Families of nearly ordinary Eisenstein series on unitary groups. With an appendix by Kai-Wen Lan (Q901781): Difference between revisions

In the author's own words, ``the motivation (of this paper, extracted from his thesis) is to provide a framework to generalize the important work of \textit{C. Skinner} and \textit{E. Urban} [Invent. Math. 195, No. 1, 1--277 (2014; Zbl 1301.11074)] on the Iwasawa main conjectures for \(\mathrm{GL}_2\) to forms on general unitary groups'' (from the introduction). Let \(p\) be an odd prime, \(\mathcal K\) a \(CM\) field in which \(p\) is totally split. Fix an isomorphism \(i :\mathbb C \cong \mathbb C_p\). For integers \(r \geq s \geq 0\), let \(U(r, s)\) be the unitary group associated to the skew-Hermitian matrix \[ \left(\begin{smallmatrix} && 1_s \\ & \zeta \\ -1_s \end{smallmatrix}\right), \] where \(\zeta\) is a diagonal matrix such that \(i^{-1}\zeta\) is positive definite. Given an irreducible cuspidal automorphic representation of \(U(r, s)\) which is nearly ordinary at all primes dividing \(p\), Eischen, Harris, Li and Skinner [``\(p\)-adic \(L\)-functions for Shimura varieties.'' (in preparation)] constructed the \(p\)-adic \(L\)-function which interpolates the algebraic part of critical values of the standard \(L\)-function of the representation twisted by general \(CM\) characters at far-from-center critical points. The present paper can be thought of as a continuation of their work. Its central task is, starting from a family of cusp forms on \(U(r, s)\) and a family of \(CM\) characters, and using a pullback formula of Shimura (actually due to \textit{P. B. Garrett} [Prog. Math. 46, 114--137 (1984; Zbl 0544.10023)]), to construct a \(p\)-adic family of Klingen Eisenstein series on \(U(r + 1, s + 1)\) and study the \(p\)-adic properties of its Fourier-Jacobi coefficients. More precisely, ``one tries to prove that the constant terms of the Klingen Eisenstein family are divisible by the standard \(p\)-adic \(L\)-function of the forms on \(U(r, s)\) and, therefore, the Eisenstein family is congruent to cuspidal families modulo this \(p\)-adic \(L\)-function. Passing to the Galois side, such congruences enable us to construct element in the Selmer groups, proving one divisibility of the corresponding Iwasawa main conjecture''. Note that in Eischen et al. [loc. cit.] the interpolation formulas are proved at all arithmetic points. Here, due to difficulties to describe in general the nearly ordinary Klingen Eisenstein sections, the author must introduce a ``generic'' condition (which puts restrictions on the ramification of the cusp forms at primes dividing \(p\)) and work with a Hida family of forms instead of a single form. Fortunately, since along a Hida family the set of ``generic'' forms is Zariski dense, the author's computations are enough to construct the whole Hida family of Klingen Eisenstein series (a similar phenomenon occurred in Skinner-Urban [loc. cit.]). The author's main results are too technical to even be summarized (their statement alone takes two whole pages), so we'll just settle for outlining the organization of the paper. Section 2 recalls various backgrounds for holomorphic automorphic forms on unitary groups, Eisenstein series and Fourier-Jacobi expansions. Section 3 collects straightforward generalizations of parts of Skinner-Urban [loc. cit.] on \(p\)-adic automorphic forms on unitary groups and Fourier-Jacobi expansions. Section 4 recalls the notion of Klingen and Siegel Eisenstein series, the pullback formulas relating them and their Fourier-Jacobi coefficients, and performs local calculations (this is the most technical part). In section 5 global calculations are made and the nearly ordinary Klingen Eisenstein series are constructed by pulling back Siegel Eisenstein series from a larger unitary group. An appendix by Kai-Wen Lan gives detailed proofs of some facts used for the \(p\)-adic \(q\)-expansion principle. Note that, using his results, the author has succeeded in proving one divisibility of the Iwasawa main conjecture for Hilbert modular forms and some kinds of Rankin-Selberg \(p\)-adic \(L\)-functions [Forum Math. Sigma 3, Article ID e18, 95 p. (2015; Zbl 1379.11089)], and \textit{C. Skinner} has been able to use this to prove a converse of a theorem of Gross, Zagier and Kolyvagin which states that, if the rank of the Selmer group of an elliptic curve is 1 and the \(p\)-part of the Shafarevich-Tate group is finite, then the Heegner point is nontorsion and the central \(L\)-value vanishes at order exactly 1 [``A converse to a theorem of Gross, Zagier, and Kolyvagin'', Preprint, \url{arXiv:1405.7294}].