betaFeedback

WorksMathematics

Modular L-functions and Chebyshev systems

Abstract

We study real zeros of completed modular L-functions and the determination of cusp forms by their periods. Using Chebyshev systems and the Riemann-Roch theorem, we obtain bounds for real zeros, counted with multiplicity, and rank estimates for evaluation maps in terms of the dimensions of cusp-form spaces and the genera of modular curves. These estimates give criteria for finite families of periods to determine a cusp form and yield explicit dual bases in cases where the relevant quotient curves have genus zero. We also bound the dimensions of rational spans of eigenform periods.

Introduction and main results

Let N≥1N \ge1, let k≥2k \ge2 be even, and let f(z)=∑n≥1af(n)e2πinzf(z) = \sum_{n \ge1} a_f(n)e^{2\pi inz} belong to Sk(Γ0(N))S_k(\Gamma_0(N)). Define

L(f,s)=∑n≥1af(n)ns,ΛN(f,s)=Ns/2(2π)−sΓ(s)L(f,s).L(f,s)=\sum_{n\ge1}\frac{a_f(n)}{n^s},\qquad\Lambda_N(f,s)=N^{s/2}(2\pi)^{-s}\Gamma(s)L(f,s).

The Dirichlet series initially converges in a right half-plane, and ΛN(f,s)\Lambda_N(f,s) extends to an entire function. Its critical values are, up to explicit nonzero factors, the periods

rn(f)=∫0i∞f(z)zn dz=in+1N−(n+1)/2ΛN(f,n+1),0≤n≤k−2.(1)r_n(f)=\int_0^{i\infty}f(z)z^n\,dz=i^{n+1}N^{-(n+1)/2}\Lambda_N(f,n+1),\qquad0\le n\le k-2. \tag*{(1)}

At level one, these periods satisfy the Eichler–Shimura relations. A fundamental question in period theory is to identify which periods suffice to determine a cusp form. The even and odd parts of the period polynomial each determine the cusp form [14]. Using the Eichler–Shimura relations, Fukuhara [7 Theorem 2.2] constructed an explicit basis of the dual space consisting of odd periods, with further applications to Hecke operators. Ni et al. [23] used the same relations to construct explicit spanning families of periods and derive spanning results for Rankin–Cohen brackets.

Lei, Ni, and Xue [15, 16] studied period independence through products and Rankin–Cohen brackets of Eisenstein series and estimates for their Fourier coefficients. Their results cover up to three periods of the same parity, subject to dimension and index conditions, and any fixed number of such periods in sufficiently large weight. Xue [35] treated pairs of opposite parity under additional index restrictions. Refining Fourier coefficient estimates for the corresponding kernels, Ni and Xue [24] obtained the bound k≫n2k\gg n^2 for certain initial families of nn odd or even periods. The present paper treats period independence through real zeros of Mellin transforms, allowing both the evaluation parameters and the level to vary.

Chebyshev systems, the main analytic tool in this paper, are classical in approximation theory, interpolation, and moment problems. They generalize the zero-counting and interpolation properties of algebraic polynomials [2, 13]. They underlie uniqueness results for best uniform approximation, generalized Gaussian quadrature, and the geometry of moment spaces [13]. They also provide zero bounds for Abelian integrals, with applications to limit cycles bifurcating under perturbations of planar Hamiltonian systems [11]. The Mellin representation of ΛN(f,s)\Lambda_N(f,s) connects period independence with the zero-counting properties of Chebyshev systems. The Riemann–Roch theorem bounds the zeros of cusp forms along suitable paths on modular curves, and the sign-change principle transfers this control to their Mellin transforms. The resulting estimates apply to arbitrary real evaluation points in the stated intervals. Counting zeros with multiplicity also allows derivative evaluations to be treated by the same argument. These bounds hold in every admissible weight, with an optimal leading dimension term at fixed level, while the genus of the relevant modular curve controls the possible loss of interpolation rank. This gives a common geometric and analytic approach to real zeros, period independence, and interpolation.

For a nonzero analytic function GG, write ZI(G)Z_I(G) for its number of real zeros in II, counted with analytic multiplicity, including an endpoint if it belongs to II and GG is analytic near it. Write [x]+=max⁡{x,0}[x]_+ = \max\{x,0\} and let 1{P}\mathbf{1}_{\{P\}} be 1 when PP holds and 0 otherwise.

For a group Γ\Gamma and a character χ\chi, let g(Γ)g(\Gamma) be the genus of the compactified quotient curve XΓX_\Gamma, and let

dk(Γ,χ)=dim⁡CSk(Γ,χ).d_k(\Gamma,\chi)=\dim_{\mathbb{C}} S_k(\Gamma,\chi).

We omit χ\chi when it is trivial. For N≥1N \ge1, write

gN=g(X0(N)),dk(N)=dim⁡CSk(Γ0(N)).g_N=g(X_0(N)), \qquad d_k(N)=\dim_{\mathbb{C}} S_k(\Gamma_0(N)).

Dimensions of cusp-form spaces and ranks of their period maps are over C\mathbb{C} unless specified otherwise. We first state the results for Γ0(N)\Gamma_0(N).

Theorem 1.1. Let N≥1N \ge1 and let k≥2k \ge2 be even, with dk(N)>0d_k(N)>0. Then every nonzero f∈Sk(Γ0(N))f \in S_k(\Gamma_0(N)) satisfies

Z(−∞,k/2)(ΛN(f,⋅))≤dk(N)+gN−1−1{k=2}.Z_{(-\infty,k/2)}(\Lambda_N(f,\cdot)) \le d_k(N)+g_N-1-\mathbf{1}_{\{k=2\}}.

For any distinct real numbers s1,…,sq<k/2s_1,\ldots,s_q<k/2,

rank⁡(f↦(ΛN(f,s1),…,ΛN(f,sq)))≥min⁡{dk(N),[q−gN+1{k=2}]+}.\operatorname{rank}\left(f\mapsto\left(\Lambda_N(f,s_1),\ldots,\Lambda_N(f,s_q)\right)\right) \ge\min\left\{d_k(N),[q-g_N+\mathbf{1}_{\{k=2\}}]_+\right\}.

Thus a sufficient condition for injectivity of the evaluation map is q≥dk(N)+gN−1{k=2}q \ge d_k(N)+g_N-\mathbf{1}_{\{k=2\}}. For N>1N>1, both assertions hold on R\mathbb{R}.

At level one, the evaluations at any dk(1)d_k(1) distinct real points below k/2k/2 therefore form a dual basis. In particular, any set of critical periods rnr_n with n+1<k/2n+1<k/2 is linearly independent whenever its cardinality does not exceed dk(1)d_k(1), with no restriction on the parity of the selected indices.

The rank estimate can also be expressed in terms of matrices of completed LL-values. Let f1,…,fdk(N)f_1,\ldots,f_{d_k(N)} be a basis of Sk(Γ0(N))S_k(\Gamma_0(N)). For distinct real points s1,…,sqs_1,\ldots,s_q, with si<k/2s_i<k/2 when N=1N=1, a sufficient condition for the evaluation matrix (ΛN(fj,si))1≤i≤q,  1≤j≤dk(N)\left(\Lambda_N(f_j,s_i)\right)_{1\le i\le q,\;1\le j\le d_k(N)} to have full column rank is q≥dk(N)+gN−1{k=2}q\ge d_k(N)+g_N-\mathbf{1}_{\{k=2\}}. When gN=1{k=2}g_N=\mathbf{1}_{\{k=2\}}, every such square matrix is nonsingular.

The dimension term in Theorem 1.1 is necessary. Write Sk(Γ0(N))RS_k(\Gamma_0(N))_{\mathbb{R}} for the space of cusp forms with real Fourier coefficients. Any dk(N)−1d_k(N)-1 homogeneous real evaluation conditions have a nonzero solution in this space. Consequently, for N>1N>1 and even k≥4k\ge4 with nonzero cusp-form space,

dk(N)−1≤max⁡0≠f∈Sk(Γ0(N))RZR(ΛN(f,⋅))≤dk(N)+gN−1.(2)d_k(N)-1 \le\max_{0\ne f\in S_k(\Gamma_0(N))_{\mathbb{R}}} Z_{\mathbb{R}}(\Lambda_N(f,\cdot)) \le d_k(N)+g_N-1. \tag*{(2)}

The zeros in the lower-bound argument may be prescribed at any dk(N)−1d_k(N)-1 distinct real points. At fixed level, the gap in (2) is at most gNg_N, independently of the weight, so the leading dimension term is optimal. Corollary 2.6 further shows that in the genus-zero case, the upper bound can be attained by forms with rational Fourier coefficients, with simple zeros in prescribed disjoint intervals. The upper bound is also attained in positive-genus cases. In particular, Theorem 3.7 shows that this holds whenever X0(N)X_0(N) has genus one. For each fixed even weight k≥2k \ge2, Corollary 3.8 asserts that only finitely many levels N>1N > 1 with nonzero cusp-form space have the following property: the Mellin transform of every nonzero f∈Sk(Γ0(N))Rf \in S_k(\Gamma_0(N))_{\mathbb{R}} has at most dk(N)−1d_k(N)-1 real zeros, counted with multiplicity. The obstruction comes from the fixed points of the Fricke involution on X0(N)X_0(N): sufficiently many such points force some Mellin transforms to exceed this zero bound.

Our method also applies to other modular groups. Theorem 3.3 extends these estimates to Γ1(N)\Gamma_1(N), Γ(N)\Gamma(N), and full simultaneous Atkin–Lehner eigenspaces, with the genus of the corresponding quotient curve. Rank bounds for evaluations of ΛN(f,s)\Lambda_N(f,s) and its derivatives appear in Theorem 3.9, including the central derivatives allowed by the functional equation. For spaces with a fixed sign in the functional equation, these estimates also bound the dimension of the subspace of cusp forms whose LL-functions vanish to at least a prescribed order at s=k/2s=k/2. They determine this dimension exactly when the corresponding modular curve has genus zero (Corollary 3.10).

For N>1N>1, choose qq distinct critical indices in {0,…,k−2}\{0,\ldots,k-2\}. By Theorem 1.1 and the identification of critical values with periods in (1), a sufficient condition for the corresponding period functionals to span Sk(Γ0(N))∗S_k(\Gamma_0(N))^* is

q≥dk(N)+gN−1{k=2}.q \ge d_k(N)+g_N-1_{\{k=2\}}.

If gN=0g_N=0, the selected periods span Sk(Γ0(N))∗S_k(\Gamma_0(N))^* if and only if q≥dk(N)q \ge d_k(N). In particular, any dk(N)d_k(N) distinct critical periods form a dual basis when dk(N)≤k−1d_k(N)\le k-1. Fukuhara and Yang [8] give explicit period bases at level two. Choie, Park, and Zagier [4 Section 7] prove injectivity of the even and odd period maps at levels 2, 3, 5 in every even weight. Theorem 1.1 recovers these results and bounds the rank for arbitrary selections of critical periods, without a parity restriction.

There are only k−1k-1 critical periods on the path from 00 to ∞\infty, so they cannot span the dual when dk(N)>k−1d_k(N)>k-1. For each fixed even k≥2k\ge2, this obstruction occurs for all sufficiently large NN. We therefore allow the initial cusp to vary. For α∈Q\alpha\in\mathbb{Q}, define the period integrals

rn(f,α)=∫αi∞f(z)(z−α)n dz=in+1∫0∞f(α+iy)yn dy,0≤n≤k−2.(3)r_n(f,\alpha)=\int_{\alpha}^{i\infty} f(z)(z-\alpha)^n\,\mathrm{d}z =i^{n+1}\int_0^\infty f(\alpha+iy)y^n\,\mathrm{d}y,\qquad0\le n\le k-2. \tag*{(3)}

References

  1. [1]A. O. L. Atkin and J. Lehner, Hecke operators on Γ₀(m), Math. Ann. 185 (1970), 134–160.DOI
  2. [2]P. Borwein and T. Erdélyi, Polynomials and polynomial inequalities, Graduate Texts in Mathematics, vol. 161, Springer-Verlag, New York, 1995.DOI
  3. [3]F. Brunault and M. Neururer, Fourier expansions at cusps, Ramanujan J. 53 (2020), no. 2, 423–437.DOI
  4. [4]Y. Choie, Y. K. Park, and D. Zagier, Periods of modular forms on Γ₀(N) and products of Jacobi theta functions, J. Eur. Math. Soc. 21 (2019), no. 5, 1379–1410.
  5. [5]H. Cohen and F. Strömberg, Modular Forms: A Classical Approach, Graduate Studies in Mathematics, vol. 179, American Mathematical Society, Providence, RI, 2017.
  6. [6]M. Dickson and M. Neururer, Products of Eisenstein series and Fourier expansions of modular forms at cusps, J. Number Theory 188 (2018), 137–164.DOI
  7. [7]S. Fukuhara, Explicit formulas for Hecke operators on cusp forms, Dedekind symbols and period polynomials, J. Reine Angew. Math. 607 (2007), 163–216.DOI
  8. [8]S. Fukuhara and Y. Yang, Period polynomials and explicit formulas for Hecke operators on Γ₀(2), Math. Proc. Cambridge Philos. Soc. 146 (2009), no. 2, 321–350.DOI
  9. [9]A. Gasull, A. Geyer, and F. Mañosas, A Chebyshev criterion with applications, J. Differential Equations 269 (2020), 6641–6655.DOI
  10. [10]L. Gavrilov and I. D. Iliev, Two-dimensional Fuchsian systems and the Chebyshev property, J. Differential Equations 191 (2003), no. 1, 105–120.arxiv.org/abs/math/0111236
  11. [11]M. Grau, F. Mañosas, and J. Villadelprat, A Chebyshev criterion for Abelian integrals, Trans. Amer. Math. Soc. 363 (2011), no. 1, 109–129.
  12. [12]J. Huang, H. Liang, and X. Zhang, A new Chebyshev criterion and its application to planar differential systems, J. Differential Equations 344 (2023), 658–695.arxiv.org/abs/2011.02857
  13. [13]S. Karlin and W. J. Studden, Tchebycheff systems: With applications in analysis and statistics, Pure and Applied Mathematics, vol. XV, Interscience Publishers, New York, 1966.
  14. [14]W. Kohnen and D. Zagier, Modular forms with rational periods, in: Modular Forms (Durham, 1983), R. A. Rankin (ed.), Ellis Horwood, Chichester, 1984, pp. 197–249.
  15. [15]A. Lei, T. Ni, and H. Xue, Linear independence of even periods of modular forms, J. Number Theory 248 (2023), 120–139.DOI
  16. [16]A. Lei, T. Ni, and H. Xue, Linear independence of odd periods of modular forms, Res. Number Theory 9 (2023), no. 2, Paper No. 33.DOI
  17. [17]Ju. I. Manin, Parabolic points and zeta-functions of modular curves, Izv. Akad. Nauk SSSR Ser. Mat. 36 (1972), 19–66; English transl., Math. USSR-Izv. 6 (1972), 19–64.
  18. [18]Ju. I. Manin, Periods of parabolic forms and p-adic Hecke series, Math. USSR-Sb. 21 (1973), no. 3, 371–393.
  19. [19]K. Martin, Root number bias for newforms, Proc. Amer. Math. Soc. 151 (2023), no. 9, 3721–3736; corrected version, arXiv:2207.08121v3 (2025).DOI
  20. [20]L. Merel, Modular Fourier expansions of modular forms, in: On Artin’s Conjecture for Odd 2-dimensional Representations, Lecture Notes in Mathematics, vol. 1585, Springer, Berlin, 1994, pp. 59–94.DOI
  21. [21]L. Merel, Symboles de Manin et valeurs de fonctions L, in: Algebra, Arithmetic, and Geometry: In Honor of Yu. I. Manin, Vol. II, Progress in Mathematics, vol. 270, Birkhäuser, Boston, 2009, pp. 283–309.DOI
  22. [22]T. Ni, Uniform non-vanishing of prime quadratic twists of standard and symmetric square L-functions, arXiv:2610.11974 [math.NT], 2026.arxiv.org/abs/2610.11974
  23. [23]T. Ni, A. Song, Y. Su, H. Xue, and A. Yin, Explicit generators of the space of modular forms, Ramanujan J. 70 (2026), Paper No. 47.
  24. [24]T. Ni and H. Xue, Periods and symmetric square periods of modular forms, Manuscripta Math. 177 (2026), Paper No. 51.DOI
  25. [25]T. Ni and H. Xue, Twisted periods of modular forms, Nagoya Math. J. 261 (2026), Paper No. e18.arxiv.org/abs/2507.17041
  26. [26]D. D. No, and J. Torregrosa, On extended Chebyshev systems with positive accuracy, J. Math. Anal. Appl. 448 (2017), no. 1, 171–186.
  27. [27]A. P. Ogg, Hyperelliptic modular curves, Bull. Soc. Math. France 102 (1974), 449–462.DOI
  28. [28]K. Ono, L. Rolen, and F. Sprung, Zeta-polynomials for modular form periods, Adv. Math. 306 (2017), 328–343.arxiv.org/abs/1602.00752
  29. [29]V. Paşol and A. A. Popa, Modular forms and period polynomials, Proc. Lond. Math. Soc. (3) 107 (2013), no. 4, 713–743.
  30. [30]G. Shimura, On the periods of modular forms, Math. Ann. 229 (1977), no. 3, 211–221.
  31. [31]V. V. Shokurov, Shimura integrals of cusp forms, Math. USSR-Izv. 16 (1981), no. 3, 603–646.
  32. [32]A. Snowden, Real components of modular curves, arXiv:1108.3131 [math.NT], 2011.arxiv.org/abs/1108.3131
  33. [33]W. Stein, Modular forms, a computational approach, Graduate Studies in Mathematics, vol. 79, American Mathematical Society, Providence, RI, 2007.DOI
  34. [34]M. Watkins, Class numbers of imaginary quadratic fields, Math. Comp. 73 (2004), no. 246, 907–938.
  35. [35]H. Xue, Linear independence between odd and even periods of modular forms, New York J. Math. 30 (2024), 521–531.
  36. [36]Y. Zhang and Y. Zhou, Dimension formulas for modular form spaces with character for Fricke groups, Acta Arith. 206 (2022), 291–311.

Paper details

vlnc:W_brvt5uqh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC0-1.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents