Papers
Topics
Authors
Recent
2000 character limit reached

Mixed Character Sum Bounds

Updated 7 January 2026
  • Mixed Character Sum Bounds are rigorous estimates for sums that combine multiplicative and additive characters over finite fields, crucial for revealing nontrivial cancellation.
  • Techniques such as subspace dissection, high-moment Hölder’s inequalities, and combinatorial geometry push beyond classical square-root barriers and Burgess limits.
  • These bounds have significant applications in analytic number theory, additive combinatorics, and arithmetic geometry, impacting subconvexity, pseudorandomness, and cryptographic analyses.

Mixed Character Sum Bounds are estimates for sums involving both multiplicative and additive characters over finite fields or residue rings, often evaluated at polynomial or rational arguments, and frequently with additional algebraic or combinatorial restrictions. This topic explores nontrivial cancellation in such sums, aiming to optimize upper bounds—typically below the square-root or Burgess barrier—by leveraging algebraic structure, combinatorial geometry, analytic techniques, and recent mean-value theorems. Applications span analytic number theory, additive combinatorics, and arithmetic geometry, impacting areas such as subconvexity, pseudorandomness, and exponential sum phenomena.

1. Foundational Definitions and Classical Bounds

Let Fq\mathbb{F}_q be a finite field of characteristic pp, r1r \geq 1, and Fqr\mathbb{F}_{q^r} its degree-rr extension. For a basis {v1,,vr}\{v_1,\dots,v_r\}, a restricted-coordinate subset Sr(A)S_r(A) is defined as

Sr(A)={x=a1v1++arvr:  aiA},S_r(A) = \left\{ x = a_1v_1 + \cdots + a_rv_r\,:\;a_i \in A \right\},

with AFqA\subseteq\mathbb{F}_q, #Sr(A)=(#A)r\#S_r(A) = (\#A)^r. Mixed character sums are formed as

Sr(A;χ,ψ;f1,f2)=xSr(A)χ(f1(x))ψ(f2(x)),S_r(A;\chi,\psi;f_1,f_2) = \sum_{x\in S_r(A)} \chi(f_1(x))\, \psi(f_2(x)),

where χ\chi is a multiplicative character (extended by χ(0)=0\chi(0)=0), ψ\psi an additive character, and f1f_1, f2f_2 rational functions.

Classically, the trivial upper bound is Sr(A;)(#A)r|S_r(A;\cdots)| \leq (\#A)^r, and Burgess-type results for incomplete sums (with additive phase) yield nontrivial savings only for sums of length >q1/4>q^{1/4} in one variable.

2. Breakthroughs in Mixed Character Sum Bounds

Iyer–Shparlinski (Restricted Coordinates and Cantor-like Sets)

For p=log#A/logqp = \log \#A / \log q and integer s1s \geq 1, Iyer–Shparlinski (Iyer et al., 2023) established

Sr(A;χ,ψ;f1,f2)(#A)rqrKs(p)S_r(A;\chi,\psi;f_1,f_2) \ll (\#A)^r\, q^{-r K_s(p)}

where

Ks(p)=sp(2p1)+(p1)4s(sp+1),K_s(p) = \frac{sp(2p-1)+(p-1)}{4s(sp+1)},

assuming nonprincipal χ\chi and f1f_1 in Qd1,nQ_{d_1,n}, or nonprincipal ψ\psi and f2f_2 in Rd2R_{d_2} (with rational function classes as above). Whenever Ks(p)>0K_s(p)>0 (p>1/2p > 1/2, large enough ss), this provides an exponential saving over the trivial bound.

Special cases include Cantor-like sets in F3r\mathbb{F}_{3^r} where A={0,2}A=\{0,2\}, yielding for s=5s=5: Sr(A;χ,ψ;f1,f2)2r30.99128rS_r(A;\chi,\psi;f_1,f_2) \ll 2^r 3^{-0.99128\,r} when p0.6309p\approx 0.6309 (Iyer et al., 2023). The method relies on subspace dissection, high-moment Hölder’s inequality, exceptional-shift counting, and subspace Weil bounds.

Sparse Coordinate Sums

Merai–Shparlinski–Winterhof (Mérai et al., 2022) analyze sums over vectors of prescribed Hamming weight: $S_{s,r}(\chi,\psi;f_1,f_2) = \sum_{\nu\in\Grs} \chi(f_1(\nu))\,\psi(f_2(\nu)),$ with $\Grs = \{ \nu:\mathrm{wt}(\nu)=s \}$, showing

Ss,r(χ,ψ;f1,f2)d1,d2,q(d1+max{d2,2})2s+1(rs)qr/2|S_{s,r}(\chi,\psi;f_1,f_2)| \ll_{d_1,d_2,q} (d_1+\max\{d_2,2\})\,2^{s+1}\binom{r}{s} q^{r/2}

for sufficiently non-degenerate f1,f2f_1, f_2. For q=2q=2, a combinatorial analysis yields bounds of the form

Ss,r(χ,ψ;f1,f2)2η(ρ)r+o(r)|S_{s,r}(\chi,\psi;f_1,f_2)| \leq 2^{\eta(\rho)r+o(r)}

with η(ρ)<H(ρ)\eta(\rho)<H(\rho) for ρ0.2145\rho\gtrsim 0.2145, thus breaking the entropy barrier for very sparse sums.

3. Multilinear and High-Dimensional Character Sums

Fouvry–Shparlinski–Xi (Fouvry et al., 2024) and Shkredov–Shparlinski (Shkredov et al., 2016) address trilinear and quadrilinear forms: T(α,β)=kKmMnNαmβk,nχ(ak+bmn)T(\alpha, \beta) = \sum_{k \le K} \sum_{m \le M} \sum_{n \le N} \alpha_m \beta_{k,n} \chi(a k + b m n)

Q(α,β)=kmnα,mβk,nχ(ak+bmn),Q(\alpha, \beta) = \sum_{k} \sum_\ell \sum_{m} \sum_n \alpha_{\ell,m} \beta_{k,n} \chi(a k\ell + b m n),

introducing bounds of the form

TM2NKΔ2po(1),QL2KNp2r2r2Δ3po(1)|T| \ll M^2 N K \Delta_2 p^{o(1)},\quad |Q| \ll L^2 K N p^{-\frac{2r}{2r-2}} \Delta_3 p^{o(1)}

with explicit nontriviality thresholds as low as K,M,Np1/8+εK, M, N \gg p^{1/8+\varepsilon}, significantly beating the naive p1/4p^{1/4} range and previous bilinear exponents. Key ingredients include repeated amplification/smoothing, application of high-moment Weil bounds, and precise incidence/energy lemmas.

4. Burgess Bounds for Mixed Sums in Multiple Variables

In higher dimensions, Pierce (Pierce, 2014), Kerr (Kerr, 2014), and Pierce–Xu (Pierce, 2020) utilize Vinogradov Mean Value Theorem (VMVT) and its multidimensional generalizations to achieve Burgess-type estimates

Sk(f;N,H)H11/rq(rMd,k+1)/(4r(rMd,k))qmax2rk/(4r(rMd,k))S_k(f;N,H) \ll \|H\|^{1-1/r} \|q\|^{(-r-M_{d,k}+1)/(4r(r-M_{d,k}))} q_{max}^{2rk/(4r(r-M_{d,k}))}

and, in the “minimal embedding” regime, stronger exponents for polynomials ff with “non-diagonal” monomial support. Nontrivial cancellation is achieved as soon as Hiqi1/k+ϵH_i \gg q_i^{1/k+\epsilon}, breaking the $1/4$ barrier for multi-variable sums.

Pierce–Xu (Pierce, 2020) further demonstrate that for nn variables,

SHn(n+1)/(2r)q[n(ΘM)+1]/[4r(ΘM)]+ϵ|S| \ll H^{n - (n+1)/(2r)} q^{[n(\Theta-M)+1]/[4r(\Theta-M)] + \epsilon}

with thresholds Hq1/21/(2(n+1))+κH \geq q^{1/2-1/(2(n+1)) + \kappa}, leveraging the algebraic stratification results of Xu and the main VMVT for translation-dilation invariant systems.

5. Advanced Geometric and Combinatorial Approaches

Recent bounds exploit additive combinatorics, small-doubling phenomena, and geometric incidence results. Schoen–Shkredov (Schoen et al., 2020) show that if AFpA \subset \mathbb{F}_p and A+A<KA|A+A| < K|A|, then for a large set BB,

S(A,B;χ)=aA,bBχ(a+b)<exp(c(ε04logplog2K)1/3)AB|S(A,B;\chi)| = \left| \sum_{a\in A,\,b\in B}\chi(a+b) \right| < \exp\left(-c \left(\frac{\varepsilon_0^4 \log p}{\log^2K}\right)^{1/3}\right)|A||B|

yielding super-polynomial savings and applications to sum-product expansions (Balog-type theorems), surpassing both Burgess and classical Pólya–Vinogradov bounds.

6. Applications and Further Directions

Mixed character sum bounds underpin deep questions in analytic number theory, e.g., equidistribution, subconvexity for LL-functions, randomness properties for cryptographic settings, and incidence geometry over finite fields. They resolve longstanding thresholds for incomplete sums, support combinatorial expansion hypotheses, and offer sharp tools for analyzing algebraic constructions such as Cantor sets, unions of intervals, and sparse coordinate sets in finite fields.

Open problems include optimally lowering multilinear exponent thresholds (e.g., pushing $1/8$ further down), understanding Paley graph phenomena in arbitrary sets, refining composite modulus analogues, and advancing geometric incidence bounds to improve combinatorial character sum estimates.


Table: Summary of Main Mixed Character-Sum Results

Paper/Authors Main Bound Form Nontriviality Thresholds
Iyer–Shparlinski (Iyer et al., 2023) (#A)rqrKs(p)(\#A)^r\,q^{-rK_s(p)} p>1/2p > 1/2, ss large; Cantor sets (p0.63p \approx 0.63)
Fouvry–Shparlinski–Xi (Fouvry et al., 2024) M2NKΔ2po(1)M^2 N K \Delta_2 p^{o(1)} K,M,Np1/8+ϵK,M,N \gg p^{1/8+\epsilon}
Shkredov–Shparlinski (Shkredov et al., 2016) (BCD)3/4A1/2q1/2+(BCD)^{3/4}A^{1/2}q^{1/2} + \cdots Nq2/5+ϵN \geq q^{2/5+\epsilon}
Kerr (Kerr, 2014), Pierce (Pierce, 2014) B11/rq|B|^{1-1/r}q^{\cdots} Hi>qi1/k+ϵH_i > q_i^{1/k+\epsilon}
Schoen–Shkredov (Schoen et al., 2020) exp(c(logp)1/3/(logK)2/3)AB\exp(-c(\log p)^{1/3}/(\log K)^{2/3})|A||B| Small doubling KK, A,B|A|,|B| large

7. Relation to Lower Bounds and Structural Obstructions

Goldmakher–Lamzouri (Goldmakher et al., 2011) unconditionally realize large character sums for odd-order characters via pretentious distance, matching (up to ϵ\epsilon) conditional GRH bounds. The extremal behavior of mixed character sums is controlled by arithmetic structure, with Paley-type phenomena for quadratic and higher odd orders.

Wang–Xu (Wang et al., 2024) show that the average size over nonprincipal characters for mixed sums with irrational phase is x\sim \sqrt{x}, while Harper establishes a strict decay o(x)o(\sqrt{x}) for rational phases—a dichotomy driven by Diophantine properties and the arithmetic distribution of summands.


Mixed character sum bounds now constitute a core toolkit for additive number theory and finite field arithmetic, with ongoing advances driven by mean-value theorems, stratification ideas, and higher-moment analytic techniques. Progress in this arena continues to deepen the connections between combinatorial geometry, algebra, and analytic number theory.

Whiteboard

Topic to Video (Beta)

Follow Topic

Get notified by email when new papers are published related to Mixed Character Sum Bounds.