A class number formula for higher derivatives of abelian L-functions (Q4463564)

The paper under review is devoted to a conjecture of Burns on an abelian extension \(K/k\) of global fields. This new conjecture is a common generalization of two conjectures in algebraic number theory that are due to Rubin and Gross, respectively. Roughly speaking, the point is that the object appearing in Rubin's conjecture (i.e., a ``higher Stark unit'') should satisfy a congruence condition in the spirit of Gross's conjecture. This paper gives evidence in favor of Burns's conjecture for quadratic extensions and verifies it (up to a factor 2 on both sides) for real abelian extensions of \(\mathbb Q\). Moreover, it interprets two further conjectures, which are due to Darmon and Gross, as base-changes of the conjecture of Burns by a quadratic extension. Though the precise statement of Burns's conjecture already needs a lot of work and these interesting results are obtained by quite involved methods, the paper is written in a surprisingly clear and understandable way so the reviewer finds it to be a very useful text for studying the topic. To be more specific, Dirichlet's analytic class number formula for a number field \(k\) gives the leading term at \(s=0\) of the zeta function of \(k\) by means of the product of the class number and the regulator of \(k\). If we consider an abelian extension \(K/k\) of global fields with Galois group \(G\), we can define (character by character) the equivariant \(S\)-truncated, \(T\)-modified \(L\)-function \(\Theta_{K/k,S,T}\) with values in the group ring \(\mathbb C[G]\); here \(S\) and \(T\) are finite disjoint sets of places of \(k\) such that \(S\) contains all infinite primes and all primes that ramify in \(K/k\) and also a subset \(S_1\subsetneq S\) consisting of primes that split completely in \(K/k\). Then \(\Theta_{K/k,S,T}\) vanishes at \(s=0\) at least at order \(r=\#S_1\). Let \(U_{K,S,T}\) be the group of all \(S\)-units of \(K\) that are congruent to \(1\) modulo every prime of \(K\) above a prime in \(T\). We suppose that \(U_{K,S,T}\) is torsion-free. Let \(\eta\) be the ``higher Stark unit'', i.e. the element of the tensor product of \(\mathbb C\) and the \(r\)-th exterior power of \(U_{K,S,T}\) which is sent to the \(r\)th term of Taylor expansion of \(\Theta_{K/k,S,T}\) at \(s=0\) by a suitable logarithmic mapping. Rubin's conjecture predicts some integral properties of \(\eta\), namely for any ``testing homomorphism'' \(\Phi\) in the \(r\)-th exterior power of \(\text{Hom}_{\mathbb Z[G]}(U_{K,S,T},\mathbb Z[G])\) Rubin's conjecture states that \(\Phi(\eta)\in\mathbb Z[G]\). The conjecture of Burns goes further: it predicts that, modulo the \(((\#S)-r)\)-th power of the augmentation ideal \(I\) in the group ring \(\mathbb Z[G]\), \(\Phi(\eta)\) is congruent to \(\pm h_{k,S,T}\text{Reg}_G^\Phi\), where \(h_{k,S,T}\) is the \((S,T)\)-class number of \(k\) and \(\text{Reg}_G^\Phi\) is a group ring-valued regulator in the spirit of Gross built up by means of values of \(\Phi\) and of local reciprocity maps (for completions modulo primes of \(K\) above primes in \(S-S_1\)) on a \(\mathbb Z\)-basis of \(U_{k,S,T}\). As \(\text{Reg}_G^\Phi\in I^{(\#S)-r-1}\) by definition, this conjecture implies that \(\Phi(\eta)\in I^{(\#S)-r-1}\). Let us mention that a method for specifying the sign in the congruence is given in a forthcoming paper of Burns.