Diophantine
Questions over all Residue Classes of a Number Field and where the
topic is today
Gist of the paper: It has
two straightforward points. First:
That expanding diophantine equations to elementary statements, while it
includes important number theory topics, it isn't sufficiently general
to have important conclusions. To wit: Even when over the integers and
asking about reductions moduli infinitely many primes, you couldn't
eliminate quantifiers, so as to resolve the questions.
Second: When questions
about primes for which certain properties hold, we
should be able to measure those primes through properties of L-series
for which natural groups are attached.
So, the idea of the paper has a clear statement. If you
generalize elementary statements to Galois stratifications, you have
happy endings to both concerns. The method looks technical because it
deals with all possible elementary statements. Statements that had
variables quantified by both
∃(exists) and ∀(for all) had previously only ad hoc approaches.
Logicians often don't care about computing anything in such a general
context. To them this procedure doesn't matter if it is no better than a non recursive
quantifier elimination procedure. That is, they thought this procedure
couldn't really compute anything. That is not correct. By involving
group theory – the most successul non-linear equation attacker – it
revealed the nature of many problems, as I had shown many cases.
What Price to pay?:
Nothing brought more pain – in an otherwise merely unhappy career –
than the reception to this paper. One (I think of a Stony Brook
colleague, for example) would say it was just a technical result;
another would say schemes was the
wrong method (algebraic number theorists with no feeling for
geometry; Grothendieck haters – did you know they abounded then?); and
still a 3rd [group] would say you
can do it in an easier way (usually someone thinking of the
pure existential case).
The first direct positive statement about this successful way to get
the power of group theory into general diophantine equations occurred
many years later at Ihara's 60th Birthday Conference in Tokyo. It came
from Francois Loeser. (He and I were two of the six invited foreign
speakers, but I hadn't met him until the event of
the next paragraph.)
He had spoken on Motivic Measure
(a la Kontsevic), and never said anything about this particular paper.
Still, though I had not touched this topic for the previous 20 years, I
recognized its connection to my topics, and wondered if he knew also.
As my wife and I came down the stairs to the street car station after
the morning session of his talk, he looked our way and put up a book in
front of him. That was Fried-Jarden "Field
Arithmetic" opened to Chap. 25 [FJ]. Said he,
"How come no one knows anything about this?" It is common for
mathematicians to praise those who are already honored, even on topics
they don't know much about. This was essentially the opposite, as I now
explain.
Passing from Galois
Stratifications to Chow Motives: For some purposes (the category
of) Galois Stratifications isn't the right place to stop. Yes, they do
give zeta functions to statements with partially quantified
variables, through quantifier elimination. Yes, the coefficients
of the Poincare series are then equivalence classes of Galois
stratifications, to be specialized at the various primes. Still, there
are two objections.
- You don't clearly see how to produce canonical Galois Stratification
coefficients – choices are made.
- You don't automatically involve the great objects of arithmetic
geometry – non-singular projective varieties (satisfying the Weil
conjectures).
As I say in in the appendix to [Ha], I showed Jan
Denef (and Diane Meuser) [LGS], with its
generalization to p-adic
topics, in the early 80s (at a Bowdoin College conference). Especially
how to formulate the p-adic Chebotarev generalization
appropriate for that. Near the time we met in Japan, Loeser and Denef
had written a version of [DL]. This used Galois
stratifications but produced the Poincare series coefficients as
Covering space versions of Chow Motives.
That is, as equivalence classes of virtual algebraic projective
nonsingular varieties over Q,
with groups and attached projectors that still encoded the Galois
Stratification data from group theory. You specialize the coefficients
at a particular prime – as did Galois stratifications – by applying the
Frobenius at that prime to the coefficients.
Interpret applying the Frobenius to mean you are applying it to something in a category from ètale cohomology groups of varieties. This has a(t least one)
virtue and also a weakness by comparison with [LGS].
- Virtue: You could canonically (and prettily) attach (series of)
Beti numbers and Euler characteristics to general diophantine
statements, because the Chow motives reduced compatibly modulo all but
a finite number of exceptional primes.
- Weakness: You could say nothing about the particular primes that
were exceptions. The Chow motive production was rife with
famous complicated constructions producing denominators, including using resolution of
singularities in characteristic 0.
The subject today and bringing Chow
Motive questions down-to-earth: Nicaise [N] does the
relative version of [DL] using Voevodsky's relative Chow motives. While there
are Galois covers in [DL] and [N], Chow motives are usually not given
by covering data. This relation between using group theory and covering
data to motivic questions needs some examples that reflect on all Chow
motives, yet have simple properties understood from covers.
[EC] gives an example of such, and is intended for
readers to grasp the practical value of Chow Motives. It uses exceptional covers. For
example [EC, §8.2.2] asks what are the implications if the
Poincare series attached to a projective curve has a certain type of
Weil relation with the Poincare series of the projective line. The
converse of a known fact would be that there is a chain of exceptional correspondences from the curve to the projective line.
Exceptional covers (over finite fields) have a huge literature. This is
not a typical question for the people who work on them, but rather its
converse. The gist being that exceptional covers, a serious aspect of
cryptology [EC, §4.3], are
related to Chow motives in a big way.
REFERENCES
[DL] J. Denef, F. Loeser, Definable sets, motives and p-adic
integrals, J. Amer. Math. Soc. 14 (2001) 424–69.
[EC] M.D. Fried, The place of exceptional covers among all
diophantine relations, Finite Fields and their Apps. 11
(2005), 367–433.
[FJ] M.D. Fried, M. Jarden, Field Arithmetic, Ergebnisse der Mathematik III,
vol. 11, Springer, Heidelberg,
1986. New edition, 2004, ISBN 3-540-22811-x.
[Ha] T.C. Hales, What is Motivic Measure?, BAMS
42, 119–135. Hales kindly put my appendix – Historical Remarks on
Galois Stratification – in after he published the
paper.
[LGS] M. Fried, L-series on a Galois Stratification,
preprint 1979 (it was never texed).
[N] J. Nicaise, Relative motives
and the theory of pseudo-finite fields, International Mathematics Research Papers 2007, 2007:rpm001-69.