Definition

Let FF be a , and let rr be a finite-dimensional representation of its or with finite inertia image. Choose a finite through which the factors, and write GiG_i for its lower-numbered ramification groups, with G0G_0 the inertia group. The Artin conductor exponent is

a(r)=i0GiG0codimVGi.a(r)= \sum_{i\geq 0} \frac{|G_i|}{|G_0|} \operatorname{codim} V^{G_i}.

This integer is independent of the chosen finite extension. It is zero exactly when rr is unramified.

Tame and wild parts

The term for i=0i=0 is dimVdimVIF\dim V-\dim V^{I_F} and measures tame failure of inertia invariance. The remaining sum is the Swan conductor, which measures . Thus

a(r)=codimVIF+Swan(r).a(r)=\operatorname{codim}V^{I_F}+\operatorname{Swan}(r).

The conductor is additive in and direct sums.

Weil–Deligne form

For a (r,N)(r,N), the conductor also detects monodromy:

a(r,N)=a(r)+dimVIFdim(kerN)IF.a(r,N) = a(r)+\dim V^{I_F}-\dim(\ker N)^{I_F}.

This is the exponent that appears in the power of qFsq_F^{-s} in a . It also records the ramification contribution of a local parameter to its .

Global conductor

For a global representation unramified outside finitely many finite places, the local exponents assemble into the conductor ideal

f(r)=vpva(rv).\mathfrak f(r)=\prod_v\mathfrak p_v^{a(r_v)}.

The global conductor is therefore arithmetic data assembled from local ramification, not a separate choice of normalization.

References
  1. Jean-Pierre Serre, Local Fields, Graduate Texts in Mathematics 67, Springer, 1979, Chapter VI, §2. Springer.
  2. Pierre Deligne, “Les constantes des équations fonctionnelles des fonctions LL,” in Modular Functions of One Variable II, Lecture Notes in Mathematics 349, Springer, 1973, 501–597. IAS copy.