Holmstrom 5 Memo notes Weibel

Memo notes from Weibel in K-theory handbook

Problem: Compute K-theory of number fields and their rings of integers. Have known for long that K-groups of ROIs are f.g. with known ranks. Ref: Borel: Cohomologie reelle stable…, and LNM 341. Recent work by Suslin, Voevodsky, Rost etc have given us (nearly complete???) torsion information.

Thm: On the odd K-groups.

Discussion: K-groups of \mathbb{Z}; Vandiver’s conjecture is equivalent to K 4m()=0K_{4m}(\mathbb{Z}) = 0 for m2m \geq 2. It is known that these groups have odd order, with no prime factor smaller than 10 710^7.

Will also survey K-groups of local fields and global function fields.

Section 2: K nK_n for n3n \leq 3, this material is classical - the groups have presentations by generators and relations.

Section 3: For odd groups, have cyclic summand which is understood through a construction by Harris and Segal: a variant of Adams e-invariant.

Section 4: Canonical free summands related to units (Borel). Other (almost periodic) summands related to Pic(R)Pic(R) and Br(R)Br(R), detected by etale Chern classes (Soule).

Section 5: Global function fields. In this case, there is a smooth projective curve XX whose higher K-groups are finite, and are related to the action of FrobFrob on Jac(X)Jac(X). The orders of these groups are related to the values of the zeta function of XX at negative integers.

Section 6: Local fields. For an extension of p\mathbb{Q}_p, we understand the pp-completion, but not the actual K-groups.

Section 7: Odd torsion, in the case of a number field. Also 2-primary torsion in totally imaginary number fields. This uses the Voevodsky-Rost theorem.

Section 8: 2-primary torsion in real number fields. Uses a theorem of Voevodsky.

Section 9: Odd torsion in K 2i()K_{2i}(\mathbb{Z}): Irregular primes, Vandiver (Soule).

Key technical tool: Motivic spectral sequence. Details. Established by Levine over a Dedekind domain. With finite coeffs, motivic cohomology groups are etale cohomology (for a field). For a Dedekind domain, have a similar but more complicated result for the E 2E_2 terms. Pic, role of etale cohomological dimension.

Also a spectral sequence from terms based mainly on etale cohomology (does he mean \ell-adic? to K pq(X;/ )K_{-p-q}(X; \mathbb{Z}/ \ell^{\infty} ) (what does he mean by this?).

Section 2:

Class groups. Bass-Milnor-Serre (1967) on K 2K_2. Localization sequence involving tame symbol/tame kernel. Relation to etale cohomology (Tate, Merkurjev, Suslin). Under some hyp, have mod mm short exact sequence with PicPic, K 2K_2 , BrBr.

Quillen and Borel: Homological techniques for the rank of higher K-groups.

Classical results for K 3K_3.

Section 3: The e-invariant

Def and details. “Harris-Segal map, Harris-Segal summand”.

It is an isomorphism for finite fields.

Def: Exceptional field.

Example: Relation to the Bott map.

Remark: Exlains homotopy-theoretic motivation. Talks about the K-theory space of a symmetric monoidal category.

Remark: Bernoulli number, relation to the Riemann zeta function and to the J-homomorphism.

Example: On the image of the natural map π n sK n()\pi_n^s \to K_n(\mathbb{Z}) (this must mean the n-th stable homotopy group of spheres, because he later talks about the image of JJ in this group).

Birch-Tate conjecture: Let FF be a number field. Then it is known that the ζ F\zeta_F has a pole of order r 2r_2 at s=1s = -1. B-T conjecture says, that for totally real number fields (r 2=0r_2 = 0), we have

ζ F(1)=(1) r 1|K 2(O F)|/w 2(F) \zeta_F(-1) = (-1)^{r_1} |K_2(O_F)| / w_2(F)

The odd part of this conjecture was proven by Wiles, using Tate’s Theorem 4. The 2-primary part would follow from the 2-adic Main Conjecture of Iwasawa theory. Hence the full Birch-Tate conjecture is known for abelian extensions of \mathbb{Q}.

Section 4: Etale Chern classes

Want to relate etale cohomology and K-groups. In one direction, can use etale Chern classes (ref: Soule: K-theories des anneaux d’entiers…) in the form given in Dwyer-Friedlander: Algebraic and etale K-theory (1985).

In this section, we describe maps in the other direction: Kahn maps, or anti-Chern classes. These are an efficient reorganization of the maps above. There are also higher Kahn maps, omitted here. Refs to Kahn: Deux theoremes de comparaison en cohomologie etale: applications; On the Lichtenbaum-Quillen conjecture; K-theory of semi-local rings with finite coeffs and etale cohomology.

If FF is a field in which \ell is invertible, there is a canonical map

K 2i1(F;/ v)H et 1(F,μ v i K_{2i-1}(F; \mathbb{Z} / \ell^v) \to H^1_{et}(F, \mu_{\ell^v}^{\otimes i}

called the first etale Chern class. It is equal to a map to etale K-theory composed with an edge map in the Atiyah-Hirzebruch spectral sequence for etale K-theory. For i=1i=1 it is the Kummer isomorphism.

For each ii and vv, can construct a splitting of the first etale Chern class, at least of \ell is odd. This is the Kahn map. Definition. It is compatible with the Harris-Segal map (diagram). Theorem on injectivity. Compatibility with change of coeffs, so get maps

H et 1(F, (i))K 2i1(F; ) H^1_{et}(F, \mathbb{Z}_{\ell}(i)) \to K_{2i-1}(F; \mathbb{Z}_{\ell})

and

H et 1(F,/ (i))K 2i1(F;/ ) H^1_{et}(F, \mathbb{Z} / {\ell}^{\infty}(i)) \to K_{2i-1}(F; \mathbb{Z} / {\ell}^{\infty})

Corollary: For ring of SS-integers.

We also have the second etale Chern class

K 2i(F;/ v)H et 2(F,μ v i+1 K_{2i}(F; \mathbb{Z} / \ell^v) \to H^2_{et}(F, \mu_{\ell^v}^{\otimes i+1}

and a Kahn map in the other direction. For i=1i=1, this is just Tate’s map, which is an isomorphism for all fields by Merkurjev-Suslin.

Theorems similar to the above (compatibility etc).

Many details omitted.

Example: The Main Conjecture of Iwasawa theory implies that for odd \ell, the order of the finite group H et 2([1/], (2k))istheH^2_{et}( \mathbb{Z}[1/ \ell], \mathbb{Z}_{\ell}(2k)) is the \ellprimarypartofthenumeratorof-primary part of the numerator of \zeta(1-2k).ByEulersformula,thisisalsothe. By Euler's formula, this is also the \ellprimarypartofthenumeratorof-primary part of the numerator of B_k / 2k`.

Theorem: For every number field FF, and all ii, the Adams operation ψ k\psi^k acts on K 2i1(F)K_{2i-1}(F) \otimes \mathbb{Q} as multiplication by k ik^i.

Proof: This follows from an isomorphism with an etale cohomology group, which commutes with the Adams op.

Section 5: Function fields

A global field of characteristic p>0p>0 is a finitely generated field FF of transcendence degree 1 over F p\mathbf{F}_p. The algebraic closure of F p\mathbf{F}_p in FF is a finite field F q\mathbf{F}_q of char pp. There is a unique smooth projective curve XX over F q\mathbf{F}_q whose function field is FF. If SS is a nonempty set of closed points of XX, then XSX-S is affine, and its coordinate ring is called the ring of SS-integers in FF. Will discuss K-groups of XX, of FF, and of the SS-integers.

Thm: For n1n \geq 1, the groups K n(X)K_n(X) are finite of order prime to pp.

Proof uses localization sequence, Weil reciprocity, and more…

More details on the above groups.

K-groups of X¯\bar{X}, with action of geometric Frob. Explicit description of these Galois modules, proof uses Suslin’s theorem saying that for i1i \geq 1 and p\ell \neq p, we have

H M n(X¯,/ (i))=H et n(X¯,/ (i)) H^n_M(\bar{X}, \mathbb{Z} / \ell^{\infty} (i)) = H^n_{et}(\bar{X}, \mathbb{Z} / \ell^{\infty} (i))

Thm describing K n(X)=K n(X¯) GK_n(X) = K_n(\bar{X})^G.

The zeta function of a curve. Definition: ζ X(s)=Z(X,q s)\zeta_X(s) = Z(X, q^{-s}) where

Z(X,t)=exp( n=1 |X(F q n|t nn) Z(X,t) = exp \big( \sum_{n=1}^{\infty} | X( \mathbf{F}_{q^n} | \cdot \frac{t^n}{n} \big)

Weil’s result for curves, explanation in terms of action of algebraic Frob on \ell-adic cohomology.

Formula for |ζ X(1i)|| \zeta_X(1-i) | in terms of \ell-adic cohomology, or equivalently in terms of K-groups.

Iwasawa modules: alternative interpretation of some etale cohomology groups. Ref to Dwyer-Mitchell: On the K-theory spectrum of a ring of algebraic integers, and to Mitchell in the Kth handboook for the number field case.

Section 6: Local fields

Details omitted. Mention of Leopoldt’s conjecture, in the following form (possibly valid only for totally real fields): The torsion free part p d1\mathbb{Z}_p^{d-1} of 𝒪 F × p\mathcal{O}_F^{\times} \otimes \mathbb{Z}_p injects into the torsion free part of p d\mathbb{Z}_p^{d} of 𝒪 E j ×\prod \mathcal{O}_{E_j}^{\times} (the completions of the number field FF above pp).

Section 7: Number fields at primes where cd=2cd=2

Here we obtain a cohomological description of the odd torsion in the K-groups of a number field, and also the 2-primary torsion in the K-groups of a totally imaginary number field. These are the cases where cd (𝒪 S)=2cd_{\ell}(\mathcal{O}_S) = 2, something which forces the motivic spectral sequence to degenerate completely.

Details omitted.

Theorem of Wiles describing all the odd special values of the zeta function of a totally real field.

Section 8: Real number fields at the prime 2

Infinite cohomological dimension: this complicates descent methods. Also other problems with this case.

Details omitted.

Section 9: The odd torsion in K *(Z)K_*(Z)

Some details omitted.

Vandiver’s conjecture in terms of the class group; the Herbrand-Ribet thm.

Vandiver’s conjecture for \ell is equivalent to non-existence of \ell-torsion in K 4k()K_{4k}(\mathbb{Z}) for all kk.

nLab page on 5 Memo notes Weibel

Created on June 9, 2014 at 21:16:13 by Andreas Holmström