study notes
number theory
We give two proofs of Baker's claim that for Baker claims that this follows "by symmetry". One proof was provided to me in a comment by user @reuns and relies on the Primitive Element Theorem and the Symmetric Function Theorem; we give a second more elementary proof based on Lagrange interpolation of , but it is less in the "by symmetry" spirit.
Both proofs can be found on math.stackexchange.
We denote by the Lebesgue measure on . Consider the set Then and thus To apply the refinement of Minkovski's theorem with the lattice and the symmetric convex set , we need that that is
Let . We recall (corollary of the primitive element theorem) that there are precisely ring (field) homomorphisms . We assume that the are real and that the remaining are complex with for .
Applying Minkovski's theorem requires
This is not quite the setup we have here. For several reasons:
Fixing these discrepancies is easy enough. I highlighted them merely to pin point precisely what made me uneasy.
The most important point, and which is quite independent from the other points, is the linear independence of the . This follows from the fact that defines a (symmetric) nondegenerate bilinear form on . Indeed, so that , .
Lemma. The bilinear pairing is nondegenerate.
Now if is some -basis of , the matrix is invertible since In conclusion,
Lemma. The field embeddings are -linearly independent.
When applying Minkovski we will implicitely apply it to and to the maps (we use the same notation for the -linear extensions of -linear maps). We also notice that simple line operations convert the matrix at the cost of multiplying the determinant by . Since the matrix on the right is invertible, we get that the matrix on the left is aswell. This is the ""-matrix in Theorem 12.2.
In this subsection we investigate why when applying Minkovski's theorem in Dirichlet's Unit Theorem, one requires rather than . The answer is, unsurprisingly, that relative to a natural volume form associated to the nondegenerate bilinear form .
Proposition. Let be a nondegenerate quadratic form on , a finite dimensional real vector space. Then there exists a unique translation invariant measure on such that for any -orthonormal basis
The is easy: is simply the Lebesgue measure on identified with by choosing a -orthonormal basis. This is independent on the choice of -orthonormal basis. Indeed, for any base change matrix , Therefore, if and are both -orthonormal bases, , and is well-defined.
Proposition. Let be a (full rank, discrete) lattice in ; let be a -basis of . Then setting , we have
Let be as above and let be a -orhtonormal basis of . We have where is the matrix of the vectors of relative to the basis . Also, if we define by , then and so . Since this imposes
Corollary. If is a number field, .
The real vector space inherits the nondegenerate bilinear form from . The statement above is w.r.t. the associated volume form. Recall that the discriminant of is defined as the determinant of any integral basis of . That is,
By construction, the kernel of the Log map where . Its kernel is comprised of those algebraic integers of all of whose "conjugates" lie on the unit circle.
Lemma. Let have all its conjugates of modulus , then is a root of unity, i.e. the kernel of 's Log map is the group of roots of unity in .
Clearly, every root of unity in is an algebraic integer and lies in the kernel of the Log map: ( is the group of all roots of unity in ). Conversely, if has all its conjugates of modulus , then its minimal polynomial lies in a finite set of integer polynomials by Kummer's argument. Therefore is a finite subgroup of . If is in the kernel, then so are all its powers, and thus is a root of unity. If , then if follows that is a subgroup of the group of -th roots of unity. Equality follows from cardinality.
Note. The lemma is wrong if one changes "" to "": the point and its (sole) conjugate have both modulus yet aren't roots of unity: if they were they would have to be integers in which they are not since their coordinates aren't integers (note ).
Note. The Log map clearly makes sense as a map where its defines a group homomorphism. For this note we only care about it along the subset .
This is simply the observation that for any unit , so that In other words, the more natural "full Log map" has its image in the hyperplane . The "missing modulus" is redundant information.
The beginning of the proof works without modification for "shorter Log maps" (i.e. by including fewer than of the ). The fact that isn't used until one invokes the proposition below.
Let be the family of units constructed in Dirichlet's Unit Theorem (DUT). By construction, the vectors , , are linearly independent in , and [1] is a free abelian group of rank within .
Proposition. The quotient group is finite.
Proof. This is the proposition that is used in the proof of the DUT. Its proof is basically that of corollary 12.1. Let :
The unit is constructed as in Baker's proof by taking a nearby vector to in the (discrete, full rank) lattice and setting Here "nearby" means w.r.t. the -norm on . We are not necessarily interested in the closest lattice point, merely in the fact that there is a uniform error bound on the coordinates . This is clear, for by construction, if was constructed by
then for :
The only noteworthy point is how one deduces the boundedness of the length of the "single missing conjugate" . We get it from the previously made observation: (not bothering with the factors one would expect on the first terms allows us not to have to differentiate between the case and .) This (trivial) computation is precisely where we use the fact that we used : had we chosen a value lower, we would not have been able to untangle the remaining moduli. Taking allows us to have a single modulus to dominate.
The remaining points are easy consequences of the boundedness of the conjugates.
Once Baker knows that the quotient is finite, the proof is nearly done. We put . Then [2], where we set . We note that includes the full rank sugroup of , hence has full rank itself. Using Lemma 11.3, there exists a basis of such that its matrix w.r.t. the basis is upper triangular. I don't believe the "upper triangular" property is important here, though.
If is a unit, then there exist such that therefore (indeed, raising the LHS to the -th power lands it in the kernel of the Log map; the kernel of the Log map is the set of roots of unity in ; the LHS is thus a root of unity in ) i.e. for some -th root of unity .
In conclusion, we get a split short exact sequence since its final term is free abelian: and so .
We use here the notation for to denote the subgroup of a group generated by the . This should not be confused with the ideal generated by the . Since ↩︎
In this subsection we write for the subgroup of some group generated by elements . Here is actually a lattice. There should be no possible confusion with the analogous notation for ideals. ↩︎