inverse $S_-$ such that $S^7 = S_- \# S_+$. If an infinite connected sum were defined, we would have $S_+ = S_+ \# (S_- \# S_+) \# (S_- \# S_+) \cdots$ $= (S_+ \# S_-) \# (S_+ \# S_-) \# \cdots$ $= S^7$ which is false as $S_+$ is exotic.
Mar 6, 2011 at 21:42 comment added Gil Kalai Dear Ryan, I vaguely remember a result about embedding in dimension 2n-b(n) where b(n) is the number of 1's in the binary expansion of n. But I dont remember if it was for the same problem.
Mar 6, 2011 at 20:56 history answered Ryan Budney CC BY-SA 2.5