betaFeedback

WorksMathematics

A lower bound for the Bohr radius of the bidisc

Abstract

We prove that the Bohr radius K2K_2 of the bidisc, whose exact value is unknown, satisfies K2>0.301665K_2>0.301665, improving Patel's bound K2>0.30158758K_2>0.30158758. As in his argument, we pass to functions with positive real part and value 11 at the origin; we bound the sum of the moduli of their coefficients of degree nn by 2Sn2S_n instead of 2n2\sqrt n, where SnS_n is the Sidon constant of {0,1,…,n}\{0,1,\dots,n\}, and we bound S5S_5, S6S_6 and S7S_7 with computer-assisted certificates. In Lean 4 we also prove K2≥0.301665K_2\ge0.301665, assuming Patel's inequality coupling the coefficients of degrees two and three.

Introduction

For d≥1d \ge1 let Dd={z∈Cd:∣z1∣,…,∣zd∣<1}\mathbb{D}^d = \{z \in\mathbb{C}^d : |z_1|,\ldots,|z_d| < 1\} be the unit polydisc and let the Schur class S(Dd)\mathcal{S}(\mathbb{D}^d) be the set of holomorphic maps f:Dd→Df:\mathbb{D}^d \to\mathbb{D}. For f(z)=∑αcαzαf(z) = \sum_{\alpha} c_{\alpha}z^{\alpha} and r>0r > 0 write ∥fr∥1=∑α∣cα∣r∣α∣\|f_r\|_1 = \sum_{\alpha}|c_{\alpha}|r^{|\alpha|}, where ∣α∣=α1+⋯+αd|\alpha| = \alpha_1 + \cdots+ \alpha_d. The Bohr radius of Dd\mathbb{D}^d is

Kd=sup⁡{r>0:∥fr∥1≤1 for all f∈S(Dd)}.K_d = \sup\{r > 0 : \|f_r\|_1 \le1 \text{ for all } f \in\mathcal{S}(\mathbb{D}^d)\}.

Bohr [9] proved that K1≥1/6K_1 \ge1/6. The sharp value K1=1/3K_1 = 1/3 was found independently by Riesz, Schur and Wiener, and Bohr’s paper includes Wiener’s proof (see [8]). Boas and Khavinson [8] introduced KdK_d for d≥2d \ge2, proved 1/(3d)<Kd<2(log⁡d)/d1/(3\sqrt{d}) < K_d < 2\sqrt{(\log d)/d} and asked for the exact value of KdK_d when d>1d > 1. Defant, Frerick, Ortega-Cerdà, Ounaïes and Seip [11] showed that KdK_d has order (log⁡d)/d\sqrt{(\log d)/d}, and Bayart, Pellegrino and Seoane-Sepúlveda [6] proved Kd∼(log⁡d)/dK_d \sim\sqrt{(\log d)/d} as d→∞d \to\infty. No exact value of KdK_d is known for any d≥2d \ge2.

For the bidisc, functions of one variable show K2≤1/3K_2 \le1/3. Knese Knese [14, Corollary 1.2] proved K2≥0.3006K_2 \ge0.3006: by his Theorem 1.1 and Andô’s inequality, K2K_2 is at least the root 0.3006282…0.3006282\ldots of ∑k≥1k rk=1/2\sum_{k\ge1}\sqrt{k}\,r^k = 1/2. Baran, Pikul, Woerdeman and Wojtylak Baran, Pikul, Woerdeman and Wojtylak [5, Theorem 6.4] proved K2<0.3177K_2 < 0.3177, and Griego Griego [12] lowered this to K2<0.3174541K_2 < 0.3174541.

In an unrefereed solution to a problem on MathDB, Patel Patel [20] proved K2>0.30158758K_2 > 0.30158758. He also reports the upper bound K2<0.302825279492K_2 < 0.302825279492, from a rational inner function and an exact computation that his reproducibility package [21] carries out; we have not checked this bound, but if it holds, it and Theorem 1.1 confine K2K_2 to an interval of width about 1.2⋅10−31.2 \cdot10^{-3}.

Sidon constants enter through the coefficients. For a finite set Λ⊂Z\Lambda\subset\mathbb{Z} the Sidon constant S(Λ)S(\Lambda) is the least CC with

∑λ∈Λ∣aλ∣≤Cmax⁡∣w∣=1∣∑λ∈Λaλwλ∣(aλ∈C),\sum_{\lambda\in\Lambda}|a_\lambda| \le C \max_{|w|=1}\left|\sum_{\lambda\in\Lambda}a_\lambda w^\lambda\right| \qquad(a_\lambda\in\mathbb{C}),

and we write

Sk=S({0,1,…,k})(k≥1).S_k = S(\{0,1,\ldots,k\}) \qquad(k \ge1).

Bernal-González, Cabana, García, Maestre, Muñoz-Fernández and Seoane-Sepúlveda Bernal-González, Cabana, García, Maestre, Muñoz-Fernández and Seoane-Sepúlveda [7, Theorem 3.3] showed that KdK_d is at least the root of ∑k≥1S(k,d) rk=1/2\sum_{k\ge1} S(k,d)\,r^k = 1/2, where S(k,d)S(k,d) is the Sidon constant of the kk-homogeneous polynomials in dd variables. For d=2d=2, S(k,2)=SkS(k,2)=S_k (Remark 2.4). Newman’s bound Sk≤kS_k\le\sqrt{k}, whose proof appears in Shapiro’s thesis [23 Chapter 3] (see also [19, (1)]), turns this equation into Knese’s, and Shapiro [23] showed that $S_k=\sqrt{k}$ if and only if $k\in\{1,2,4\}$. Few values of $S_k$ are known; {xref}Table 1` lists them with their sources.

We give Patel’s argument for the lower bound in full, confirm the exact rational computation behind it with our own programs, and improve the bound.

Theorem 1.1. The Bohr radius of the bidisc satisfies K2>0.301665K_2>0.301665.

The gain over Patel’s bound, about 8⋅10−58\cdot10^{-5}, comes from the terms of degrees five, six and seven.

Patel reduces the problem to the coefficients of holomorphic functions HH on D2\mathbb{D}^2 with Re⁡H>0\operatorname{Re}H>0 and H(0)=1H(0)=1 (§ 2). Carathéodory’s inequality on slices bounds sums of moduli of the coefficients of degree two and, with the help of a positive semidefinite Toeplitz form, of degree three, and a third inequality (Lemma 3.4) couples the two degrees; Patel combines the three into a bound for the coefficients of degrees two and three together (§ 3). The third inequality comes from the unitary realization of Schur functions on the bidisc [1, 4], which rests on Andô’s inequality [3]. In the degrees n≥4n\ge4 Patel uses the bound 2n2\sqrt{n} for the sum of the absolute values of the coefficients of degree nn of HH, which he derives from Knese’s estimate [14 Lemma 2.1].

Our improvement starts from a simple observation: Carathéodory’s inequality on slices also bounds this sum by 2Sn2S_n (Lemma 2.3). By Newman’s bound and Shapiro’s theorem, Sn<nS_n<\sqrt{n} for every n≥5n\ge5, while S4=2S_4=2. We prove S5≤2.187S_5\le2.187, S6≤2.363S_6\le2.363 and S7≤2.552S_7\le2.552 with computer-assisted certificates built from positive definite Toeplitz matrices (Section 4); for S5S_5 our bound is slightly above the value S5=2.186529771892569…S_5=2.186529771892569\ldots that Ye et al. [24] determined. In the degrees n≥8n\ge8 we keep Newman’s bound Sn≤nS_n\le\sqrt{n}, since the size of the certificates grows quickly with nn (Table 2). With rational upper bounds for n\sqrt{n} and an explicit tail, an exact rational computation shows that Patel’s majorant, with these bounds, is less than 11 at r=0.301665r=0.301665, and by continuity also at slightly larger rr, which proves Theorem 1.1 (§ 5). § 6 describes the computer checks and a formal proof in Lean, conditional on Lemma 3.4, of the non-strict bound K2≥0.301665K_2\ge0.301665, which needs only the computation at r=0.301665r=0.301665.

k=5k=5k=6k=6k=7k=7
bound bkb_{k}2.1872.3632.552
initial grid848^{4}858^{5}2⋅452\cdot4^{5}
boxes8,91261,118210,960
distinct vertex labels269,7694,333,58546,344,804
smallest margin bk2−v∗T(y)−1vb_{k}^{2}-v^{*}T(y)^{-1}v1.49⋅10−51.49\cdot10^{-5}8.48⋅10−88.48\cdot10^{-8}8.88⋅10−88.88\cdot10^{-8}

Table 2. The certificates of Proposition 4.2.

The Carathéodory normalization and slices

Let the Carathéodory class P2\mathcal{P}_2 be the set of holomorphic functions HH on D2\mathbb{D}^2 with Re⁡H>0\operatorname{Re}H>0 and H(0)=1H(0)=1. We write

H(z)=1+∑n≥1Hn(z),Hn(z)=∑j+k=nhjkz1jz2k.H(z)=1+\sum_{n\ge1}H_n(z),\qquad H_n(z)=\sum_{j+k=n}h_{jk}z_1^jz_2^k.

and set ∥Hn∥1=∑j+k=n∣hjk∣\lVert H_n\rVert_1=\sum_{j+k=n}|h_{jk}| and Mr(H)=∑n≥1∥Hn∥1rnM_r(H)=\sum_{n\ge1}\lVert H_n\rVert_1r^n.

We use the following reduction.

Lemma 2.1. Let 0<r<10<r<1. If Mr(H)≤1M_r(H)\le1 for every H∈P2H\in\mathcal{P}_2, then K2≥rK_2\ge r.

Proof. Constant functions satisfy ∥fr∥1≤1\lVert f_r\rVert_1\le1. Let f∈S(D2)f\in\mathcal{S}(\mathbb{D}^2) be nonconstant. By the maximum principle ∣f∣<1|f|<1 on D2\mathbb{D}^2. Multiplying ff by a unimodular constant does not change ∥fr∥1\lVert f_r\rVert_1, so we may assume that t=f(0)t=f(0) lies in [0,1)[0,1). Then H=(1−f)/(1−t)H=(1-f)/(1-t) lies in P2\mathcal{P}_2. Its coefficients of positive degree are those of ff multiplied by −1/(1−t)-1/(1-t). Hence ∥fr∥1=t+(1−t)Mr(H)≤1\lVert f_r\rVert_1=t+(1-t)M_r(H)\le1. □

Remark 2.2. The normalization loses nothing: if r<K2r<K_2, then Mr(H)≤1M_r(H)\le1 for every H∈P2H\in\mathcal{P}_2. Indeed, for 0<s<10<s<1 the function z↦H(sz)z\mapsto H(sz) maps D2\mathbb{D}^2 into a compact subset of the right half-plane, hence into a disc {w:∣w−ρ∣<ρ}\{w:|w-\rho|<\rho\} with ρ≥1\rho\ge1. Then f(z)=1−H(sz)/ρf(z)=1-H(sz)/\rho lies in S(D2)\mathcal{S}(\mathbb{D}^2) with f(0)=1−1/ρf(0)=1-1/\rho, and ∥fr∥1≤1\lVert f_r\rVert_1\le1 becomes ∑n∥Hn∥1snrn≤1\sum_n\lVert H_n\rVert_1s^nr^n\le1. Let s→1s\to1. Thus K2=sup⁡{r:Mr(H)≤1 for all H∈P2}K_2=\sup\{r:M_r(H)\le1\text{ for all }H\in\mathcal{P}_2\}.

Lemma 2.3. Let H∈P2H \in\mathcal{P}_2 and n≥1n \ge1. Then ∣Hn(ζ)∣≤2|H_n(\zeta)| \le2 for ζ∈T2\zeta\in\mathbb{T}^2, and ∥Hn∥1≤2Sn\lVert H_n\rVert_1 \le2S_n.

Proof. Fix ζ∈T2\zeta\in\mathbb{T}^2. The function φ(λ)=H(λζ)=1+∑m≥1Hm(ζ)λm\varphi(\lambda)=H(\lambda\zeta)=1+\sum_{m\ge1}H_m(\zeta)\lambda^m is holomorphic on D\mathbb{D} with Re⁡φ>0\operatorname{Re}\varphi>0. For 0<ρ<10<\rho<1,

Hn(ζ)ρn=1π∫02πRe⁡φ(ρeiθ)e−inθ dθ.H_n(\zeta)\rho^n=\frac{1}{\pi}\int_0^{2\pi}\operatorname{Re}\varphi(\rho e^{i\theta})e^{-in\theta}\,d\theta.

so ∣Hn(ζ)∣ρn≤2Re⁡φ(0)=2|H_n(\zeta)|\rho^n\le2\operatorname{Re}\varphi(0)=2. Letting ρ→1\rho\to1, we obtain Carathéodory’s inequality ∣Hn(ζ)∣≤2|H_n(\zeta)|\le2. Now write Hn(ζ1,ζ2)=ζ2nqn(ζ1/ζ2)H_n(\zeta_1,\zeta_2)=\zeta_2^n q_n(\zeta_1/\zeta_2) with qn(w)=∑j=0nhj,n−jwjq_n(w)=\sum_{j=0}^n h_{j,n-j}w^j. As ζ\zeta runs over T2\mathbb{T}^2, the quotient ζ1/ζ2\zeta_1/\zeta_2 runs over T\mathbb{T} and ∣Hn(ζ)∣=∣qn(ζ1/ζ2)∣|H_n(\zeta)|=|q_n(\zeta_1/\zeta_2)|. Hence max⁡T∣qn∣≤2\max_{\mathbb{T}}|q_n|\le2, and by the definition of SnS_n,

∥Hn∥1=∑j=0n∣hj,n−j∣≤Snmax⁡T∣qn∣≤2Sn.\lVert H_n\rVert_1=\sum_{j=0}^n|h_{j,n-j}|\le S_n\max_{\mathbb{T}}|q_n|\le2S_n.

Remark 2.4. The same identification shows that S(k,2)=SkS(k,2)=S_k: on T2\mathbb{T}^2 a kk-homogeneous polynomial in two variables is, up to the factor ζ2k\zeta_2^k, a polynomial in ζ1/ζ2\zeta_1/\zeta_2 of degree at most kk. With this identification, the next corollary is the case d=2d=2 of [7 Theorem 3.3].

Corollary 2.5. If 0<r<10<r<1 and ∑k≥1Skrk≤1/2\sum_{k\ge1}S_kr^k\le1/2, then K2≥rK_2\ge r.

Proof. By Lemma 2.3, Mr(H)≤2∑k≥1Skrk≤1M_r(H)\le2\sum_{k\ge1}S_kr^k\le1 for every H∈P2H\in\mathcal{P}_2. Apply Lemma 2.1.

In [7] the corresponding estimate is proved for ff itself. There Wiener’s inequality takes the place of Carathéodory’s: the Taylor coefficients g^(k)\widehat{g}(k) of a holomorphic g:D→D‾g:\mathbb{D}\to\overline{\mathbb{D}} satisfy ∣g^(k)∣≤1−∣g(0)∣2|\widehat{g}(k)|\le1-|g(0)|^2 for k≥1k\ge1. Patel derives the bound ∥Hn∥1≤2n\lVert H_n\rVert_1\le2\sqrt{n} for all H∈P2H\in\mathcal{P}_2 from Knese’s estimate [14 Lemma 2.1], by a limiting argument with (1−ϵH)/(1+ϵH)(1-\epsilon H)/(1+\epsilon H) [20]. Lemma 2.3 with Newman’s bound gives it directly.

With the values of SkS_k for k≤5k\le5 in Table 1 and Sk≤kS_k\le\sqrt{k} for k≥6k\ge6, Corollary 2.5 gives K2≥0.301339K_2\ge0.301339. If the value of Ye et al. for S5S_5 and Newman’s bounds for S6S_6 and S7S_7 are replaced by the bounds of Proposition 4.2, it gives K2≥0.301370K_2\ge0.301370. Patel’s coupling of degrees two and three goes beyond this.

kkSkS_ksource
11elementary
22\sqrt{2}Newman (see [18])
35/35/3Yu et al. [25]; lower bound Neuwirth [19]
42Newman’s bound; equality by Shapiro [23]
52.186529771892569 . . .; ≤2.187\le2.187Ye et al. [24]; Proposition 4.2
6≤2.363\le2.363Proposition 4.2
7≤2.552\le2.552Proposition 4.2
k≥8k \ge8≤k\le\sqrt{k}Newman [23 Chapter 3]

Table 1. Sidon constants Sk=S({0,1,…,k})S_k = S(\{0,1,\ldots,k\}).

The coefficients of degrees two and three

Throughout this section H∈P2H\in\mathcal{P}_2, and

A2=∣h20∣+∣h02∣,B2=∣h11∣,A3=∣h30∣+∣h03∣,B3=∣h21∣+∣h12∣,A_2=|h_{20}|+|h_{02}|,\qquad B_2=|h_{11}|,\qquad A_3=|h_{30}|+|h_{03}|,\qquad B_3=|h_{21}|+|h_{12}|,

so that ∥H2∥1=A2+B2\lVert H_2\rVert_1=A_2+B_2 and ∥H3∥1=A3+B3\lVert H_3\rVert_1=A_3+B_3. Inner products are linear in the first argument, ⟨x,y⟩=y∗x\langle x,y\rangle=y^*x.

Lemma 3.1 (Patel [20]). A22+B22≤4A_2^2+B_2^2\le4.

Proof. Choose ζ0∈T\zeta_0\in\mathbb{T} such that h20ζ02h_{20}\zeta_0^2 and h02h_{02} have the same argument, and put a=h20ζ02+h02a=h_{20}\zeta_0^2+h_{02} and b=h11ζ0b=h_{11}\zeta_0. Then ∣a∣=A2|a|=A_2, ∣b∣=B2|b|=B_2 and H2(±ζ0,1)=a±bH_2(\pm\zeta_0,1)=a\pm b. By Lemma 2.3, ∣a±b∣≤2|a\pm b|\le2, and therefore ∣a∣2+∣b∣2=12(∣a+b∣2+∣a−b∣2)≤4|a|^2+|b|^2=\frac{1}{2}(|a+b|^2+|a-b|^2)\le4.

Lemma 3.2. Let RR be a positive semidefinite Hermitian Toeplitz matrix of size k+1k+1, and let q(ζ)=∑j=0kcjζjq(\zeta)=\sum_{j=0}^k c_j\zeta^j. Then c∗Rc≤R00max⁡T∣q∣2c^*Rc\le R_{00}\max_{\mathbb{T}}|q|^2, where c=(c0,…,ck)Tc=(c_0,\ldots,c_k)^{\mathsf{T}}.

Proof. Let e0,…,eke_0,\ldots,e_k be vectors in Ck+1\mathbb{C}^{k+1} with ⟨el,ej⟩=Rjl\langle e_l,e_j\rangle=R_{jl}, for example the columns of R1/2R^{1/2}. Since RR is Toeplitz, the tuples (e0,…,ek−1)(e_0,\ldots,e_{k-1}) and (e1,…,ek)(e_1,\ldots,e_k) have the same Gram matrix. Hence ej↦ej+1e_j\mapsto e_{j+1} (0≤j<k)(0\le j<k) extends to a well-defined linear isometry from the span of e0,…,ek−1e_0,\ldots,e_{k-1}

onto the span of e1,…,eke_1,\ldots,e_k, and then to a unitary operator Ω\Omega on the span of all the eje_j. Then ej=Ωje0e_j=\Omega^j e_0, so ∑jcjej=q(Ω)e0\sum_j c_j e_j=q(\Omega)e_0. As Ω\Omega is unitary, ∥q(Ω)∥≤max⁡T∣q∣\lVert q(\Omega)\rVert\leq\max_{\mathbb{T}}|q|, and

c∗Rc=∥∑jcjej∥2=∥q(Ω)e0∥2≤max⁡T∣q∣2∥e0∥2=R00max⁡T∣q∣2.c^*Rc=\left\lVert\sum_j c_j e_j\right\rVert^2=\lVert q(\Omega)e_0\rVert^2\leq\max_{\mathbb{T}}\lvert q\rvert^2\lVert e_0\rVert^2=R_{00}\max_{\mathbb{T}}\lvert q\rvert^2.

$\xrefLemma 3.2 is a consequence of the Carathéodory–Toeplitz theorem. Ye et al. prove it in the same way [24 Lemma 3.1].

Lemma 3.3 (Patel [20]). Let p(ζ)=∑j=03cjζjp(\zeta)=\sum_{j=0}^{3}c_j\zeta^j satisfy max⁡T∣p∣≤1\max_{\mathbb{T}}\lvert p\rvert\leq1. Then

35(∣c0∣+∣c3∣)2+28(∣c1∣+∣c2∣)2≤44.35\left(\lvert c_0\rvert+\lvert c_3\rvert\right)^2+28\left(\lvert c_1\rvert+\lvert c_2\rvert\right)^2\leq44.

Consequently 35A32+28B32≤17635A_3^2+28B_3^2\leq176.

Proof. Replacing p(ζ)p(\zeta) by p(eiθζ)p(e^{i\theta}\zeta) for a suitable θ\theta, we may assume that c1c_1 and c2c_2 have the same argument. Choose ξ∈T\xi\in\mathbb{T} with ∣ξ‾c0+c3∣=∣c0∣+∣c3∣\lvert\overline{\xi}c_0+c_3\rvert=\lvert c_0\rvert+\lvert c_3\rvert, and let

u=(3,4,−4,−3)T,v=(0,1,1,0)T,w=(ξ,0,0,1)T,R=uu∗+28vv∗+35ww∗.u=(3,4,-4,-3)^{\mathsf{T}},\qquad v=(0,1,1,0)^{\mathsf{T}},\qquad w=(\xi,0,0,1)^{\mathsf{T}},\qquad R=uu^*+28vv^*+35ww^*.

The matrix RR is positive semidefinite. Since ∣ξ∣=1\lvert\xi\rvert=1, it is also Toeplitz, with first row (44,12,−12,−9+35ξ)(44,12,-12,-9+35\xi). By \xref`Lemma 3.2`, withc=(c_0,\ldots,c_3)^{\mathsf{T}}$,

∣u∗c∣2+28∣v∗c∣2+35∣w∗c∣2=c∗Rc≤44.\lvert u^*c\rvert^2+28\lvert v^*c\rvert^2+35\lvert w^*c\rvert^2=c^*Rc\leq44.

Here ∣v∗c∣=∣c1+c2∣=∣c1∣+∣c2∣\lvert v^*c\rvert=\lvert c_1+c_2\rvert=\lvert c_1\rvert+\lvert c_2\rvert and ∣w∗c∣=∣ξ‾c0+c3∣=∣c0∣+∣c3∣\lvert w^*c\rvert=\lvert\overline{\xi}c_0+c_3\rvert=\lvert c_0\rvert+\lvert c_3\rvert, which proves the first claim. For the second, apply it to p=q3/2p=q_{3/2} from the proof of \xref`Lemma 2.3`. Then\lvert c_0\rvert+\lvert c_3\rvert=A_3/2$ and ∣c1∣+∣c2∣=B3/2\lvert c_1\rvert+\lvert c_2\rvert=B_3/2.

The proof above extends to a one-parameter family. For −1≤t≤2-1\leq t\leq2 and ut=(t,1,−1,−t)Tu_t=(t,1,-1,-t)^{\mathsf{T}}, the matrix utut∗+(t+1)vv∗+(2+t−t2)ww∗u_tu_t^*+(t+1)vv^*+(2+t-t^2)ww^* is positive semidefinite and Toeplitz, with first row (2+t,t,−t,−t2+(2+t−t2)ξ)(2+t,t,-t,-t^2+(2+t-t^2)\xi), so the same argument shows

(2+t−t2)(∣c0∣+∣c3∣)2+(t+1)(∣c1∣+∣c2∣)2≤2+t.(2+t-t^2)(\lvert c_0\rvert+\lvert c_3\rvert)^2+(t+1)(\lvert c_1\rvert+\lvert c_2\rvert)^2\leq2+t.

$\xrefLemma 3.3 is the case t=3/4t=3/4. The case t=1/2t=1/2 is the inequality 9(∣c0∣+∣c3∣)2+6(∣c1∣+∣c2∣)2≤109(\lvert c_0\rvert+\lvert c_3\rvert)^2+6(\lvert c_1\rvert+\lvert c_2\rvert)^2\leq10, from which Yu et al. obtain S3≤5/3S_3\leq5/3 by the Cauchy–Schwarz inequality [25 (6)]. On its own, $\xrefLemma 3.3 gives only ∥H3∥1≤396/35≈3.364\lVert H_3\rVert_1\leq\sqrt{396/35}\approx3.364, more than 2S3=10/32S_3=10/3. It is used together with the next inequality, which bounds A3A_3 in terms of B2B_2.

Lemma 3.4 (Patel [20]). A3+12B22≤2A_3+\frac{1}{2}B_2^2\leq2.

Proof. We use the transfer-function realization of Schur functions on the bidisc. The Schur–Agler class consists of the g∈S(D2)g\in\mathcal{S}(\mathbb{D}^2) with ∥g(X1,X2)∥≤1\lVert g(X_1,X_2)\rVert\leq1 for every pair of commuting strict contractions X1,X2X_1,X_2 on a Hilbert space. By Andô’s inequality [3], every g∈S(D2)g\in\mathcal{S}(\mathbb{D}^2) belongs to this class. By Agler’s theorem [1] (see also [4, 2]) there are a Hilbert space K\mathcal{K}, complementary orthogonal projections PP and QQ on K\mathcal{K}, and an isometry V=(κx∗yW)V=\begin{pmatrix}\kappa&x^*\\y&W\end{pmatrix} of C⊕K\mathbb{C}\oplus\mathcal{K}, the colligation, such that

g(z)=κ+x∗Z(I−WZ)−1y,Z=z1P+z2Q(z∈D2).g(z)=\kappa+x^*Z(I-WZ)^{-1}y,\qquad Z=z_1P+z_2Q\qquad(z\in\mathbb{D}^2).

For z∈D2z\in\mathbb{D}^2 we have ∥Z∥≤max⁡j∣zj∣<1\lVert Z\rVert\leq\max_j\lvert z_j\rvert<1, so I−WZI-WZ and the other operators I−ZWI-ZW, I−ZUI-ZU inverted below are invertible. We may take VV unitary, which matters if the realization is given with an isometric colligation, as in [14 Section 2]. Indeed, the operator (V00V∗)\begin{pmatrix}V&0\\0&V^*\end{pmatrix} on (C⊕K)⊕(C⊕K)(\mathbb{C}\oplus\mathcal{K})\oplus(\mathbb{C}\oplus\mathcal{K}) is unitary and agrees with VV on the first summand. We use it as the new colligation, with state space K⊕(C⊕K)\mathcal{K}\oplus(\mathbb{C}\oplus\mathcal{K}) and projections P⊕IP\oplus I and Q⊕0Q\oplus0. Since the new yy is the old one, the new WW and x∗x^* agree with the old ones on K\mathcal{K}, and the new projections map K\mathcal{K} into itself, each term x∗Z(WZ)myx^{*}Z(WZ)^{m}y of the expansion of gg is unchanged. From now on K\mathcal{K}, VV, xx, yy, WW, PP and QQ denote the new data.

Apply this to g=(H−1)/(H+1)g=(H-1)/(H+1), which lies in S(D2)\mathcal{S}(\mathbb{D}^{2}) and vanishes at 0. Then κ=0\kappa=0, and since VV is unitary,

∥x∥=∥y∥=1,Wx=0,W∗y=0,xx∗+W∗W=I,yy∗+WW∗=I.\|x\|=\|y\|=1,\qquad Wx=0,\qquad W^{*}y=0,\qquad xx^{*}+W^{*}W=I,\qquad yy^{*}+WW^{*}=I.

Put U=W+yx∗U=W+yx^{*}. These identities imply U∗U=UU∗=IU^{*}U=UU^{*}=I. Let G=I−ZWG=I-ZW. Then Gx=xGx=x and x∗G−1Zy=x∗Z(I−WZ)−1y=gx^{*}G^{-1}Zy=x^{*}Z(I-WZ)^{-1}y=g. Since I−ZU=G−Zyx∗I-ZU=G-Zyx^{*} is a rank-one perturbation of GG and 1−g≠01-g\ne0,

x∗(I−ZU)−1x=x∗G−1x+(x∗G−1Zy)(x∗G−1x)1−x∗G−1Zy=1+g1−g=11−g.x^{*}(I-ZU)^{-1}x=x^{*}G^{-1}x+\frac{(x^{*}G^{-1}Zy)(x^{*}G^{-1}x)}{1-x^{*}G^{-1}Zy}=1+\frac{g}{1-g}=\frac{1}{1-g}.

Hence x∗(I+ZU)(I−ZU)−1x=2x∗(I−ZU)−1x−1=(1+g)/(1−g)=Hx^{*}(I+ZU)(I-ZU)^{-1}x=2x^{*}(I-ZU)^{-1}x-1=(1+g)/(1-g)=H, and expanding the resolvent,

Hn(z)=2x∗(ZU)nx(n≥1).H_n(z)=2x^{*}(ZU)^n x\qquad(n\ge1).

Relative to K=PK⊕QK\mathcal{K}=P\mathcal{K}\oplus Q\mathcal{K} write

U=(U11U12U21U22),x=(ab),Ux=(a~b~),U=\begin{pmatrix}U_{11}&U_{12}\\U_{21}&U_{22}\end{pmatrix},\qquad x=\begin{pmatrix}a\\b\end{pmatrix},\qquad Ux=\begin{pmatrix}\tilde{a}\\\tilde{b}\end{pmatrix},

so that ∥a∥2+∥b∥2=∥a~∥2+∥b~∥2=1\|a\|^{2}+\|b\|^{2}=\|\tilde{a}\|^{2}+\|\tilde{b}\|^{2}=1. Collecting the monomials z1z2z_{1}z_{2}, z13z_{1}^{3} and z23z_{2}^{3} in 2x∗(ZU)nx2x^{*}(ZU)^n x, we find

12h11=a∗U12b~+b∗U21a~,12h30=a∗U112a~,12h03=b∗U222b~.\frac{1}{2}h_{11}=a^{*}U_{12}\tilde{b}+b^{*}U_{21}\tilde{a},\qquad\frac{1}{2}h_{30}=a^{*}U_{11}^{2}\tilde{a},\qquad\frac{1}{2}h_{03}=b^{*}U_{22}^{2}\tilde{b}.

Put σ=∥U11∗a∥2+∥U22∗b∥2\sigma=\|U_{11}^{*}a\|^{2}+\|U_{22}^{*}b\|^{2} and τ=∥U11a~∥2+∥U22b~∥2\tau=\|U_{11}\tilde{a}\|^{2}+\|U_{22}\tilde{b}\|^{2}. Since U∗U^{*} is an isometry, ∥U11∗a∥2+∥U12∗a∥2=∥a∥2\|U_{11}^{*}a\|^{2}+\|U_{12}^{*}a\|^{2}=\|a\|^{2} and ∥U21∗b∥2+∥U22∗b∥2=∥b∥2\|U_{21}^{*}b\|^{2}+\|U_{22}^{*}b\|^{2}=\|b\|^{2}, so ∥U12∗a∥2+∥U21∗b∥2=1−σ\|U_{12}^{*}a\|^{2}+\|U_{21}^{*}b\|^{2}=1-\sigma. Since UU is an isometry, ∥U12b~∥2+∥U21a~∥2=1−τ\|U_{12}\tilde{b}\|^{2}+\|U_{21}\tilde{a}\|^{2}=1-\tau in the same way. Applying the Cauchy–Schwarz inequality to 12h11=⟨b~,U12∗a⟩+⟨a~,U21∗b⟩=⟨U12b~,a⟩+⟨U21a~,b⟩\frac{1}{2}h_{11}=\langle\tilde{b},U_{12}^{*}a\rangle+\langle\tilde{a},U_{21}^{*}b\rangle=\langle U_{12}\tilde{b},a\rangle+\langle U_{21}\tilde{a},b\rangle in the two groupings, we obtain

14B22≤1−σand14B22≤1−τ.\frac{1}{4}B_{2}^{2}\le1-\sigma\qquad\text{and}\qquad\frac{1}{4}B_{2}^{2}\le1-\tau.

Finally 12∣h30∣=∣⟨U11a~,U11∗a⟩∣≤12(∥U11∗a∥2+∥U11a~∥2)\frac{1}{2}|h_{30}|=|\langle U_{11}\tilde{a},U_{11}^{*}a\rangle|\le\frac{1}{2}(\|U_{11}^{*}a\|^{2}+\|U_{11}\tilde{a}\|^{2}), and likewise for h03h_{03}, so A3≤σ+τ≤2−12B22A_{3}\le\sigma+\tau\le2-\frac{1}{2}B_{2}^{2}.

Remark 3.5. Lemmas 3.1 and 3.4 are sharp. Equality holds in both for H=(1+z1)/(1−z1)H=(1+z_{1})/(1-z_{1}), where A2=A3=2A_{2}=A_{3}=2 and B2=0B_{2}=0, and for H=(1+z1z2)/(1−z1z2)H=(1+z_{1}z_{2})/(1-z_{1}z_{2}), where A2=A3=0A_{2}=A_{3}=0 and B2=2B_{2}=2.

Lemma 3.6 (Patel [20]). Let α,β>0\alpha,\beta>0, γ≥0\gamma\ge0 and r>0r>0, and put

B(r)=4α+176β+2γ+14α+14α+2γ+(r−γ)2140β+r2112β.B(r)=4\alpha+176\beta+2\gamma+\frac{1}{4\alpha}+\frac{1}{4\alpha+2\gamma}+\frac{(r-\gamma)^{2}}{140\beta}+\frac{r^{2}}{112\beta}.

Then ∥H2∥1+r∥H3∥1≤B(r)\|H_{2}\|_{1}+r\|H_{3}\|_{1}\le B(r).

Proof. Expanding the squares verifies the identity

B(r)−(A2+B2+r(A3+B3))=α(4−A22−B22)+β(176−35A32−28B32)+γ(2−A3−12B22)+α(A2−12α)2+(α+γ2)(B2−12α+γ)2+35β(A3−r−γ70β)2+28β(B3−r56β)2.\begin{aligned} B(r)-\bigl(A_{2}+B_{2}+r(A_{3}+B_{3})\bigr) ={}&\alpha(4-A_{2}^{2}-B_{2}^{2})+\beta(176-35A_{3}^{2}-28B_{3}^{2})+\gamma\left(2-A_{3}-\frac{1}{2}B_{2}^{2}\right)\\ &+\alpha\left(A_{2}-\frac{1}{2\alpha}\right)^{2}+\left(\alpha+\frac{\gamma}{2}\right)\left(B_{2}-\frac{1}{2\alpha+\gamma}\right)^{2}\\ &+35\beta\left(A_{3}-\frac{r-\gamma}{70\beta}\right)^{2}+28\beta\left(B_{3}-\frac{r}{56\beta}\right)^{2}. \end{aligned}

By Lemmas 3.1, 3.3 and 3.4, each term on the right is nonnegative. □

Thus B(r)B(r) is the Lagrangian bound, with multipliers α\alpha, β\beta and γ\gamma, for the maximum of A2+B2+r(A3+B3)A_2+B_2+r(A_3+B_3) under the three inequalities of Lemmas 3.1, 3.3 and 3.4.

Sidon constants and certificates for small kk

The proof of Theorem 1.1 uses upper bounds for S1S_1 and for SkS_k with k≥4k \ge4. Degrees two and three are covered by Lemma 3.6.

The proof of Theorem 1.1 uses neither the value S3=5/3S_3=5/3 nor the value of S5S_5 below, so that it does not rely on these recent preprints. Ye et al. [24 Theorem 1.1] show that S5S_5 is the root in (2.18,2.19)(2.18,2.19) of 675X4−1026X3−936X2−414X+677675X^4-1026X^3-936X^2-414X+677. The proof of Proposition 4.2 rests on the following lemma.

Lemma 4.1. Let k≥1k \ge1. For w∈Ck+1w \in\mathbb{C}^{k+1} put

ν(w)=max⁡{Re⁡∑j=0kw‾jcj:max⁡T∣∑j=0kcjzj∣≤1}.\nu(w)=\max\left\{\operatorname{Re}\sum_{j=0}^{k}\overline{w}_j c_j:\max_{\mathbb{T}}\left|\sum_{j=0}^{k}c_jz^j\right|\le1\right\}.

(a) Sk=max⁡{ν(w):w0=w1=1, ∣wj∣=1 (2≤j≤k)}S_k=\max\{\nu(w):w_0=w_1=1,\ |w_j|=1\ (2\le j\le k)\}.

(b) ν\nu is convex.

(c) Let y∈Cky\in\mathbb{C}^k, and let T(y)T(y) be the Hermitian Toeplitz matrix with diagonal entries 11 and entries T(y)jl=yl−jT(y)_{jl}=y_{l-j} for l>jl>j. If T(y)T(y) is positive definite, then ν(w)2≤w∗T(y)−1w\nu(w)^2\le w^*T(y)^{-1}w.

(d) For 0<δ≤π/40<\delta\le\pi/4, the arc {eiφ:∣φ−φ0∣≤δ}\{e^{i\varphi}:|\varphi-\varphi_0|\le\delta\} lies in the triangle with vertices ei(φ0−δ)e^{i(\varphi_0-\delta)}, ei(φ0+δ)e^{i(\varphi_0+\delta)} and eiφ0/cos⁡δe^{i\varphi_0}/\cos\delta.

(e) ν(w‾)=ν(w)\nu(\overline{w})=\nu(w), where w‾=(w‾0,…,w‾k)\overline{w}=(\overline{w}_0,\ldots,\overline{w}_k).

Proof. (a) Write q(z)=∑jcjzjq(z)=\sum_j c_jz^j. If ∣wj∣=1|w_j|=1 for all jj, then ν(w)≤max⁡{∑j∣cj∣:max⁡T∣q∣≤1}=Sk\nu(w)\le\max\{\sum_j|c_j|:\max_{\mathbb{T}}|q|\le1\}=S_k. Equality holds when cc attains this maximum and wj=cj/∣cj∣w_j=c_j/|c_j| (any unimodular number if cj=0c_j=0). The substitution cj↦cjei(ψ+jθ)c_j\mapsto c_je^{i(\psi+j\theta)} replaces q(z)q(z) by eiψq(eiθz)e^{i\psi}q(e^{i\theta}z) and wjw_j by wje−i(ψ+jθ)w_je^{-i(\psi+j\theta)} without changing ν\nu, and a choice of ψ\psi and θ\theta makes w0=w1=1w_0=w_1=1. (b) ν\nu is a supremum of real-linear functions of ww. (c) By Lemma 3.2, c∗T(y)c≤max⁡T∣q∣2c^*T(y)c\le\max_{\mathbb{T}}|q|^2 for c=(c0,…,ck)Tc=(c_0,\ldots,c_k)^{\mathsf T}. Hence, by the Cauchy–Schwarz inequality,

Re⁡w∗c≤(w∗T(y)−1w)1/2(c∗T(y)c)1/2≤(w∗T(y)−1w)1/2max⁡T∣q∣.\operatorname{Re}w^*c\le(w^*T(y)^{-1}w)^{1/2}(c^*T(y)c)^{1/2}\le(w^*T(y)^{-1}w)^{1/2}\max_{\mathbb{T}}|q|.

(d) After a rotation, φ0=0\varphi_0=0. The triangle is {u∈C:Re⁡u≥cos⁡δ, Re⁡(ue−iδ)≤1, Re⁡(ueiδ)≤1}\{u\in\mathbb{C}:\operatorname{Re}u\ge\cos\delta,\ \operatorname{Re}(ue^{-i\delta})\le1,\ \operatorname{Re}(ue^{i\delta})\le1\}, and u=eiφu=e^{i\varphi} with ∣φ∣≤δ|\varphi|\le\delta satisfies cos⁡φ≥cos⁡δ\cos\varphi\ge\cos\delta and cos⁡(φ∓δ)≤1\cos(\varphi\mp\delta)\le1. (e) The map c↦c‾c\mapsto\overline{c} maps the set of admissible cc onto itself, since ∑jc‾jzj=q(z‾)‾\sum_j\overline{c}_jz^j=\overline{q(\overline{z})}, and Re⁡∑jwjc‾j=Re⁡∑jw‾jcj\operatorname{Re}\sum_jw_j\overline{c}_j=\operatorname{Re}\sum_j\overline{w}_jc_j. Hence ν(w‾)=max⁡Re⁡∑jwjcj=max⁡Re⁡∑jwjc‾j=ν(w)\nu(\overline{w})=\max\operatorname{Re}\sum_jw_jc_j=\max\operatorname{Re}\sum_jw_j\overline{c}_j=\nu(w). □

Parts (a) and (c) are variants of the reduction to phase functionals and of the Toeplitz domination bound used by Ye et al. [24 (14) and Proposition 3.2], who normalize the phases differently [24 (17)]. Proposition 4.2 also differs from their proof in the covering of the phase space.

Proposition 4.2. S5≤2.187S_{5} \le2.187, S6≤2.363S_{6} \le2.363 and S7≤2.552S_{7} \le2.552.

Proof (computer-assisted). Let (k,bk)(k,b_{k}) be (5,2.187)(5,2.187), (6,2.363)(6,2.363) or (7,2.552)(7,2.552). By Lemma 4.1(a) it suffices to show that ν(1,1,eiφ2,…,eiφk)≤bk\nu(1,1,e^{i\varphi_{2}},\ldots,e^{i\varphi_{k}}) \le b_{k} for (φ2,…,φk)∈[−π,π]k−1(\varphi_{2},\ldots,\varphi_{k}) \in[-\pi,\pi]^{k-1}. By Lemma 4.1(e), which replaces (φ2,…,φk)(\varphi_{2},\ldots,\varphi_{k}) by (−φ2,…,−φk)(-\varphi_{2},\ldots,-\varphi_{k}), we may also assume φ2≥0\varphi_{2} \ge0. For k=5k=5 and k=6k=6 the certificate covers [−π,π]k−1[-\pi,\pi]^{k-1} by boxes, obtained from a uniform grid of 8k−18^{k-1} boxes with half-widths π/8\pi/8 by repeated bisection. For k=7k=7 it covers [0,π]×[−π,π]5[0,\pi]\times[-\pi,\pi]^{5} in the same way, starting from a uniform grid of 2⋅452\cdot4^{5} boxes with half-widths π/4\pi/4. By Lemma 4.1(d), the vectors (1,1,eiφ2,…,eiφk)(1,1,e^{i\varphi_{2}},\ldots,e^{i\varphi_{k}}) with φ\varphi in the box lie in a product of k−1k-1 triangles. By Lemma 4.1(b), the maximum of ν\nu over this product is attained at one of its 3k−13^{k-1} vertices. For each vertex vv that occurs, the certificate gives y∈Cky \in\mathbb{C}^{k} with T(y)T(y) positive definite and v∗T(y)−1v<bk2v^{*}T(y)^{-1}v < b_{k}^{2}, so that ν(v)<bk\nu(v) < b_{k} by Lemma 4.1(c). A vertex with a coordinate −1-1 occurs with two labels, phase −π-\pi and phase π\pi, and has a certificate for each. Table 2 gives the sizes of the three certificates, and § 6 describes the check. □

Numerical maximization, which is not part of the proof, gives S6≈2.36256S_{6}\approx2.36256 and S7≈2.55166S_{7}\approx2.55166, so the bounds 2.363 and 2.552 are close to the true values. Further gains along these lines need bounds for SnS_{n} with n≥8n\ge8 or sharper inequalities in the degrees two and three.

Proof of Theorem 1.1

Throughout the paper, lower bounds are rounded down and upper bounds up. Let N=40N=40. Let s4=2s_{4}=2, s5=2.187s_{5}=2.187, s6=2.363s_{6}=2.363, s7=2.552s_{7}=2.552, and sn=⌈108n⌉/108s_{n}=\lceil10^{8}\sqrt{n}\rceil/10^{8} for 8≤n≤N8\le n\le N. By Table 1 and Proposition 4.2, Sn≤snS_{n}\le s_{n} for 4≤n≤N4\le n\le N, and Sn≤n≤nS_{n}\le\sqrt{n}\le n for all nn. Let H∈P2H\in\mathcal{P}_{2}, let 0<r<10<r<1, and fix α,β>0\alpha,\beta>0 and γ≥0\gamma\ge0. Lemma 2.3 bounds ∥Hn∥1\lVert H^{n}\rVert_{1} by 2 for n=1n=1, by 2sn2s_{n} for 4≤n≤N4\le n\le N and by 2n2n for n>Nn>N, and Lemma 3.6 bounds ∥H2∥1+r∥H3∥1\lVert H^{2}\rVert_{1}+r\lVert H^{3}\rVert_{1}. Summing the series ∑n>Nnrn=rN+1(N+1−Nr)/(1−r)2\sum_{n>N}nr^{n}=r^{N+1}(N+1-Nr)/(1-r)^{2}, we obtain the majorant

Mr(H)≤E(r):=2r+r2B(r)+2∑n=4Nsnrn+2rN+1(N+1−Nr)(1−r)2.M_{r}(H)\le E(r):=2r+r^{2}B(r)+2\sum_{n=4}^{N}s_{n}r^{n}+\frac{2r^{N+1}(N+1-Nr)}{(1-r)^{2}}.

We take

r∗=60333200000,α=2067405362500000,β=39014571562500000,γ=101990031109.r_{*}=\frac{60333}{200000},\qquad\alpha=\frac{20674053}{62500000},\qquad\beta=\frac{3901457}{1562500000},\qquad\gamma=\frac{101990031}{10^{9}}.

All quantities in E(r∗)E(r_{*}) are then rational, and an exact computation shows E(r∗)<0.99999556<1E(r_{*})<0.99999556<1. Since α\alpha, β\beta and γ\gamma are fixed, EE is continuous on (0,1)(0,1), so there is r′>r∗r'>r_{*} with E(r′)<1E(r')<1. By Lemma 2.1, K2≥r′>r∗K_{2}\ge r'>r_{*}, which proves Theorem 1.1.

Remark 5.1 (Patel’s computation). With Patel’s parameters α=264631/800000\alpha= 264631/800000, β=249627/108\beta= 249627/10^{8} and γ=10197021/108\gamma= 10197021/10^{8}, with N=20N = 20 and with sn=⌈108n⌉/108s_n = \lceil10^{8}\sqrt{n} \rceil/10^{8} for 4≤n≤204 \le n \le20, the function EE is the rational majorant that he evaluates in [20]; its value at his radius rP=15079379/(5⋅107)r_P = 15079379/(5 \cdot10^{7}) is less than 0.99999997857, below his bound 1−1/(5⋅107)1 - 1/(5 \cdot10^{7}).

Verification

Computer checks

The certificates of Proposition 4.2 are checked in the ball arithmetic of Arb [13], through python-flint [22]. The checker reads the numbers of the certificates as binary64 numbers, exactly, rebuilds the boxes from the recorded bisections and verifies at every vertex the condition stated in the proof of Proposition 4.2; for S6S_6 and S7S_7, whose certificates are split into files, a further program checks, from the outputs of the file checks, that the files together cover the phase domain. The majorant E(r∗)E(r_*) of § 5 is evaluated in exact rational arithmetic, and so is Patel’s majorant in Remark 5.1. The programs and their output are archived with the certificates (see Data availability).

Formal proof

The formal proof is written in the Lean 4 proof assistant [15] with the Mathlib library [16]. The formal statements use Mathlib’s definitions. A family c:N2→Cc:\mathbb{N}^{2}\to\mathbb{C} is admissible if the series ∑αcαzα\sum_{\alpha}c_{\alpha}z^{\alpha} converges absolutely at every point of D2\mathbb{D}^{2} and its value there has modulus less than 1. Bohr’s inequality holds at rr if ∑α∈F∣cα∣r∣α∣≤1\sum_{\alpha\in F}|c_{\alpha}|r^{|\alpha|}\le1 for every admissible family cc and every finite set F⊂N2F\subset\mathbb{N}^{2}, and K2K_{2} is the supremum of the radii r>0r>0 at which it holds. It agrees with the definition in the introduction: a nonconstant f∈S(D2)f\in S(\mathbb{D}^{2}) maps D2\mathbb{D}^{2} into D\mathbb{D}, and a constant of modulus 1 satisfies Bohr’s inequality with equality. The main theorem is the following.

Theorem 6.1 (formalized in Lean). Suppose that A3+12B22≤2A_{3}+\frac{1}{2}B_{2}^{2}\le2, the inequality of Lemma 3.4, holds for every family hh with h00=1h_{00}=1 whose series converges absolutely on D2\mathbb{D}^{2} to values with positive real part, that is, for the coefficients of every H∈P2H\in\mathcal{P}_{2}. Then K2≥0.301665K_{2}\ge0.301665.

The proof of the hypothesis rests on Andô’s inequality and Agler’s realization, which are not in Mathlib. The formal proof follows Sections 2 to 5 and checks the certificates of Proposition 4.2 in one of two ways.

(1) A checking program written in Lean validates the certificates, converted into Lean source files, in exact rational arithmetic and ball arithmetic, and a Lean proof shows that every part it accepts implies the bound of Lemma 4.1(c) on the boxes of that part, so that the converted files need not be trusted. These checks run as compiled code, through native_decide, so the trusted base includes Lean’s compiler and runtime; the rest of the proof uses only the standard axioms.

(2) A second formalization proves the three bounds of Proposition 4.2 with every check run by Lean’s kernel. With the step of the first proof that derives the main theorem from them, the main theorem follows from four theorems in which every check is run by the kernel; these are not combined into a single Lean theorem.

Data availability

The certificates of Proposition 4.2, the programs of § 6, the programs that produced the certificates, the controls and the output of all runs, and the Lean files of § 6.2, are archived at doi:10.5281/zenodo.23716440. The same files, except the 48 files of the certificate for S6S_6, the 576 files of the certificate for S7S_7 and an archive of generated Lean files of the kernel-only proof of S7≤2.552S_7\le2.552, are also in the repository https://github.com/tejasvi/bidisc-bohr-radius/tree/v6. The README of the archive lists the Lean theorems and projects, with the versions of Lean and Mathlib, and gives the commands that run the checks and build the proofs. The certificate for S5S_{5} has the SHA-256 checksum ce3e9dd41a0dbd9c3a8ea49b04aa7aaf21f0fe557291a2084c7739461a460a14, and the archive lists the checksums of the other files.

Use of AI tools

Generative AI tools were used to find and prove the results, to write the programs, and to draft the text. The author reviewed the arguments and the programs and takes responsibility for the content.

References

ORCID: https://orcid.org/0000-0003-4668-3639

References

  1. [1]J. Agler, On the representation of certain holomorphic functions defined on a polydisc, in: Topics in Operator Theory: Ernst D. Hellinger Memorial Volume, Oper. Theory Adv. Appl. 48, Birkhäuser, Basel, 1990, 47–66.
  2. [2]J. Agler and J. E. McCarthy, Pick interpolation and Hilbert function spaces, Grad. Stud. Math. 44, Amer. Math. Soc., Providence, RI, 2002.DOI
  3. [3]T. Andō, On a pair of commutative contractions, Acta Sci. Math. (Szeged) 24 (1963), 88–90.
  4. [4]J. A. Ball and T. T. Trent, Unitary colligations, reproducing kernel Hilbert spaces, and Nevanlinna–Pick interpolation in several variables, J. Funct. Anal. 157 (1998), 1–61.DOI
  5. [6]F. Bayart, D. Pellegrino and J. B. Seoane-Sepúlveda, The Bohr radius of the n-dimensional polydisk is equivalent to √(log n)/n, Adv. Math. 264 (2014), 726–746.
  6. [7]L. Bernal-González, H. J. Cabana, D. García, M. Maestre, G. A. Muñoz-Fernández and J. B. Seoane-Sepúlveda, A new approach towards estimating the n-dimensional Bohr radius, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Math. RACSAM 115 (2021), no. 2, Paper No. 44.
  7. [8]H. P. Boas and D. Khavinson, Bohr’s power series theorem in several variables, Proc. Amer. Math. Soc. 125 (1997), 2975–2979.arxiv.org/abs/math/9606203
  8. [9]H. Bohr, A theorem concerning power series, Proc. London Math. Soc. (2) 13 (1914), 1–5.DOI
  9. [11]A. Defant, L. Frerick, J. Ortega-Cerdà, M. Ounaïes and K. Seip, The Bohnenblust–Hille inequality for homogeneous polynomials is hypercontractive, Ann. of Math. (2) 174 (2011), 485–497.DOI
  10. [12]S. Griego, Improve C₅₉ upper bound, pull request no. 75 to the repository teorth/optimizationproblems, GitHub, May 2026, https://github.com/teorth/optimizationproblems/pull/75.
  11. [13]F. Johansson, Arb: efficient arbitrary-precision midpoint-radius interval arithmetic, IEEE Trans. Comput. 66 (2017), 1281–1292.arxiv.org/abs/1611.02831
  12. [14]G. Knese, Three radii associated to Schur functions on the polydisk, Proc. Amer. Math. Soc. Ser. B 12 (2025), 48–63; arXiv:2410.21693v3.DOI
  13. [15]L. de Moura and S. Ullrich, The Lean 4 theorem prover and programming language, in: Automated Deduction – CADE 28, Lecture Notes in Comput. Sci. 12699, Springer, 2021, 625–635.DOI
  14. [16]The mathlib Community, The Lean mathematical library, in: Proc. 9th ACM SIGPLAN Int. Conf. Certified Programs and Proofs (CPP 2020), 2020, 367–381.DOI
  15. [18]S. Neuwirth, The Sidon constant of sets with three elements, arXiv:math/0102145, 2001.DOI
  16. [19]S. Neuwirth, On the (Fourier analytic) Sidon constant of {0, 1, 2, 3}, arXiv:2603.28229, 2026 (note dated 2013).arxiv.org/abs/2603.28229
  17. [20]S. Patel, Improved lower and upper bounds for the bidisc Bohr radius, solution to the problem “Exact value of the Bohr radius of the bidisc”, MathDB, second solution on the page, posted 27 August 2026 (UTC) and edited the same day, https://mathdb.com/p/392062/exact-value-of-the-bohr-radius-of-the-bidisc#s2 (consulted 21 October 2026).
  18. [21]S. Patel, Complex phases and a certified upper bound for the bidisc Bohr radius: reproducibility package, version 1.0.0, Zenodo, 5 September 2026, doi:10.5281/zenodo.22341928.DOI
  19. [22]python-flint: Python bindings for FLINT, version 0.9.0 (FLINT 3.6.0), https://github.com/flintlib/python-flint.
  20. [23]H. S. Shapiro, Extremal problems for polynomials and power series, Master’s thesis, Massachusetts Institute of Technology, 1951.
  21. [24]Z. Ye et al., The exact Sidon constant of six consecutive frequencies, preprint, October 2026, https://github.com/ABCDEFGCoder/The-Sidon-Constant-of-n-Consecutive-Frequencies.
  22. [25]H. Yu et al., The exact Sidon constant of {0,1,2,3} and its formalization, preprint, ai.viXra.org:2606.0009, 2026 (dated 10 May 2026); Lean formalization at https://github.com/yhx-12243/Sidon3.

Paper details

vlnc:W_7rg9rmkd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Mathematics

Generative AI tools

View statement
Looking for related papers…
Contents