5
$\begingroup$

This is a follow up on my previous question for Lectures on Condensed Mathematics. I am reading ahead at page 43. But it is not directly clear to me from the results that:

  1. How do we know $\Bbb Z[[T]], \Bbb R, \Bbb Z_p$ are solid modules?

  2. How did we obtain that $\Bbb R^{L \blacksquare}=0?$

  3. For the last equation 6.4 to holds, does $-\otimes^{L\blacksquare}-$ commute with filtered limits in each variable?

Some elaborations would help. I'd also like to know what are the formal aspects of the deductions (which I suppose is the great part this new category).


Basically all I know which are solid: are

  1. those of the form $(-)^{L\blacksquare}$,
  2. Those which are local objects. Characterization 5.8.
  3. $\prod_I \Bbb Z$, the cpt. projective generators.

$\endgroup$
1
  • 1
    $\begingroup$ For 2, see Cor. 6.1(iii). In particular, $\mathbf R$ is very far from solid. As for $\mathbf Z[[T]]$, it is isomorphic to a product of copies of $\mathbf Z$, so that is clearly solid. For $\mathbf Z_p$ I think you can use the short exact sequence $0 \to (T-p)\mathbf Z[[T]] \to \mathbf Z[[T]] \to \mathbf Z_p \to 0$. The last equation of 6.4 follows immediately from Prop. 6.3. $\endgroup$ Dec 16, 2020 at 18:36

1 Answer 1

5
$\begingroup$

Let me basically repeat what R. says in the comments.

For 1, $\mathbb{Z}[[T]]$ and $\mathbb{Z}_p$ are solid becuase $\mathbb{Z}$ is solid and solid abelian groups are closed under all limits and colimits. But $\mathbb{R}$ is not solid, see 2.

For 2, the first point is that $\mathbb{R}$ is pseudocoherent as a condensed abelian group, i.e. the following equivalent conditions are satisfied:

  1. $Ext^i(\mathbb{R},-)$ commutes with filtered colimits for all i;
  2. $\mathbb{R}$ admits a projective resolution where the terms are compact projective condensed abelian groups.

(The equivalence of 1 and 2 is valid in any abelian category generated by compact projectives. It's a good exercise in homological algebra if you'd like to try.)

The pseudocoherence of $\mathbb{R}$ follows from the short exact sequence $\mathbb{Z}\rightarrow\mathbb{R}\rightarrow\mathbb{R}/\mathbb{Z}$. Indeed, $\mathbb{Z}$ is clearly pseudo-coherent (it is projective and compact) and the Breen-Deligne resolution implies that any compact abelian group is psuedo-coherent. Equivalent condition 1) shows that pseudocoherent modules have the 2 out of 3 property in short exact sequences, so this implies that $\mathbb{R}$ is pseudo-coherent.

The second point is that for a pseudocoherent condensed abelian group $M$ the derived solidification of $M$ identifies with $\underline{RHom}(\underline{RHom}(M,\mathbb{Z}),\mathbb{Z})$. Indeed, using condition 2 one reduces to checking that when $M=\mathbb{Z}[S]$ for $S$ extr. disconnected we have that $\underline{RHom}(\mathbb{Z}[S],\mathbb{Z})$ lives in degree $0$ and $\underline{RHom}(-,\mathbb{Z})$ on it is the derived solidification of $\mathbb{Z}[S]$. But these we verified in producing the solid theory (they follow from Specker's theorem and the definition of solidification).

Thus it suffices to see that $\underline{RHom}(\mathbb{R},\mathbb{Z})=0$. But this was verified in the proof of the existence of the solid theory. It follows from the Breen-Deligne resolution and the fact that $\mathbb{R}$ is contractible (so its cohomology with $\mathbb{Z}$-coefficients vanishes).

For 3, the answer is yes (Edit: no, I misread and thought the question was about filtered colimits. See comments) . For ordinary tensor product this is clear, and it follows for derived tensor product because filtered colimits are exact so they have vanishing higher derived functors.

$\endgroup$
9
  • $\begingroup$ P.S.: if this answer doesn't make sense to you please do feel free to ask for clarification. $\endgroup$ Dec 16, 2020 at 21:56
  • $\begingroup$ Thanks again! For 1. This follows by Solid is a reflective localization? For 3. I thought we want the case for "inverse system \projective of limits " rather for colimits. And in this case, I don't see why $\otimes$ commutes with it: i.e. in $\Bbb Z$, $\Bbb Q \otimes (\varprojlim \Bbb Z/p^n) = 0 \not= \Bbb Z_p \otimes \Bbb Q=\Bbb Q_p$. $\endgroup$
    – Bryan Shih
    Dec 16, 2020 at 23:02
  • $\begingroup$ Sorry, in 3 I misread and thought you asked about filtered colimits. Filtered limits do not commute with solid tensor product in general. If they did, so would arbitrary products, by writing a product as a filtered limit of finite products which are direct sums. But you can see that if $M= \oplus \mathbb{Z}$ then $M \otimes \prod \mathbb{Z} = \oplus \prod \mathbb{Z}\neq \prod \oplus \mathbb{Z}$ when you have infinite product and direct sum. $\endgroup$ Dec 17, 2020 at 6:42
  • $\begingroup$ Now i'm confused how we computed $\Bbb Z[[U]] \otimes^{L \blacksquare} \Bbb Z[[T]] $ : The text said 6.3 is used, but to me: $\Bbb Z[[U]] \not= \prod_{\Bbb Z} \Bbb Z$. but is an inverse limit. $\endgroup$
    – Bryan Shih
    Dec 17, 2020 at 7:48
  • 1
    $\begingroup$ As a condensed abelian group, $\mathbb{Z}[[U]] = \prod_{\mathbb{Z}} \mathbb{Z}$, they are the same inverse limit of finite products. $\endgroup$ Dec 17, 2020 at 8:51

Your Answer

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

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