18
$\begingroup$

In a vague sense, what I want to ask is to what extent can the most general Riemann-Roch theorems be regarded as adelic duality statements?

For example, the Riemann-Roch theorem for a (proper, irreducible) curve follows from self duality of the adeles of its function field. This duality can either be proved directly for a completely adelic proof, or viewed as a simple consequence of Serre duality. Direct proofs (through residues) often require the assumption that the base field is perfect, but I'd like to neglect that here.

Self-duality of the additive group of the adeles of proper, irreducible surface $S$ over a field is sufficient to imply the following statement for a Cartier divisor $D$ on $S$:

$(D,C-D)=2(\chi(\mathcal{O}_S)-\chi(\mathcal{O}_S(D)))$,

where $C$ is the canonical divisor. This statement differs from the Hirzebruch-Riemann-Roch theorem by the following formula (Noether's formula):

$\chi(\mathcal{O}_S)=\frac{1}{12}(C^2+c_2)$,

where $c_2$ is the second Chern class. Is it reasonable to expect that this can be interpreted as a duality statement, perhaps of the K-groups of the adeles?

What then, of higher dimensions, and the Grothendieck-Riemann-Roch theorem?

$\endgroup$
1
  • $\begingroup$ There is a link between duality for coherent sheaves and Grothendieck-Riemann-Roch. Consider the Lefschetz-Verdier formula for coherent sheaves (see SGA 5, Exp. III, Appendix) and (with the terminology of the appendix) let $X_1=X_2=X$, $Y_1=Y_2=Y$ and $C'=C''=$diagonal, $D'=D''$=diagonal. Then the formula gives (although this is difficult to show...) Grothendieck-Riemann-Roch with values in Hodge cohomology, ie $\oplus_{p}H^p(Y,\Omega^p)$. The Lefschetz-Verdier formula depends only on duality. $\endgroup$ Jun 4, 2013 at 19:06

1 Answer 1

1
$\begingroup$

The closest I have seen to what you are asking, is the paper

Hübl, Reinhold; Yekutieli, Amnon Adelic Chern forms and applications. Amer. J. Math. 121 (1999), no. 4, 797–839.

Not quite Riemann-Roch, but probably the right approach.

$\endgroup$

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.