The Stacks project

Lemma 42.46.11. In Situation 42.7.1 let $X$ be locally of finite type over $S$. Let $E$ and $F$ be perfect objects of $D(\mathcal{O}_ X)$ whose Chern classes are defined. Then we have

\[ c_1(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F) = r(E) c_1(\mathcal{F}) + r(F) c_1(\mathcal{E}) \]

and for $c_2(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F)$ we have the expression

\[ r(E) c_2(F) + r(F) c_2(E) + {r(E) \choose 2} c_1(F)^2 + (r(E)r(F) - 1) c_1(F)c_1(E) + {r(F) \choose 2} c_1(E)^2 \]

and so on for higher Chern classes in $A^*(X)$. Similarly, we have $ch(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F) = ch(E) ch(F)$ in $A^*(X) \otimes \mathbf{Q}$. More precisely, we have

\[ P_ p(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F) = \sum \nolimits _{p_1 + p_2 = p} {p \choose p_1} P_{p_1}(E) P_{p_2}(F) \]

in $A^ p(X)$.

Proof. After choosing an envelope $f : Y \to X$ such that $Lf^*E$ and $Lf^*F$ can be represented by locally bounded complexes of finite locally free $\mathcal{O}_ X$-modules this follows by a computation from the corresponding result for vector bundles in Lemmas 42.43.4 and 42.45.3. A better proof is probably to use the splitting principle as in Remark 42.46.8 and reduce the lemma to computations in polynomial rings which we describe in the next paragraph.

Let $A$ be a commutative ring (for us this will be the subring of the bivariant chow ring of $X$ generated by Chern classes). Let $S$ be a finite set together with maps $\epsilon : S \to \{ \pm 1\} $ and $f : S \to A$. Define

\[ P_ p(S, f , \epsilon ) = \sum \nolimits _{s \in S} \epsilon (s) f(s)^ p \]

in $A$. Given a second triple $(S', \epsilon ', f')$ the equality that has to be shown for $P_ p$ is the equality

\[ P_ p(S \times S', f + f' , \epsilon \epsilon ') = \sum \nolimits _{p_1 + p_2 = p} {p \choose p_1} P_{p_1}(S, f, \epsilon ) P_{p_2}(S', f', \epsilon ') \]

To see this is true, one reduces to the polynomial ring on variables $S \amalg S'$ and one shows that each term $f(s)^ if'(s')^ j$ occurs on the left and right hand side with the same coefficient. To verify the formulas for $c_1(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F)$ and $c_2(E \otimes _{\mathcal{O}_ X}^\mathbf {L} F)$ we use the splitting principle to reduce to checking these formulae in a torsion free ring. Then we use the relationship between $P_ j(E)$ and $c_ i(E)$ proved in Remark 42.46.8. For example

\[ c_1(E \otimes F) = P_1(E \otimes F) = r(F)P_1(E) + r(E)P_1(F) = r(F)c_1(E) + r(E)c_1(F) \]

the middle equation because $r(E) = P_0(E)$ by definition. Similarly, we have

\begin{align*} & 2c_2(E \otimes F) \\ & = c_1(E \otimes F)^2 - P_2(E \otimes F) \\ & = (r(F)c_1(E) + r(E)c_1(F))^2 - r(F)P_2(E) - P_1(E)P_1(F) - r(E)P_2(F) \\ & = (r(F)c_1(E) + r(E)c_1(F))^2 - r(F)(c_1(E)^2 - 2c_2(E)) - c_1(E)c_1(F) - \\ & \quad r(E)(c_1(F)^2 - 2c_2(F)) \end{align*}

which the reader can verify agrees with the formula in the statement of the lemma up to a factor of $2$. $\square$

Comments (0)

Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.

In order to prevent bots from posting comments, we would like you to prove that you are human. You can do this by filling in the name of the current tag in the following input field. As a reminder, this is tag 0FAG. Beware of the difference between the letter 'O' and the digit '0'.