scientific article; zbMATH DE number 15986
\(L\)-functions of symmetric powers of elliptic curvesBeilinson's conjectureBloch-Beilinson regulatorde Rham cohomologyKronecker seriesmotivic cohomology groups
Generalizations (algebraic spaces, stacks) (14A20) Applications of methods of algebraic (K)-theory in algebraic geometry (14C35) (Co)homology theory in algebraic geometry (14F99) Zeta functions and related questions in algebraic geometry (e.g., Birch-Swinnerton-Dyer conjecture) (14G10) Elliptic curves (14H52)
The study of Beilinson's conjecture in a more general context started with the work of Bloch and was particularly influenced by numerical studies of elliptic curves over \(\mathbb{Q}\) [cf. \textit{S. Bloch} and \textit{D. Grayson} in Applications of algebraic \(K\)-theory to algebraic geometry and number theory, Proc. AMS-ISM-SIAM Joint Summer Res. Conf., Boulder 1983, Part I, Contemp. Math. 55, 79-88 (1986; Zbl 0629.14002)]. Deninger gave an explicit version of Beilinson's conjecture with respect to the values \(L(\text{Sym}^ kE,k+1)\), for \(k\geq 1\) [cf. \textit{C. Deninger}, in Sémin. Théorie Nombres, Paris 1986/87, Prog. Math. 75, 111-128 (1988; Zbl 0701.14008) and Invent. Math. 96, No. 1, 1-69 (1989; Zbl 0721.14004)]. In the paper under review the authors verify numerically the relations conjectured by Beilinson and Deninger in the case \(k=2\), for elliptic curves for which this statement is non-trivial (the number of curves among these admitting a rational torsion point is finite). Let \(E\) be an elliptic curve over \(\mathbb{Q}\) and \({\mathcal E}\) its minimal regular model over \(\mathbb{Z}\). Given a non-zero differential \(\omega\) in \(E\) defined over \(\mathbb{R}\), let \(\omega_ 1=\left|\int_{E(\mathbb{R})^ 0}\omega\right|\) be the real period of \((E,\omega)\) and choose \(\omega_ 2\) such that the lattice \(\Lambda=\omega_ 1\mathbb{Z}+\omega_ 2\mathbb{Z}\subset\mathbb{C}\) gives an isomorphism between \((\mathbb{C}/\Lambda,dz)\) and \((E,\omega)\) over \(\mathbb{C}\) and such that \(\tau=\omega_ 2/\omega_ 1\) satisfies \(y=\Im(\tau)>0\). Let \(A(\Lambda)=\frac{\omega_ 1^ 2y}{\pi}\) and \(\langle\gamma,x\rangle=\exp(\frac{\gamma\overline{x}- x\overline{\gamma}}{A(\Lambda)})\). Given a divisor \({\mathfrak a}=\sum_{x\in\mathbb{C}/\Lambda}a_ x(x)\) over \(E(\mathbb{C})\), an integer \(\nu\geq 0\) and \(s\in\mathbb{C}\), with \(\Re(s)>(\nu/2)+1\), the double Kronecker series is defined by \[ K_ \nu(0,{\mathfrak a},s,\Lambda)= \sum_{x\in\mathbb{C}/\Lambda}a_ x \mathop{{\sum}'}_{\gamma\in\Lambda} \langle\gamma,x\rangle \frac{{\overline{\gamma}}^{\nu}} {(\gamma\overline{\gamma})^ s}. \] This series can be extended analytically to \(\mathbb{C}\). Given an integer \(d>1\) let \(\mathbb{Q}[E_ d]^ 0\) be the \(\mathbb{Q}\)-vector space of divisors of \(E\) defined over \(\mathbb{Q}\) of degree zero whose support is contained in the points of \(d\)-torsion. For a smooth quasi-projective variety \(X\) defined over \(\mathbb{Q}\) the motivic cohomology groups are given as follows: given \(j\geq 0\) and \(i\leq 2j\), let \(H^ i_{{\mathcal M}}(X,\mathbb{Q}(j))=\text{grad}^ j_ \omega(K_{2j-i}(X)\otimes\mathbb{Q})\), where the graduate ring is taken with respect to the filtration \(\omega\) of the Quillen \(K\)-group tensored by \(\mathbb{Q}\) [cf. \textit{A. A. Beilinson}, J. Sov. Math. 30, 2036-2070 (1985); translation from Itogi Nauki Tekh., Ser. Sovrem. Probl. Mat. 24, 181-238 (1984; Zbl 0588.14013) and \textit{P. Schneider} in: Beilinson's Conjectures on Special Values of \(L\)-functions, Meet. Oberwolfach 1986, Perspect. Math. 4, 1-35 (1988; Zbl 0673.14007)]. It is known that the image of \(K_{2j-i}(\mathcal X)\otimes\mathbb{Q}\) in \(H^ i_{\mathcal M}(X,\mathbb{Q}(j))\) does not depend on the regular proper model \(\mathcal X\) of \(X\) over \(\mathbb{Z}\), if such a model exists. This image is denoted by \(H^ i_{\mathcal M}(X,\mathbb{Q}(j))_ \mathbb{Z}\). We have the following exact sequence \[ H^ 2_{\mathcal M}(E,\mathbb{Q}(2))_ \mathbb{Z}\hookrightarrow H^ 2(E,\mathbb{Q}(2))\xrightarrow{\partial=\coprod\partial_ p}\coprod_ p H^ 3_{{\mathcal M},{\mathcal E}(p)}(E,\mathbb{Q}(2))_ \mathbb{Z}\to H^ 3_{\mathcal M}(E,\mathbb{Q}(2))_ \mathbb{Z}. \] If \(L(E,s)\) denotes the Hasse-Weil \(L\)- function of \(E\), Beilinson conjectures that: \[ \dim_ \mathbb{Q} H^ 2_{\mathcal M}(E,\mathbb{Q}(2))_ \mathbb{Z}=\text{ord}_{s=0}L(E,s)=1.\tag{1} \] \[ H^ 3_{\mathcal M}(E,\mathbb{Q}(2))_ \mathbb{Z}=0.\tag{2} \] \textit{U. Jannsen} [``Mixed motives and algebraic \(K\)-theory, Lect. Notes Math. 1400 (1990; Zbl 0691.14001); \S12] showed that the ``motivic homology \(H^ 3_{{\mathcal M},{\mathcal E}(p)}(E,\mathbb{Q}(2))_ \mathbb{Z}=K_ 1^ \prime({\mathcal E}(p))\otimes_ \mathbb{Z}\mathbb{Q}\) is non-zero if and only if the local factor \(L_ p(E,s)^{-1}\) in \(p\) of \(L(E,s)=\prod_ p L_ p(E,s)\) is zero at \(s=0\), which is equivalent to \(E\) having bad reduction of multiplicative type at \(p\). In this case the motivic homology has dimension one. This implies the following conjecture: \[ \dim_ \mathbb{Q} H^ 2_{\mathcal M}(E,\mathbb{Q}(2))=\text{ord}_{s=0}L_ S(E,s)=1+\#\{p\mid L_ p(E,0) ^{-1}=0\},\tag{3} \] where \(S\) denotes the set of primes of bad reduction for \(E\) and \(L_ S=\prod_{p\notin S}L_ p\). This is a particular case of the first part of an \(S\)-conjecture of Beilinson [cf. \textit{D. Ramakrishnan}, Contemp. Math. 83, 183-310 (1989; Zbl 0694.14002)]. Let \(r:H^ 2_{\mathcal M}(E,\mathbb{Q}(2))\to H^ 2_{dR}(E_{/\mathbb{R}},\mathbb{R}(2))=H^ 1(E(\mathbb{C}),\mathbb{R}(1))^{\overline{F_ \infty}}\) be the Bloch-Beilinson regulator, where \(\overline{F_ \infty}\) denotes the involution corresponding to the real structure of \(H^ 1_{dR}(E/\mathbb{R})\otimes\mathbb{R}(1)\), \(H^ i_{dR}\) denotes the de Rham cohomology and \(\mathbb{R}(1)=2\pi i\mathbb{R}\). Bloch conjectured that \(r\) is injective [cf. \textit{S. Bloch} and \textit{D. Grayson}, in Contemp. Math. 55, 79-88 (1986; loc. cit.)]. According to the general context of Beilinson's conjecture one is reduced at least to conjecturing the injectivity of \(r\) in \(H^ 2_{\mathcal M}(E,\mathbb{Q}(2))_ \mathbb{Z}\). One consequence of this fact is that there exists a unique \(\mathbb{Q}\)-linear map \(\xi_ d:\mathbb{Q}[E_ d]^ 0\to H^ 2_{\mathcal M}(E,\mathbb{Q}(2))\) such that for every divisor \({\mathfrak a}=\sum_{dx=0}a_ x(x)\) we have \([r(\xi({\mathfrak a})),\omega]=A(\Lambda)^ 2K_ 1(0,{\mathfrak a},2,\Lambda)\), where \([\alpha,\beta]=\frac{1}{2\pi i}\int_{E(\mathbb{C})}\alpha\wedge\beta\), for any 1-forms \(\alpha\) and \(\beta\) in \(E\). It follows from the functional equation of \(L(E,s)\) and elementary properties of Kronecker series that, if \(\partial(\xi({\mathfrak a}))=0\), then \(y^ 2K_ 1(0,{\mathfrak a},2)\in L(E,2)\mathbb{Q}\). Denote by \({\mathcal M}_ \mathbb{Q}\) the category of motifs over \(\mathbb{Q}\). In this category one considers the motif \(\text{Sym}^ k E\) and the corresponding motivic cohomology groups \(H^ i_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(j))\). Its integral part is defined as the image of \(H^ i(\tilde{\mathcal E}^ k,\mathbb{Q}(j))\) in the former group and is denoted as before by \(H^ i(\text{Sym}^ k E,\mathbb{Q}(j))_ \mathbb{Z}\), where \(\tilde{\mathcal E}\) is the normalization of the \(k\)-th power of \({\mathcal E}\) over \(\mathbb{Z}\). We have the following three consequences of Beilinson's conjecture, which are analogues of the previous conjectures: \[ \dim_ \mathbb{Q} H^{k+1}_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(k+1))_ \mathbb{Z}=\text{ord}_{s=0}L(\text{Sym}^ k E,s)=[\frac{k}{2}]+1.\tag{4} \] \[ H^{k+2}_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(k+1))_ \mathbb{Z}=0.\tag{5} \] \[ \dim_ \mathbb{Q} H^{k+1}_{\mathcal M} (\text{Sym}^ k E,\mathbb{Q}(k+1)) =\left[\frac{k+2}{2}\right]+\#\{p\mid L_ p(\text{Sym}^ k E,0)^{- 1}=0\}.\tag{6} \] As before we have a unique map \(\xi:\mathbb{Q}[E_ d]^ 0\to H^{k+1}_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(k+1))\), which coincides with the previous map \(\xi\) in the case \(k=1\) [cf. \textit{C. Deninger}, Invent. Math. 96, No. 1, 1-69 (1989; loc. cit.)]. In this case Beilinson's regulator is given by \[ r:H^{k+1}_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(k+1))\to H^{k+1}_{dR}(\text{Sym}^ k E_{/\mathbb{R}},\mathbb{R}(k+1)). \] If \(\omega,\eta\) is a basis of \(H^ 1_{dR}(E_{/\mathbb{Q}})\) then a basis of \(H^ 1_{dR}(\text{Sym}^ k E,\mathbb{R}(1))\) is given by the projection by \(\mathbb{C}\to\mathbb{R}(k)\) of the differentials \(\omega^{(k)},\omega^{(k- 1)}\eta,\dots,\omega^{(k- [\frac{k}{2}])}\eta^{([\frac{k}{2}])}\) as an \(\mathbb{R}\)-basis [\textit{P. Deligne}, in Automorphic forms, representations and \(L\)-functions, Proc. Symp. Am. Math. Soc., Corvallis 1977, Proc. Symp. Pure Math. 33, 313-346 (1979; Zbl 0449.10022)]. The projection of these differentials is denoted by a tilde. Once again it is conjectured that given two \(k\)-forms \(\alpha\) and \(\beta\) in \(E^ k\), let \([\alpha,\beta]=\frac{1}{(2\pi i)^ k}\int_{E^ k(\mathbb{C})}\alpha\wedge\beta\), then \(\xi\) is uniquely characterized by \([r(\xi({\mathfrak a})),\tilde\omega^{(k- j)}\tilde\eta^{(j)}]=A(\Lambda)^{k- j+1}K_{k-2j}(0,{\mathfrak a},k- j+1)\), where \({\mathfrak a}=\sum_{dx=0}a_ x(x)\in\mathbb{Q}[E_ d]^ 0\), for \(j=0,\dots,[\frac{k}{2}]\). Beilinson's conjecture in this case allows one to compute the period \(c^ +(\text{Sym}^ k E(k))\) as follows: \[ c^ +(\text{Sym}^ k E(k))^{- 1}\text{det}([r(\xi({\mathfrak a}_ i)),\tilde{\omega}^{(k-j)}\tilde{\eta}^{(j)}]_{i,j}\in L^{([\frac{k}{2}]+1)}(\text{Sym}^ k E,0)\mathbb{Q}, \] where \({\mathfrak a}_ 0,\dots,{\mathfrak a}_{[\frac{k}{2}]}\in\mathbb{Q}[E_ d]^ 0\) are such that their images by \(\xi\) lie in \(H^{k+1}_{\mathcal M}(\text{Sym}^ k E,\mathbb{Q}(2))_ \mathbb{Z}\). The authors develop a numerical study of the following conjecture, first in the case where the elliptic curve \(E\) has potentially good reduction, and then in the case where it has semi-stable reduction. \textit{Conjecture}: Let \({\mathfrak a},{\mathfrak b}\in\mathbb{Q}[E_ d]^ 0\) such that their images lie in \(H^ 3_{\mathcal M}(\text{Sym}^ 2E,\mathbb{Q}(3))_ \mathbb{Z}\), \textit{i.e.} \(\partial(\xi({\mathfrak a}))=\partial(\xi({\mathfrak b}))=0\). Then up to a factor \(\varrho\in\mathbb{Q}\) we have the following identity \[ \frac{y^ 4}{{\pi}^ 2}\text{det}\begin{pmatrix} K_ 2(0,{\mathfrak a},3) & K_ 0(0,{\mathfrak a},2)\\ K_ 2(0,{\mathfrak b},3) & K_ 0(0,{\mathfrak b},2)\end{pmatrix}=\varrho C^{3/2} L(\text{Sym}^ 2E,3), \] where \(C\) is the conductor of the \(\ell\)-adic system of representations \(H_ \ell(\text{Sym}^ 2 E)\). For the entire collection see [Zbl 0749.14025].
- Algorithmic Number Theory
- p-adic Eisenstein-Kronecker series for CM elliptic curves and the Kronecker limit formulas
- L-functions of symmetric powers of Kloosterman sums (unit root L-functions and p-adic estimates)
- On the coefficients of symmetric power L-functions
- Special values of motivic \(L\)-functions and zeta-polynomials for symmetric powers of elliptic curves
- scientific article; zbMATH DE number 691440
- Symmetric power L-functions for families of generalized Kloosterman sums
- \(A_{\infty}\)-algebras associated with elliptic curves and Eisenstein-Kronecker series
- On symmetric powers of -recurrent sequences and deformations of Eisenstein series
- On \(L\)-functions of elliptic curves and cyclotomic towers
- Higher regulators and Hecke \(L\)-series of imaginary quadratic fields. I
- scientific article; zbMATH DE number 4023445 (Why is no real title available?)
- ON A NONCRITICAL SYMMETRIC SQUARE -VALUE OF THE CONGRUENT NUMBER ELLIPTIC CURVES
- scientific article; zbMATH DE number 4187961 (Why is no real title available?)
- Explicit version of Beilinson's theorem for the modular curve \(X_{1}(N)\)
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3972633)
