Let $A \rightarrow B$ and $C \rightarrow B$ be two maps of schemes. How can I compute the derived fiber product $A \otimes^L_B C$? I'm guessing this is a dg-scheme.

For instance - let $B=\mathbb{A}^1$ and $A = 0 \in \mathbb{A}^1$, and $C \in \mathbb{A}^1$ some point (either $0$ or $1$).

Question: The example I'm really interested in is $\tilde{\mathfrak{g}} \otimes^L_{\mathfrak{g}} 0$, where $\mathfrak{g}=\mathfrak{sl}_2$ and $\tilde{\mathfrak{g}}=\{ X,(0 \subset V \subset \mathbb{C}^2) | X \in \mathfrak{sl}_2, X V \subset V \}$ is the Grothendieck-Springer resolution.

  • 2
    $\begingroup$ Is there even an ordinary "tensor product" of schemes? $\endgroup$ – Dylan Wilson Sep 19 '13 at 2:46
  • 2
    $\begingroup$ "tensor" = "fiber". $\endgroup$ – Sasha Sep 19 '13 at 3:05
  • 1
    $\begingroup$ The tensor product of commutative rings corresponds to the fiber product of the associated affine schemes, but I think one should be careful with terminology and notation. $\endgroup$ – Fernando Muro Sep 19 '13 at 16:39

As usual, you replace $A$ (or $C$) with a (sheaf of) DG-algebras which is flat over $B$ and compute the usual tensor product of it with $O_A$ over $O_B$. In case when $A$ is a complete intersection subscheme a good choice for such DG-algebra is the Koszul complex. Then the derived fiber product is given by the pullback to $C$ of the Koszul resolution of $O_A$.

For example, if $A = 0 \subset \mathbf{A}^1$ then $O_A \cong \{k[t] \stackrel{t}\to k[t]\}$, so the derived fiber product is empty if $C = 1$ and $\{k \stackrel{0}\to k\} = k[\epsilon]/\epsilon^2$ with $\deg\epsilon = -1$, $d\epsilon = 0$ if $C = 0$. The same computation works as well in your second example.

  • $\begingroup$ Thanks - that makes sense! But for the example, $A=\tilde{\mathfrak{g}} \rightarrow \mathfrak{g}=B$ -- what does the corresponding sheaf of dg-algebras (ie. the Koszul complex) look like? $\endgroup$ – Vinoth Sep 21 '13 at 3:11
  • $\begingroup$ It is a simple exercise, try to do it yourself. $\endgroup$ – Sasha Sep 21 '13 at 4:03
  • $\begingroup$ When $A$ is not a complete intersection you can rewrite the morphism $A \to B$ as a composition of a closed embedding $A \to \tilde B$ and of a flat morphism $\tilde B \to B$. Then $A\times^L_B C=A\times^L_{\tilde B}(\tilde B\times_BC) = A\times^L_{\tilde B}(\tilde B\times_B C)$. $\endgroup$ – Sasha Sep 23 '13 at 5:55
  • $\begingroup$ Ah okay. But what is the precise definition of a "complete intersection subscheme"? Hartshorne doesn't have a definition either. $\endgroup$ – Vinoth Sep 23 '13 at 18:36
  • $\begingroup$ I understand what a flat module over a ring is but -- given a map $A \rightarrow B$ what does it mean for a sheaf of DG-algebras on $A$ to be flat over $B$? Thanks. $\endgroup$ – Vinoth Sep 23 '13 at 18:48

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.