betaFeedback

WorksMathematics

Truncated Dirichlet estimates for Bloch functions

Abstract

We study the truncated Dirichlet functional Φn(f)=∑k=1nk∣ak∣2,f(z)=∑k=0∞akzk, Φ_n (f) = \sum_{k = 1}^n k |a_k|^2, \quad f(z) = \sum_{k = 0}^\infty a_k z^k, when ff belongs to the Bloch space. If n≥27n \geq 27, we find a normalized Bloch function fnf_n such that Φn(fn)>nBn2, Φ_n (f_n) > n B_n^2, where Bn=max⁡{∣an∣:∥f∥B=1}B_n = \max \{ |a_n| : \| f \|_{\mathcal{B}} = 1 \}, thus proving that an inequality by Kayumov and Wirths cannot be extended to large values of nn. Nevertheless, we show that the inequality Φn(f)≤nBn2Φ_n(f) \leq n B_n^2 holds for all normalized functions when n≤9n \leq 9.

Introduction

Let D\mathbb{D} be the open unit disc of the complex plane C\mathbb{C}. The Bloch space B\mathcal{B} is defined as the space of all analytic functions in D\mathbb{D} such that

ρB(f)=sup⁡∣z∣<1{∣f′(z)∣(1−∣z∣2)}<∞.(1)\rho_{\mathcal{B}}(f)=\sup_{|z|<1}\{|f'(z)|(1-|z|^2)\}<\infty. \tag*{(1)}

Note that ρB\rho_{\mathcal{B}} is a seminorm, and it is well-known that B\mathcal{B} becomes a Banach space when endowed with the norm ∥f∥B=∣f(0)∣+ρB(f)\|f\|_{\mathcal{B}}=|f(0)|+\rho_{\mathcal{B}}(f). An important property of B\mathcal{B} is that it is Möbius invariant, i.e., ρB(f)=ρB(f∘φ)\rho_{\mathcal{B}}(f)=\rho_{\mathcal{B}}(f\circ\varphi) for every disk automorphism φ\varphi. Actually, as shown by Rubel and Timoney [11], it is the largest Banach space of holomorphic functions (admitting a nontrivial continuous linear functional) with this property.

Another Möbius invariant space is the Dirichlet space D\mathcal{D}. We say that an analytic function ff belongs to D\mathcal{D} if

ρD(f)2=∫D∣f′(z)∣2 dA(z)<∞,\rho_{\mathcal{D}}(f)^2=\int_{\mathbb{D}}|f'(z)|^2\,dA(z)<\infty,

where dA(z)dA(z) represents the normalized area measure of D\mathbb{D}. Actually, D\mathcal{D} is the unique Hilbert space with this property [2] and it is easy to see that the inclusion D⊂B\mathcal{D}\subset\mathcal{B} is proper. In the case that ff is univalent, the change of variables ζ=f(z)\zeta=f(z) shows that ρD2(f)\rho_{\mathcal{D}}^2(f) is precisely the normalized area of f(D)f(\mathbb{D}). If f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_kz^k, Parseval’s identity yields the alternative representation

ρD2(f)=∑k=1∞k∣ak∣2.\rho_{\mathcal{D}}^2(f)=\sum_{k=1}^{\infty}k|a_k|^2.

We refer the reader to the seminal paper [1] and the monograph [4] for further information about B\mathcal{B} and D\mathcal{D}, respectively.

In 1978, Wirths [12] showed that if n≥2n \ge2 and f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_kz^k is a normalized Bloch function (i.e., a function with ∥f∥B=1\|f\|_{\mathcal{B}}=1), then

∣an∣≤n+12n(n+1n−1)n−12=:Bn.(2)|a_n| \le\frac{n+1}{2n}\left(\frac{n+1}{n-1}\right)^{\frac{n-1}{2}}=:B_n. \tag*{(2)}

Furthermore, equality is possible if and only if ff is a rotation of z↦Bnznz\mapsto B_nz^n (see [7 Corollary 2]). If n=1n=1, it trivially follows from (1) that ∣a1∣≤1|a_1|\le1 so we define B1=1B_1=1 accordingly.

Related to this bound, consider the quadratic functional

Φn(f)=∑k=1nk∣ak∣2,f∈B.\Phi_n(f)=\sum_{k=1}^{n}k|a_k|^2,\qquad f\in\mathcal{B}.

In other words, Φn(f)\Phi_n(f) is the truncation of the (perhaps divergent) series defining ρD2(f)\rho_{\mathcal{D}}^2(f). In 2017, Kayumov [5] proved that

Φn(f)≤nBn2(3)\Phi_n(f)\le nB_n^2 \tag*{(3)}

for all f∈Bf\in\mathcal{B} with ∥f∥B≤1\|f\|_{\mathcal{B}}\le1 when n≤5n\le5. Actually, his proof yields that

∑k=[n/2]nk∣ak∣2≤nBn2,\sum_{k=[n/2]}^{n}k|a_k|^2\le nB_n^2,

where [x][x] denotes the integer part of xx, for all nn [5 p. 467]. These results were later complemented by Kayumov and Wirths [6], who showed that (3) holds for n=6n=6 as well. Other related extremal problems for coefficients of Bloch functions have been studied since then (e.g., [8] and [10]).

In [3], it was proved that (3) holds for all nn when the coefficients of ff are non-negative. Furthermore,

nBn2≤max⁡{Φn(f):∥f∥B=1}≤nn(n−1)n−1<4enBn2,nB_n^2\le\max\{\Phi_n(f):\|f\|_{\mathcal{B}}=1\}\le\frac{n^n}{(n-1)^{n-1}}<\frac{4}{e}nB_n^2,

so the bound nBn2nB_n^2 is of the correct order of magnitude. Even though it was not made explicit in the original paper [5], it was mentioned in [9 p. 6] that (3) could be valid for every nn.

Despite this evidence, the first main result of this work is to show that (3) does not hold when nn is large enough.

Theorem 1. Let n≥27n\ge27. Then, there exists f∈Bf\in\mathcal{B} with ∥f∥B=1\|f\|_{\mathcal{B}}=1 such that

Φn(f)>nBn2.\Phi_n(f)>nB_n^2.

Nevertheless, we are able to show that (3) holds for all n≤9n\le9. Namely, we use a refinement of the methods of [5] to prove the following:

Theorem 2. Suppose that 6≤n≤96\le n\le9. Then we have

Φn(f)≤nBn2\Phi_n(f)\le nB_n^2

for all normalized Bloch functions. Furthermore, equality holds if and only if f(z)=λBnznf(z)=\lambda B_nz^n for some unimodular constant λ\lambda.

Observe that, as stated before, the case n=6n = 6 was already known [6]. However, the present argument will represent a new proof of this fact.

To the best of our knowledge, the cases 10≤n≤2610 \le n \le26 are still open.

This work is organized as follows: the construction of the counterexamples to (3) which lead to the proof of Theorem 1 can be found in Section 2. Indeed, we will present one function for the case n=27n = 27 and another that covers all other cases n≥28n \ge28. Section 3 is devoted to the positive solution for n∈{6,7,8,9}n \in\{6,7,8,9\}. In Section 4, we extend Theorem 1 to the family of generalized weighted Dirichlet functionals Φnt\Phi_n^t. Finally, an appendix with a series of elementary but technical results is included for the correct understanding of the main sections.

AI statement

AI statement. This work has been carried out with assistance from Claude Opus 5.5 (Anthropic). It began with the aim of continuing the work [3] and attempting to prove (3) for every nn. When asked about the case n=7n = 7, it produced a first draft of solution for 6≤n≤96 \le n \le9. In particular, it proposed the use of Lemma 5 and the restricted parameter κ\kappa that can be found in Section 3. While exploring a suitable replacement of Lemma 5 for larger nn, it found a counterexample for (3) when n≥28n \ge28. Further discussions led to the failure of the case n=27n = 27 as well. All these proofs have been distilled, simplified and written in their final form by the authors.

Acknowledgements

Acknowledgements. The three authors were partially supported by the grant PID2025-168349NB-I00 from Ministerio de Ciencia, Innovación y Universidades and by the project CIGE/2026/195 from Conselleria d’Educació, Cultura i Universitats, Generalitat Valenciana. The work of the second author was also supported by the grant PID2024-160326NA-I00 from the same ministry.

Failure of (3) for large nn

In this section we will construct the counterexample needed to prove Theorem 1. In light of [3 Theorem 2], the coefficients of this function must not have constant argument. Actually, the same proof yields that it is not possible that

arg⁡ak=α+kβ,1≤k,\operatorname{arg} a_k = \alpha+ k\beta,\qquad1 \le k,

for certain α,β∈R\alpha,\beta\in\mathbb{R}. This represents the main difficulty of the following lemma.

Lemma 3. Let n≥4n \ge4. Consider the function

fε(z)=εz+Bnzn−ε2n−1(2nn+1)2Bn2z2n−1,ε>0.(4)f_\varepsilon(z) = \varepsilon z + B_n z^n - \frac{\varepsilon}{2n-1}\left(\frac{2n}{n+1}\right)^2 B_n^2 z^{2n-1},\qquad\varepsilon> 0. \tag*{(4)}

Then, there exists εn>0\varepsilon_n > 0 such that ρB2(fε)=1+16ε2(n+1)2\rho_{\mathcal{B}}^2(f_\varepsilon) = 1 + \frac{16\varepsilon^2}{(n+1)^2} for all 0<ε<εn0 < \varepsilon< \varepsilon_n.

Proof. Write z=reitz=re^{it}. Then we have that

∣fε′(z)∣2=∣nBnzn−1+ε(1−(2nn+1)2Bn2z2n−2)∣2=n2Bn2r2n−2+2nBnεRe⁡{zn−1(1−(2nn+1)2Bn2z2n−2)}+ε2∣1−(2nn+1)2Bn2z2n−2∣2=n2Bn2r2n−2+2nBnA(n,r,t)ε+B(n,r,t)ε2,(5)\begin{aligned} |f'_{\varepsilon}(z)|^2 &=\left|nB_nz^{n-1}+\varepsilon\left(1-\left(\frac{2n}{n+1}\right)^2B_n^2z^{2n-2}\right)\right|^2\\ &=n^2B_n^2r^{2n-2}+2nB_n\varepsilon\operatorname{Re}\left\{z^{n-1}\left(1-\left(\frac{2n}{n+1}\right)^2B_n^2z^{2n-2}\right)\right\}+\varepsilon^2\left|1-\left(\frac{2n}{n+1}\right)^2B_n^2z^{2n-2}\right|^2\\ &=n^2B_n^2r^{2n-2}+2nB_nA(n,r,t)\varepsilon+B(n,r,t)\varepsilon^2, \tag*{(5)} \end{aligned}

where

A(n,r,t)=(1−(2nn+1)2Bn2r2n−2)rn−1cos⁡((n−1)t)=(1−(n+1n−1r2)n−1)rn−1cos⁡((n−1)t),(6)\begin{aligned} A(n,r,t) &=\left(1-\left(\frac{2n}{n+1}\right)^2B_n^2r^{2n-2}\right)r^{n-1}\cos((n-1)t)\\ &=\left(1-\left(\frac{n+1}{n-1}r^2\right)^{n-1}\right)r^{n-1}\cos((n-1)t), \tag*{(6)} \end{aligned}
B(n,r,t)=1−2(2nn+1)2Bn2r2n−2cos⁡(2(n−1)t)+(2nn+1)4Bn4r4n−4=(1+(n+1n−1r2)n−1)2−4(n+1n−1r2)n−1cos⁡2((n−1)t).(7)\begin{aligned} B(n,r,t) &=1-2\left(\frac{2n}{n+1}\right)^2B_n^2r^{2n-2}\cos(2(n-1)t)+\left(\frac{2n}{n+1}\right)^4B_n^4r^{4n-4}\\ &=\left(1+\left(\frac{n+1}{n-1}r^2\right)^{n-1}\right)^2-4\left(\frac{n+1}{n-1}r^2\right)^{n-1}\cos^2((n-1)t). \tag*{(7)} \end{aligned}

(recall the definition of BnB_n (2)). If r>0r>0 and ε\varepsilon are fixed, observe from (5), (6) and (7) that ∣fε′(reit)∣2|f'_{\varepsilon}(re^{it})|^2 is a concave quadratic function of cos⁡((n−1)t)\cos((n-1)t). That quadratic function attains its maximum at its vertex

vε(r)=n+18ε((n+1n−1r2)−n−12−(n+1n−1r2)n−12).v_{\varepsilon}(r)=\frac{n+1}{8\varepsilon}\left(\left(\frac{n+1}{n-1}r^2\right)^{-\frac{n-1}{2}}-\left(\frac{n+1}{n-1}r^2\right)^{\frac{n-1}{2}}\right).

For the critical radius rn=n−1n+1r_n=\sqrt{\frac{n-1}{n+1}} we have that vε(rn)=0v_{\varepsilon}(r_n)=0 and therefore on that radius we deduce that

∣fε′(rneit)∣2(1−rn2)2≤(n2Bn2rn2n−2+(1+(n+1n−1rn2)n−1)2ε2)(1−rn2)2=1+16ε2(n+1)2,(8)\begin{aligned} |f'_{\varepsilon}(r_ne^{it})|^2(1-r_n^2)^2 &\leq\left(n^2B_n^2r_n^{2n-2}+\left(1+\left(\frac{n+1}{n-1}r_n^2\right)^{n-1}\right)^2\varepsilon^2\right)(1-r_n^2)^2\\ &=1+\frac{16\varepsilon^2}{(n+1)^2}, \tag*{(8)} \end{aligned}

and equality in (8) is possible when e(n−1)tie^{(n-1)ti} is a purely imaginary number. This in particular shows that ρ∞2(fε)≥1+16ε2(n+1)2\rho^2_{\infty}(f_{\varepsilon})\geq1+\frac{16\varepsilon^2}{(n+1)^2} for every ε>0\varepsilon>0.

We will now show that the converse inequality holds if ε\varepsilon is small enough. If r=0r=0, it is clear that ∣fε′(0)∣2=ε2<1+16ε2(n+1)2|f'_{\varepsilon}(0)|^2=\varepsilon^2<1+\frac{16\varepsilon^2}{(n+1)^2} if ε<1\varepsilon<1. From now on assume that 0<r<10<r<1 and thus vε(r)v_{\varepsilon}(r) is well defined. In order to simplify the later notation, write R=(n+1n−1r2)n−1R=\left(\frac{n+1}{n-1}r^2\right)^{n-1} so that

vε(r)=n+18ε(1R−R).v_{\varepsilon}(r)=\frac{n+1}{8\varepsilon}\left(\sqrt{\frac{1}{R}}-\sqrt{R}\right).

Observe that vε(r)v_{\varepsilon}(r) is a continuous and strictly decreasing function of RR (and therefore of rr) and monotonic with respect to ε\varepsilon.

Even if there is no tt such that cos⁡((n−1)t)=vε(r)\cos((n-1)t)=v_{\varepsilon}(r), the following inequality holds:

∣fε′(reit)∣2≤(n+1)24R+(n+1)2(1−R)28−(n+1)216R(R−1R)2+(1+R)2ε2=(n+1)2(1+R)216+(1+R)2ε2=(1+16ε2(n+1)2)(n+1)2(1+R)216.\begin{aligned} \left|f_{\varepsilon}'(re^{it})\right|^{2}\leq\frac{(n+1)^{2}}{4}R+\frac{(n+1)^{2}(1-R)^{2}}{8}-\frac{(n+1)^{2}}{16}R\left(\sqrt{R}-\sqrt{\frac{1}{R}}\right)^{2}+(1+R)^{2}\varepsilon^{2} \\ &=\frac{(n+1)^{2}(1+R)^{2}}{16}+(1+R)^{2}\varepsilon^{2}=\left(1+\frac{16\varepsilon^{2}}{(n+1)^{2}}\right)\frac{(n+1)^{2}(1+R)^{2}}{16}. \end{aligned}

According to Remark 2 in the Appendix, there exists r0<rnr_{0}<r_{n} independent of ε\varepsilon such that (1+R)(1−r2)≤4n+1(1+R)(1-r^{2})\leq\frac{4}{n+1} if r0≤r<1r_{0}\leq r<1. We split the proof by cases.

(1) If r≥r0r\geq r_{0}, then

∣fε′(reit)∣2(1−r2)2≤(1+16ε2(n+1)2)(n+1)2(1+R)2(1−r2)216≤1+16ε2(n+1)2.\left|f_{\varepsilon}'(re^{it})\right|^{2}(1-r^{2})^{2}\leq\left(1+\frac{16\varepsilon^{2}}{(n+1)^{2}}\right)\frac{(n+1)^{2}(1+R)^{2}(1-r^{2})^{2}}{16}\leq1+\frac{16\varepsilon^{2}}{(n+1)^{2}}.

(2) When 0<r<r00<r<r_{0}, in particular vε(r)>vε(r0)>vε(rn)=0v_{\varepsilon}(r)>v_{\varepsilon}(r_{0})>v_{\varepsilon}(r_{n})=0. Pick εn<1\varepsilon_{n}<1 such that vεn(r0)≥1v_{\varepsilon_{n}}(r_{0})\geq1. For every 0<ε<εn0<\varepsilon<\varepsilon_{n} the quantity ∣fε′(reit)∣2\left|f_{\varepsilon}'(re^{it})\right|^{2} is then maximized when e(n−1)ti=1e^{(n-1)ti}=1. Thus

∣fε′(reit)∣2≤(n+1)24R+(n+1)R(1−R)ε+((1+R)2−4R)ε2=(n+12R+(1−R)ε)2.\left|f_{\varepsilon}'(re^{it})\right|^{2}\leq\frac{(n+1)^{2}}{4}R+(n+1)\sqrt{R(1-R)}\varepsilon+\left((1+R)^{2}-4R\right)\varepsilon^{2}=\left(\frac{n+1}{2}\sqrt{R}+(1-R)\varepsilon\right)^{2}.

Summing up, we have that

∣fε′(reit)∣(1−r2)≤(n+12R+(1−R)ε)(1−r2)≤nBnrnn−1(1−rn2)+ε,\left|f_{\varepsilon}'(re^{it})\right|(1-r^{2})\leq\left(\frac{n+1}{2}\sqrt{R}+(1-R)\varepsilon\right)(1-r^{2})\leq nB_{n}r_{n}^{n-1}(1-r_{n}^{2})+\varepsilon,

since R<1R<1 because r<r0<rnr<r_{0}<r_{n}. The monotonicity of the function rn−1(1−r2)r^{n-1}(1-r^{2}) (see Lemma 10 in the Appendix) yields that we may also choose εn\varepsilon_{n} such that nBnr0n−1(1−r02)+εn<1nB_{n}r_{0}^{n-1}(1-r_{0}^{2})+\varepsilon_{n}<1. Under this new assumption, we have that ∣fε′(reit)∣(1−r2)<1\left|f_{\varepsilon}'(re^{it})\right|(1-r^{2})<1 if 0<r<r00<r<r_{0}. This concludes the proof.

Proof of Theorem 1 for n≥28n\geq28. Let fεf_{\varepsilon} be the function defined in (4) and assume that 0<ε<εn0<\varepsilon<\varepsilon_{n}. Then we have that

Φn(fε∥fε∥B)=ε2+nBn21+16ε2(n+1)2=nBn21+4nε2(n+1)2(n−1n+1)n−11+16ε2(n+1)2.\begin{aligned} \Phi_{n}\left(\frac{f_{\varepsilon}}{\left\|f_{\varepsilon}\right\|_{\mathcal{B}}}\right)=\frac{\varepsilon^{2}+nB_{n}^{2}}{1+\frac{16\varepsilon^{2}}{(n+1)^{2}}} \\ &=nB_{n}^{2}\frac{1+\frac{4n\varepsilon^{2}}{(n+1)^{2}}\left(\frac{n-1}{n+1}\right)^{n-1}}{1+\frac{16\varepsilon^{2}}{(n+1)^{2}}}. \end{aligned}

Thus, it is enough to check that

n(n−1n+1)n−1>4.(9)n\left(\frac{n-1}{n+1}\right)^{n-1}>4. \tag*{(9)}

The proof concludes by observing that the sequence in the left-hand side of (9) is increasing and that 28(2729)27>428\left(\frac{27}{29}\right)^{27}>4. □

Observe that n=28n=28 is the smallest integer such that (9) holds. Hence, a counterexample for n=27n=27 must be necessarily different. To this end, we will add an extra high-order term which will not contribute to Φn\Phi_n.

Proof of Theorem 1 for n=27n=27. Given ε>0\varepsilon>0, consider the polynomial

fε(z)=14εz+B27z27−1453(1413)26εz53+179(1413)39ε2z79.(10)f_\varepsilon(z)=14\varepsilon z+B_{27}z^{27}-\frac{14}{53}\left(\frac{14}{13}\right)^{26}\varepsilon z^{53}+\frac{1}{79}\left(\frac{14}{13}\right)^{39}\varepsilon^2z^{79}. \tag*{(10)}

Thus,

Φ27(fε∥fε∥B)=27B2721+27(1314)26ε2∥fε∥B2.\Phi_{27}\left(\frac{f_\varepsilon}{\|f_\varepsilon\|_{\mathcal B}}\right)=27B_{27}^{2}\frac{1+27\left(\frac{13}{14}\right)^{26}\varepsilon^2}{\|f_\varepsilon\|_{\mathcal B}^{2}}.

We will check that ∥fε∥B2<1+27(1314)26ε2\|f_\varepsilon\|_{\mathcal B}^{2}<1+27\left(\frac{13}{14}\right)^{26}\varepsilon^2 if ε\varepsilon is small enough.

It is plain that

fε′(z)−27B27z26=14ε−14(1413)26εz52+(1413)39ε2z78.f_\varepsilon'(z)-27B_{27}z^{26}=14\varepsilon-14\left(\frac{14}{13}\right)^{26}\varepsilon z^{52}+\left(\frac{14}{13}\right)^{39}\varepsilon^2z^{78}.

Therefore, fε→B27z27f_\varepsilon\to B_{27}z^{27} in the Bloch topology when ε→0+\varepsilon\to0^+. In particular, if

∣fε′(zε)∣(1−∣zε∣2)=ρB(fε)|f_\varepsilon'(z_\varepsilon)|(1-|z_\varepsilon|^2)=\rho_{\mathcal B}(f_\varepsilon)

then ∣zε∣2→1314|z_\varepsilon|^2\to\frac{13}{14} (see Lemma 10). In particular, if ε\varepsilon is small enough we may use Remark 3 to see that

∥fε∥B2=sup⁡r1≤∣z∣≤r2{∣fε′(z)∣2(1−∣z∣2)2}<1+27(1314)26ε2,\|f_\varepsilon\|_{\mathcal B}^{2}=\sup_{r_1\le|z|\le r_2}\left\{|f_\varepsilon'(z)|^2(1-|z|^2)^2\right\}<1+27\left(\frac{13}{14}\right)^{26}\varepsilon^2,

so fεf_\varepsilon is a valid counterexample of (3). □

Positive solutions for 6≤n≤96 \le n \le9

If ∥f∥B≤1\|f\|_{\mathcal B}\le1, the definition of the Bloch seminorm yields that

∣f′(z)∣2≤1(1−r2)2,∣z∣=r.|f'(z)|^2\le\frac{1}{(1-r^2)^2},\qquad|z|=r.

Integrating over the circle {∣z∣=r}\{|z|=r\} and applying Parseval’s identity imply that

∑k=1∞k2∣ak∣2r2k−2=12π∫02π∣f′(reit)∣2 dt≤1(1−r2)2,0<r<1.(11)\sum_{k=1}^{\infty}k^2|a_k|^2r^{2k-2}=\frac{1}{2\pi}\int_0^{2\pi}|f'(re^{it})|^2\,dt\le\frac{1}{(1-r^2)^2},\qquad0<r<1. \tag*{(11)}

This elementary but ingenious inequality allows us to deduce (3) for 1≤n≤51\le n\le5 when substituting by the critical radius rn=n−1n+1r_n=\sqrt{\frac{n-1}{n+1}} (see [5]). However, it is not sufficient by itself to cover the cases 6≤n≤96\le n\le9.

We will divide the analysis into two complementary propositions based on the relative magnitude of the nn-th Taylor coefficient ∣an∣|a_n| with respect to the first one ∣a1∣|a_1|.

We first consider the case where the nn-th coefficient is relatively small. In this case, the defect generated by the intermediate coefficients combined with Parseval’s identity is sufficient to establish the desired inequality under some additional technical hypotheses which will be checked later in Proposition 8.

Proposition 4. Let n≥3n \ge3. Consider the sequence

(wk)wk=k(n−1n+1)k−1,1≤k,(w_k) \qquad w_k = k\left(\frac{n-1}{n+1}\right)^{k-1}, \qquad1 \le k,

and define

(wmin⁡)wmin⁡=min⁡2≤k≤n−1{wk}.(w_{\min}) \qquad w_{\min} = \min_{2 \le k \le n-1}\{w_k\}.

Assume that there exists a number κ\kappa such that the conditions

(H0) 0<κ<1wn0 < \kappa< \frac{1}{w_n} and wmin⁡>wn>1w_{\min} > w_n > 1;

and

(H1) wmin⁡−1wmin⁡−wn+4(1−wn)2κ2wn3≤nBn2\frac{w_{\min}-1}{w_{\min}-w_n}+\frac{4(1-w_n)^2}{\kappa^2w_n^3}\le nB_n^2

are satisfied.

If f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_kz^k with ∥f∥B≤1\lVert f\rVert_{\mathcal{B}}\le1 satisfies

n∣an∣2<4(1−wn)2κ2wn3∣a1∣2,(12)n|a_n|^2 < \frac{4(1-w_n)^2}{\kappa^2w_n^3}|a_1|^2, \tag*{(12)}

then

∑k=1nk∣ak∣2<nBn2.\sum_{k=1}^{n} k|a_k|^2 < nB_n^2.

Proof. In order to simplify the notation, write

yk=kwk∣ak∣,k≥1,(13)y_k=\sqrt{kw_k}|a_k|,\qquad k\ge1, \tag*{(13)}

so that

∑k=1nk∣ak∣2=∑k=1nyk2wk.\sum_{k=1}^{n} k|a_k|^2=\sum_{k=1}^{n}\frac{y_k^2}{w_k}.

By the definition of wmin⁡w_{\min}, we will have that

∑k=1nk∣ak∣2≤y12+yn2wn+1wmin⁡∑k=2n−1yk2.(14)\sum_{k=1}^{n} k|a_k|^2\le y_1^2+\frac{y_n^2}{w_n}+\frac{1}{w_{\min}}\sum_{k=2}^{n-1}y_k^2. \tag*{(14)}

Applying (11) at rn=n−1n+1r_n=\sqrt{\frac{n-1}{n+1}} we see that

∑k=2n−1yk2≤nwnBn2−y12−yn2.\sum_{k=2}^{n-1}y_k^2\le nw_nB_n^2-y_1^2-y_n^2.

Then, substituting in (14) gives

∑k=1nk∣ak∣2≤wnwmin⁡nBn2+wmin⁡−1wmin⁡y12+wmin⁡−wnwnwmin⁡yn2.\sum_{k=1}^{n} k|a_k|^2\le\frac{w_n}{w_{\min}}nB_n^2+\frac{w_{\min}-1}{w_{\min}}y_1^2+\frac{w_{\min}-w_n}{w_nw_{\min}}y_n^2.

Factoring out wmin⁡−wnwmin⁡\frac{w_{\min}-w_n}{w_{\min}} and using y1≤1y_1 \le1 together with (12) and (H0), we obtain the strict inequality

∑k=1nk∣ak∣2<wnwmin⁡nBn2+wmin⁡−wnwmin⁡(wmin⁡−1wmin⁡−wn+4(1−wn)2κ2wn3).\sum_{k=1}^{n} k|a_k|^2 < \frac{w_n}{w_{\min}}nB_n^2+\frac{w_{\min}-w_n}{w_{\min}}\left(\frac{w_{\min}-1}{w_{\min}-w_n}+\frac{4(1-w_n)^2}{\kappa^2w_n^3}\right).

Applying hypothesis (H1) guarantees that ∑k=1nk∣ak∣2<nBn2\sum_{k=1}^{n} k|a_k|^2<nB_n^2. □\square

We now turn to the complementary statement of Proposition 4 where the nn-th coefficient is dominant. To this end, we first need the following refinement of (11).

Lemma 5. Let r∈(0,1)r \in(0,1) and m≥1m \ge1 be an integer. If f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_kz^k and ∥f∥B≤1\|f\|_{\mathcal B}\le1, then

∑k=1∞k2∣ak∣2r2k−2+∣∑j=1∞j(j+m)aj+maj‾r2j+m−2∣≤1(1−r2)2.\sum_{k=1}^{\infty} k^2|a_k|^2r^{2k-2}+\left|\sum_{j=1}^{\infty}j(j+m)a_{j+m}\overline{a_j}r^{2j+m-2}\right|\le\frac{1}{(1-r^2)^2}.

Proof. Consider 0<r<10<r<1 and the real-valued function G(t)=∣f′(reit)∣2G(t)=|f'(re^{it})|^2, which is continuous on [0,2π][0,2\pi] and clearly satisfies

0≤G(t)≤1(1−r2)2,∀t∈[0,2π].0\le G(t)\le\frac{1}{(1-r^2)^2},\qquad\forall t\in[0,2\pi].

Writing G(t)=f′(reit)f′(reit)‾G(t)=f'(re^{it})\overline{f'(re^{it})} and taking into account that the power series of f′(z)f'(z) converges absolutely on {∣z∣=r}\{|z|=r\}, we obtain that the Fourier coefficients (cl)l∈Z(c_l)_{l\in\mathbb Z} of GG are given by

c0=∑k=1∞k2∣ak∣2r2k−2andcl=c−l‾=∑j=1∞j(j+l)aj+laj‾r2j+l−2,l≥1.c_0=\sum_{k=1}^{\infty}k^2|a_k|^2r^{2k-2}\quad\text{and}\quad c_l=\overline{c_{-l}}=\sum_{j=1}^{\infty}j(j+l)a_{j+l}\overline{a_j}r^{2j+l-2},\qquad l\ge1.

Now, fix φ∈R\varphi\in\mathbb R and consider the trigonometric polynomial Pφ(t)=1+cos⁡(mt−φ)P_\varphi(t)=1+\cos(mt-\varphi). Observe that PφP_\varphi is non-negative and, furthermore,

12π∫02πPφ(t) dt=1.\frac{1}{2\pi}\int_0^{2\pi}P_\varphi(t)\,dt=1.

On the one hand, we get

12π∫02πG(t)(1+cos⁡(mt−φ)) dt≤1(1−r2)2.(15)\frac{1}{2\pi}\int_0^{2\pi}G(t)(1+\cos(mt-\varphi))\,dt\le\frac{1}{(1-r^2)^2}. \tag*{(15)}

On the other hand, taking into account that

12π∫02πG(t)cos⁡(mt−φ) dt=Re⁡(eiφ12π∫02πG(t)e−imt dt)=Re⁡(cmeiφ),\frac{1}{2\pi}\int_0^{2\pi}G(t)\cos(mt-\varphi)\,dt=\operatorname{Re}\left(e^{i\varphi}\frac{1}{2\pi}\int_0^{2\pi}G(t)e^{-imt}\,dt\right)=\operatorname{Re}\left(c_me^{i\varphi}\right),

expressing the integral on the left-hand side of (15) via Fourier coefficients implies that

c0+Re⁡(cmeiφ)≤1(1−r2)2.c_0+\operatorname{Re}\left(c_me^{i\varphi}\right)\le\frac{1}{(1-r^2)^2}.

Choosing φ=−arg⁡(cm)\varphi=-\arg(c_m) we get Re⁡(cmeiφ)=∣cm∣\operatorname{Re}\left(c_me^{i\varphi}\right)=|c_m|. Substituting the explicit series representations of c0c_0 and cmc_m yields

∑k=1∞k2∣ak∣2r2k−2+∣∑j=1∞j(j+m)aj+maj‾r2j+m−2∣≤1(1−r2)2.\sum_{k=1}^{\infty} k^2|a_k|^2r^{2k-2}+\left|\sum_{j=1}^{\infty}j(j+m)a_{j+m}\overline{a_j}r^{2j+m-2}\right|\le\frac{1}{(1-r^2)^2}.

which completes the proof. □

Lemma 6. Let n≥2n \ge2. Consider the sequence (wk)k(w_k)_k defined in (wk)(w_k). For every f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_k z^k in B\mathcal{B} with ∥f∥B≤1\lVert f\rVert_{\mathcal{B}}\le1, we have

wn(nBn2−Φn(f))≥∑k=2n−1k(wk−wn)∣ak∣2+∑k=n+1∞kwk∣ak∣2+∣R∣−(wn−1)∣a1∣2,w_n\left(nB_n^2-\Phi_n(f)\right)\ge\sum_{k=2}^{n-1}k(w_k-w_n)|a_k|^2+\sum_{k=n+1}^{\infty}kw_k|a_k|^2+|R|-(w_n-1)|a_1|^2,

where R=∑j≥1j(j+n−1)aj+n−1aj‾rn2j+n−3R=\sum_{j\ge1}j(j+n-1)a_{j+n-1}\overline{a_j}r_n^{2j+n-3}.

Proof. Applying Lemma 5 for m=n−1m=n-1 at rn=n−1n+1r_n=\sqrt{\frac{n-1}{n+1}}, we have

∑k=1∞kwk∣ak∣2+∣R∣≤1(1−rn2)2=nwnBn2.\sum_{k=1}^{\infty}kw_k|a_k|^2+|R|\le\frac{1}{(1-r_n^2)^2}=nw_nB_n^2.

Subtracting wnΦn(f)w_n\Phi_n(f) from both sides yields the desired inequality since w1=1w_1=1. □

Similarly to Proposition 4, we will need to control the interaction between a1a_1, the dominant ana_n and a2n−1a_{2n-1}. Observe that this last coefficient does not play a role for Φn(f)\Phi_n(f) but it is present in the remainder RR introduced in Lemma 6. This task will require delicate estimates based on further technical hypotheses which we can safely assume when 6≤n≤96\le n\le9.

Proposition 7. Let n≥3n\ge3 and consider the sequence introduced in (wk)(w_k) and the quantity wmin⁡w_{\min} defined in (wmin⁡)(w_{\min}). Define

wmax⁡=max⁡2≤k≤n−1{wk}.w_{\max}=\max_{2\le k\le n-1}\{w_k\}.

Assume that there exists a number κ\kappa satisfying (H0) and the additional conditions

κwn2+κ2wn2wmax⁡4(wmin⁡−wn)<1.\frac{\kappa w_n}{2}+\frac{\kappa^2w_n^2w_{\max}}{4(w_{\min}-w_n)}<1.

and

κwn2+wn−1<1.\frac{\kappa w_n}{2}+w_n-1<1.

Suppose that f(z)=∑k≥0akzkf(z)=\sum_{k\ge0}a_kz^k with ∥f∥B≤1\lVert f\rVert_{\mathcal{B}}\le1 satisfies

n∣an∣2≥4(1−wn)2κ2wn3∣a1∣2.(16)n|a_n|^2\ge\frac{4(1-w_n)^2}{\kappa^2w_n^3}|a_1|^2. \tag*{(16)}

then

∑k=1nk∣ak∣2≤nBn2.\sum_{k=1}^{n}k|a_k|^2\le nB_n^2.

Furthermore, equality holds if and only if f(z)=λBnznf(z)=\lambda B_nz^n for a certain unimodular constant λ\lambda.

Proof. Recall Lemma 6:

wn(nBn2−Φn(f))≥∑k=2n−1k(wk−wn)∣ak∣2+∑k=n+1∞kwk∣ak∣2+∣R∣−(wn−1)∣a1∣2.w_n\left(nB_n^2-\Phi_n(f)\right)\ge\sum_{k=2}^{n-1}k(w_k-w_n)|a_k|^2+\sum_{k=n+1}^{\infty}kw_k|a_k|^2+|R|-(w_n-1)|a_1|^2.

If we can prove that the right-hand side of this inequality is non-negative, we are done.

Using the notation (13) and triangle inequality we get

∣R∣≥y1yn−yny2n−1−S,|R| \ge y_1 y_n-y_n y_{2n-1}-S,

where S=∑k=2n−1ykyk+n−1+∑k≥n+1ykyk+n−1S=\sum_{k=2}^{n-1}y_k y_{k+n-1}+\sum_{k\ge n+1}y_k y_{k+n-1}.

On the other hand, we will see that

∑k=2n−1k(wk−wn)∣ak∣2+∑k=n+1∞kwk∣ak∣2≥κwnS+(wn−1)y2n−12\sum_{k=2}^{n-1}k(w_k-w_n)|a_k|^2+\sum_{k=n+1}^{\infty}kw_k|a_k|^2\ge\kappa w_nS+(w_n-1)y_{2n-1}^2

which, in terms of the notation (13), is

∑k=2n−1wk−wnwkyk2+∑k=n+1∞yk2≥κwnS+(wn−1)y2n−12(17)\sum_{k=2}^{n-1}\frac{w_k-w_n}{w_k}y_k^2+\sum_{k=n+1}^{\infty}y_k^2\ge\kappa w_nS+(w_n-1)y_{2n-1}^2 \tag*{(17)}

We will prove this completing squares and using the hypotheses. We split the computations in four steps depending on the range of kk.

Firstly, observe that

∑k=2n−1wk−wnwkyk2−κwn∑k=2n−1ykyk+n−1=∑k=2n−1(wk−wnwkyk−κwn2wkwk−wnyk+n−1)2−∑k=2n−1κ2wn2wk4(wk−wn)yk+n−12≥−κ2wn2wmax⁡4(wmin⁡−wn)∑k=2n−1yk+n−12.(18)\begin{aligned} \sum_{k=2}^{n-1}\frac{w_k-w_n}{w_k}y_k^2-\kappa w_n\sum_{k=2}^{n-1}y_k y_{k+n-1} \\ &=\sum_{k=2}^{n-1}\left(\sqrt{\frac{w_k-w_n}{w_k}}y_k-\frac{\kappa w_n}{2}\sqrt{\frac{w_k}{w_k-w_n}}y_{k+n-1}\right)^2-\sum_{k=2}^{n-1}\frac{\kappa^2w_n^2w_k}{4(w_k-w_n)}y_{k+n-1}^2 \\ &\ge-\frac{\kappa^2w_n^2w_{\max}}{4(w_{\min}-w_n)}\sum_{k=2}^{n-1}y_{k+n-1}^2. \tag*{(18)} \end{aligned}

Secondly, using hypothesis (H2) and the change of index k=j+n−1k=j+n-1 we see that

∑k=n+12n−2yk2≥(κwn2+κ2wn2wmax⁡4(wmin⁡−wn))∑k=n+12n−2yk2=κwn2∑k=n+12n−2yk2+κ2wn2wmax⁡4(wmin⁡−wn)∑j=2n−1yj+n−12.(19)\sum_{k=n+1}^{2n-2}y_k^2\ge\left(\frac{\kappa w_n}{2}+\frac{\kappa^2w_n^2w_{\max}}{4(w_{\min}-w_n)}\right)\sum_{k=n+1}^{2n-2}y_k^2=\frac{\kappa w_n}{2}\sum_{k=n+1}^{2n-2}y_k^2+\frac{\kappa^2w_n^2w_{\max}}{4(w_{\min}-w_n)}\sum_{j=2}^{n-1}y_{j+n-1}^2. \tag*{(19)}

Thirdly, using (H3) we get

y2n−12≥(wn−1)y2n−12+κwn2y2n−12(20)y_{2n-1}^2\ge(w_n-1)y_{2n-1}^2+\frac{\kappa w_n}{2}y_{2n-1}^2 \tag*{(20)}

and (H0) and another change of index yield

∑k=2n∞yk2≥κwn∑k=2n∞yk2=κwn2∑k=2n∞yk2+κwn2∑j=n+1∞yj+n−12.(21)\sum_{k=2n}^{\infty}y_k^2\ge\kappa w_n\sum_{k=2n}^{\infty}y_k^2=\frac{\kappa w_n}{2}\sum_{k=2n}^{\infty}y_k^2+\frac{\kappa w_n}{2}\sum_{j=n+1}^{\infty}y_{j+n-1}^2. \tag*{(21)}

Finally, by combining all these inequalities, we obtain

∑k=2n−1wk−wnwkyk2+∑k=n+1∞yk2≥κwn∑k=2n−1ykyk+n−1+(wn−1)y2n−12+κwn2∑k=n+1∞(yk2+yk+n−12)≥κwn∑k=2n−1ykyk+n−1+(wn−1)y2n−12+κwn∑k=n+1∞ykyk+n−1\begin{aligned} \sum_{k=2}^{n-1}\frac{w_k-w_n}{w_k}y_k^2+\sum_{k=n+1}^{\infty}y_k^2\ge\kappa w_n\sum_{k=2}^{n-1}y_k y_{k+n-1}+(w_n-1)y_{2n-1}^2+\frac{\kappa w_n}{2}\sum_{k=n+1}^{\infty}(y_k^2+y_{k+n-1}^2) \\ &\ge\kappa w_n\sum_{k=2}^{n-1}y_k y_{k+n-1}+(w_n-1)y_{2n-1}^2+\kappa w_n\sum_{k=n+1}^{\infty}y_k y_{k+n-1} \end{aligned}

that yields (17). Therefore, using the non-negativity of ∣R∣|R|, y1=∣a1∣y_1=|a_1| and Lemma 13 in the Appendix, we obtain that

wn(nBn2−Φn(f))≥max⁡{0,y1yn−yny2n−1−S}+κwnS+(wn−1)(y2n−12−y12)≥κwnynmax⁡{0,y1−y2n−1}+(wn−1)(y2n−12−y12).\begin{aligned} w_n\left(nB_n^2-\Phi_n(f)\right)\ge\max\{0,y_1y_n-y_ny_{2n-1}-S\}+\kappa w_nS+(w_n-1)\left(y_{2n-1}^2-y_1^2\right) \\ &\ge\kappa w_ny_n\max\{0,y_1-y_{2n-1}\}+(w_n-1)\left(y_{2n-1}^2-y_1^2\right). \end{aligned}

Hence, if y2n−1≥y1y_{2n-1}\ge y_1,

wn(nBn2−Φn(f))≥(wn−1)(y2n−12−y12)≥0,w_n\left(nB_n^2-\Phi_n(f)\right)\ge(w_n-1)\left(y_{2n-1}^2-y_1^2\right)\ge0,

and if y2n−1<y1y_{2n-1}<y_1 we have instead that

wn(nBn2−Φn(f))≥κwnyn(y1−y2n−1)+(wn−1)(y2n−12−y12)=(κwnyn+(1−wn)(y2n−1+y1))(y1−y2n−1)>(κwnyn+2(1−wn)y1)(y1−y2n−1)≥0.\begin{aligned} w_n\left(nB_n^2-\Phi_n(f)\right)\ge\kappa w_ny_n(y_1-y_{2n-1})+(w_n-1)\left(y_{2n-1}^2-y_1^2\right) \\ &= \left(\kappa w_ny_n+(1-w_n)(y_{2n-1}+y_1)\right)(y_1-y_{2n-1}) \\ &> \left(\kappa w_ny_n+2(1-w_n)y_1\right)(y_1-y_{2n-1})\ge0. \end{aligned}

In the last inequality we have used (16).

It remains to characterize the equality case Φn(f)=nBn2\Phi_n(f)=nB_n^2. Suppose that Φn(f)=nBn2\Phi_n(f)=nB_n^2, so that wn(nBn2−Φn(f))=0w_n(nB_n^2-\Phi_n(f))=0. Since y2n−1<y1y_{2n-1}<y_1 gives a strict inequality above, we only need to analyze the case y2n−1≥y1y_{2n-1}\ge y_1. We split up this analysis into two parts

  • If y1>0y_1>0, since y2n−1≥y1>0y_{2n-1}\ge y_1>0 and the hypothesis (H3) is strict, we get that the inequality in (20) is actually strict. Therefore, wn(nBn2−Φn(f))>0w_n(nB_n^2-\Phi_n(f))>0.

  • If y1=0y_1=0, by the previous case we can suppose that y2n−1=0y_{2n-1}=0. Now, since hypotheses (H0) and (H2) are strict, we obtain the inequalities in (21) and (19) force yk=0y_k=0 for k≥n+1k\ge n+1. Then, analysing the squares in (18) we get yk=0y_k=0 for 2≤k≤n−12\le k\le n-1. Therefore, Φn(f)=nBn2\Phi_n(f)=nB_n^2 holds if and only if f(z)=λBnznf(z)=\lambda B_nz^n for a certain unimodular constant λ\lambda.

This finishes the proof. □

Proposition 8. For n∈{6,7,8,9}n\in\{6,7,8,9\} there exists κ=κ(n)\kappa=\kappa(n) such that the hypotheses (H0), (H1), (H2) and (H3) are satisfied.

Proof. To complete the proof, we summarize the required numerical values in the Table 1 and Table 2.

nnw1w_1w2w_2w3w_3w4w_4w5w_5w6w_6w7w_7w8w_8w9w_9wmax⁡w_{\max}wmin⁡−wnw_{\min}-w_n
61107\frac{10}{7}7549\frac{75}{49}500343\frac{500}{343}31252401\frac{3125}{2401}1875016807\frac{18750}{16807}7549\frac{75}{49}312516807\frac{3125}{16807}
7132\frac{3}{2}2716\frac{27}{16}2716\frac{27}{16}405256\frac{405}{256}729512\frac{729}{512}51034096\frac{5103}{4096}2716\frac{27}{16}7294096\frac{729}{4096}
81149\frac{14}{9}4927\frac{49}{27}1372729\frac{1372}{729}120056561\frac{12005}{6561}3361419683\frac{33614}{19683}823543531441\frac{823543}{531441}65883444782969\frac{6588344}{4782969}1372729\frac{1372}{729}8235434782969\frac{823543}{4782969}
9185\frac{8}{5}4825\frac{48}{25}256125\frac{256}{125}256125\frac{256}{125}61443125\frac{6144}{3125}2867215625\frac{28672}{15625}13107278125\frac{131072}{78125}589824390625\frac{589824}{390625}256125\frac{256}{125}35176390625\frac{35176}{390625}

Table 1. Values of (wj)1≤j≤n(w_j)_{1\le j\le n} for 6≤n≤96\le n\le9.

nnκ\kappa2(wn−1)κwn\frac{2(w_n-1)}{\kappa w_n}(H1)(H2)(H3)
612\frac{1}{2}38869375\frac{3886}{9375}1.77577<10.98061.77577 < 10.980630281253294172<1\frac{3028125}{3294172} < 11326133614<1\frac{13261}{33614} < 1
714\frac{1}{4}80565103\frac{8056}{5103}4.38177<12.84264.38177 < 12.842616176514194304<1\frac{1617651}{4194304} < 11315932768<1\frac{13159}{32768} < 1
814\frac{1}{4}1805375823543\frac{1805375}{823543}6.6811<14.70106.6811 < 14.701017302638433486784401<1\frac{1730263843}{3486784401} < 1292102531441<1\frac{292102}{531441} < 1
914\frac{1}{4}19919973728\frac{199199}{73728}11.4974<16.556811.4974 < 16.5568214468927488214697265625<1\frac{214468927488}{214697265625} < 1272927390625<1\frac{272927}{390625} < 1

Table 2. Verification of (H1), (H2) and (H3) for 6≤n≤96 \le n \le9.

Proof of Theorem 2. The result follows directly by combining Propositions 4, 7 and 8.

Remark 1. The proof technique presented here reaches its natural limit at n=9n = 9 and cannot be extended to n≥10n \ge10. Indeed, hypothesis (H0) strictly requires that wmin⁡>wnw_{\min} > w_n and, for example, for n=10n = 10 we have

w10−w2=10(911)9−1811≈0.0067.w_{10} - w_2 = 10\left(\frac{9}{11}\right)^9 - \frac{18}{11} \approx0.0067.

Generalized weighted Dirichlet functionals

In [3], the authors also investigated a generalized weighted truncated Dirichlet functional defined for any parameter t≥0t \ge0 by

Φnt(f)=∑k=1nkt∣ak∣2,\Phi_n^t(f) = \sum_{k=1}^{n} k^t |a_k|^2,

and analyzed the validity of the corresponding bound

Φnt(f)≤ntBn2(22)\Phi_n^t(f) \le n^t B_n^2 \tag*{(22)}

for all functions f(z)=∑k=0∞akzkf(z) = \sum_{k=0}^{\infty} a_k z^k in B\mathcal{B} satisfying ∥f∥B≤1\lVert f\rVert_{\mathcal{B}} \le1. As noted in [3], if t1≤t2t_1 \le t_2 then the validity of (22) for t1t_1 implies its validity for t2t_2. Hence, the inequality (22) holds unconditionally for all t≥2t \ge2 by applying (11), while for functions with non-negative Taylor coefficients it remains true for t≥1t \ge1 (see [3]). However, when t<1t < 1, the estimate fails for all sufficiently large nn, establishing t=1t = 1 as the critical threshold for this inequality among functions with non-negative coefficients.

The function considered in Theorem 1 to disprove inequality (3) for large nn is also suitable as a counterexample for the generalized inequality (22) when t<2t < 2.

Corollary 9. Let 1≤t<21 \le t < 2. Then, there exists nt≥2n_t \ge2 such that for every n≥ntn \ge n_t we have that

Φnt(f)>ntBn2.\Phi_n^t(f) > n^t B_n^2.

for a function f∈Bf \in\mathcal{B} with ∥f∥B=1\lVert f \rVert_{\mathcal{B}} = 1.

Proof. Let ntn_t be such that

n2−t(n−1n+1)n−1>4,n≥nt.n^{2-t}\left(\frac{n-1}{n+1}\right)^{n-1} > 4,\qquad n \ge n_t.

The index ntn_t certainly exists because the sequence is unbounded. Now, consider n≥ntn \ge n_t and fεf_\varepsilon the function defined in (4) and assume that 0<ε<εn0 < \varepsilon< \varepsilon_n, where εn\varepsilon_n is the constant defined in Lemma 3. Then, proceeding as in the proof of Theorem 1 we have that

Φnt(fε∥fε∥B)=ε2+ntBn21+16ε2(n+1)2=ntBn21+4n2−tε2(n+1)2(n−1n+1)n−11+16ε2(n+1)2>ntBn2,\Phi_n^t\left(\frac{f_\varepsilon}{\lVert f_\varepsilon\rVert_{\mathcal{B}}}\right) = \frac{\varepsilon^2+n^tB_n^2}{1+\frac{16\varepsilon^2}{(n+1)^2}} = n^tB_n^2\frac{1+\frac{4n^{2-t}\varepsilon^2}{(n+1)^2}\left(\frac{n-1}{n+1}\right)^{n-1}}{1+\frac{16\varepsilon^2}{(n+1)^2}} > n^tB_n^2,

which completes the proof.

Appendix A. Auxiliary lemmata

In this Appendix we include some technical results which are needed for the correct understanding of the work.

Lemma 10. Let n≥2n \ge2. Then the function g(x)=xn−12(1−x)g(x)=x^{\frac{n-1}{2}}(1-x), 0≤x≤10 \le x \le1, is strictly increasing in [0,n−1n+1]\left[0,\frac{n-1}{n+1}\right] and decreasing in [n−1n+1,1]\left[\frac{n-1}{n+1},1\right].

Proof. It is a straightforward consequence of the identity

g′(x)=xn−322(n−1−(n+1)x).g'(x)=\frac{x^{\frac{n-3}{2}}}{2}\left(n-1-(n+1)x\right).

Lemma 11. Let n≥4n \ge4. Consider the function

g(x)=(1+xn−1)(1−n−1n+1x),0≤x≤n+1n−1.g(x)=(1+x^{n-1})\left(1-\frac{n-1}{n+1}x\right),\qquad0 \le x \le\frac{n+1}{n-1}.

There exists 0<x0<10<x_0<1 such that g(x)≤g(1)=4n+1g(x)\le g(1)=\frac{4}{n+1} for all x∈[x0,n+1n−1]x\in\left[x_0,\frac{n+1}{n-1}\right].

Remark 2. Using the change of variables r2=n−1n+1xr^2=\frac{n-1}{n+1}x, r>0r>0, the above lemma shows that there exists r0<n−1n+1=rnr_0<\sqrt{\frac{n-1}{n+1}}=r_n such that

(1+(n+1n−1r2)n−1)(1−r2)≤(1+(n+1n−1rn2)n−1)(1−rn2)=4n+1,r0≤r≤1.\left(1+\left(\frac{n+1}{n-1}r^2\right)^{n-1}\right)(1-r^2) \le \left(1+\left(\frac{n+1}{n-1}r_n^2\right)^{n-1}\right)(1-r_n^2) =\frac{4}{n+1},\qquad r_0\le r\le1.

Proof. A direct computation shows that

g′(x)=n−1n+1((n+1)xn−2−nxn−1−1)andg′′(x)=n−1n+1xn−3((n+1)(n−2)−n(n−1)x).g'(x)=\frac{n-1}{n+1}\left((n+1)x^{n-2}-nx^{n-1}-1\right) \quad\text{and}\quad g''(x)=\frac{n-1}{n+1}x^{n-3}\left((n+1)(n-2)-n(n-1)x\right).

Let x1=(n+1)(n−2)n(n−1)x_1=\frac{(n+1)(n-2)}{n(n-1)} be the unique zero of g′′g''. Observe that x1<1x_1<1 because of the fact that (n+1)(n−2)=n2−n−2<n(n−1)(n+1)(n-2)=n^2-n-2<n(n-1).

Since g′g' is strictly decreasing in (x1,n+1n−1]\left(x_1,\frac{n+1}{n-1}\right] and g′(1)=0g'(1)=0, we have that g′(x1)>0g'(x_1)>0. But g′(0)<0g'(0)<0, so Bolzano’s theorem yields that there exists x2∈(0,x1)x_2\in(0,x_1) such that g′(x2)=0g'(x_2)=0. Given the monotonicity of g′g' on that interval, x2x_2 must be the unique root there.

Summing up, we deduced that gg is decreasing on [0,x2][0,x_2] and on [1,n+1n−1]\left[1,\frac{n+1}{n-1}\right]; and increasing on [x2,1][x_2,1]. Hence

g(x2)<g(1)=4n+1≤1=g(0).g(x_2)<g(1)=\frac{4}{n+1}\leq1=g(0).

Again, Bolzano’s intermediate values theorem yields that there exists a (unique) x0∈(0,x2)x_0\in(0,x_2) such that g(x0)=4n+1g(x_0)=\frac{4}{n+1}. Given the monotonicity of gg, we will have that

max⁡x0≤x≤n+1n−1{g(x)}=max⁡{g(x0),g(1)}=4n+1.\max_{x_0\leq x\leq\frac{n+1}{n-1}}\{g(x)\} =\max\{g(x_0),g(1)\} =\frac{4}{n+1}.

Lemma 12. Consider the polynomial

gε(ζ)=14ε+14ζ−14εζ2+εζ3.g_\varepsilon(\zeta)=14\varepsilon+14\zeta-14\varepsilon\zeta^2+\varepsilon\zeta^3.

Then there exists 0<R1<1<R2<(1413)130<R_1<1<R_2<\left(\frac{14}{13}\right)^{13} and ε0=ε0(R1,R2)\varepsilon_0=\varepsilon_0(R_1,R_2) such that

∣gε(ζ)∣2(1−1314∣ζ∣13)2<1+27(1314)26ε2,R1≤∣ζ∣≤R2,(23)|g_\varepsilon(\zeta)|^2\left(1-\frac{13}{14}\sqrt[13]{|\zeta|}\right)^2 <1+27\left(\frac{13}{14}\right)^{26}\varepsilon^2,\qquad R_1\leq|\zeta|\leq R_2, \tag*{(23)}

when 0<ε<ε00<\varepsilon<\varepsilon_0.

Proof. On the one hand, using the second order Taylor expansion at x=1x=1 of the function

x↦142x2(1−1314x13)2,x\mapsto14^2x^2\left(1-\frac{13}{14}\sqrt[13]{x}\right)^2,

we see that

142∣ζ∣2(1−1314∣ζ∣13)2=1−1413(1−∣ζ∣)2+O(∣1−∣ζ∣∣3).(24)14^2|\zeta|^2\left(1-\frac{13}{14}\sqrt[13]{|\zeta|}\right)^2 =1-\frac{14}{13}(1-|\zeta|)^2+O\left(|1-|\zeta||^3\right). \tag*{(24)}

In particular, there exists 0<R1<1<R2<(1413)130<R_1<1<R_2<\left(\frac{14}{13}\right)^{13} such that

142∣ζ∣2(1−1314∣ζ∣13)2≤1−(1−∣ζ∣)2,R1≤∣ζ∣≤R2.(25)14^2|\zeta|^2\left(1-\frac{13}{14}\sqrt[13]{|\zeta|}\right)^2 \leq1-(1-|\zeta|)^2,\qquad R_1\leq|\zeta|\leq R_2. \tag*{(25)}

By shrinking [R1,R2][R_1,R_2] if necessary, we may assume further that

∣1∣ζ∣−∣ζ∣∣≤3∣1−∣ζ∣∣,(26)\left|\frac{1}{|\zeta|}-|\zeta|\right|\le3|1-|\zeta||, \tag*{(26)}
∣(1∣ζ∣+∣ζ∣)2+4−3∣ζ∣27∣≤8,(27)\left|\left(\frac{1}{|\zeta|}+|\zeta|\right)^2+4-\frac{3|\zeta|^2}{7}\right|\le8, \tag*{(27)}
∣(1∣ζ∣+∣ζ∣)2−∣ζ∣27∣≤274((1314)26+37)(28)\left|\left(\frac{1}{|\zeta|}+|\zeta|\right)^2-\frac{|\zeta|^2}{7}\right| \le\frac{27}{4}\left(\left(\frac{13}{14}\right)^{26}+\frac{3}{7}\right) \tag*{(28)}

and

∣2∣ζ∣27−4+1314(1∣ζ∣+1)2∣≤274((1314)26−17)(29)\left|\frac{2|\zeta|^2}{7}-4+\frac{13}{14}\left(\frac{1}{|\zeta|}+1\right)^2\right| \le\frac{27}{4}\left(\left(\frac{13}{14}\right)^{26}-\frac{1}{7}\right) \tag*{(29)}

when R1≤∣ζ∣≤R2R_1\le|\zeta|\le R_2.

On the other hand, if ζ=∣ζ∣eit\zeta=|\zeta|e^{it}, then

∣gϵ(ζ)∣2142∣ζ∣2=∣1+(1ζ−ζ)ϵ+ϵ2ζ214∣=1+((1∣ζ∣+∣ζ∣)2−∣ζ∣27)ϵ2+2(1∣ζ∣−∣ζ∣)ϵcos⁡t+(2∣ζ∣27−4)ϵ2cos⁡2t+O(ϵ3),(30)\begin{aligned} \frac{|g_\epsilon(\zeta)|^2}{14^2|\zeta|^2} = \left|1+\left(\frac{1}{\zeta}-\zeta\right)\epsilon+\frac{\epsilon^2\zeta^2}{14}\right| \\ &=1+\left(\left(\frac{1}{|\zeta|}+|\zeta|\right)^2-\frac{|\zeta|^2}{7}\right)\epsilon^2 +2\left(\frac{1}{|\zeta|}-|\zeta|\right)\epsilon\cos t +\left(\frac{2|\zeta|^2}{7}-4\right)\epsilon^2\cos^2t+O(\epsilon^3), \tag*{(30)} \end{aligned}

where the big-OO constant depends only on R1R_1 and R2R_2.

We will split the proof in two parts:

  • If ∣1−∣ζ∣∣≥8ϵ|1-|\zeta||\ge8\epsilon, then

(1−(1−∣ζ∣)2)(1+8ϵ2+6ϵ∣1−∣ζ∣∣)≤(1−64ϵ2)(1+56ϵ2)<1,(1-(1-|\zeta|)^2)(1+8\epsilon^2+6\epsilon|1-|\zeta||) \le(1-64\epsilon^2)(1+56\epsilon^2)<1,

because the polynomial

x↦(1−x2)(1+8ϵ2+6ϵx)x\mapsto(1-x^2)(1+8\epsilon^2+6\epsilon x)

has negative derivative in 8ϵ<x8\epsilon<x. Thus, we will have

∣gϵ(ζ)∣2(1−1314∣ζ∣13)2<1+O(ϵ3),8ϵ≤∣1−∣ζ∣∣|g_\epsilon(\zeta)|^2\left(1-\frac{13}{14}\sqrt[13]{|\zeta|}\right)^2<1+O(\epsilon^3),\qquad8\epsilon\le|1-|\zeta||

by (25), (26), (27) and (30).

  • If ∣1−∣ζ∣∣<8ϵ|1-|\zeta||<8\epsilon, the identity

2(1∣ζ∣−∣ζ∣)ϵcos⁡t−1413(1−∣ζ∣)2=1314(1∣ζ∣+1)2ϵ2cos⁡2t−(1413(1−∣ζ∣)−1314(1∣ζ∣+1)ϵcos⁡t)22\left(\frac{1}{|\zeta|}-|\zeta|\right)\epsilon\cos t-\frac{14}{13}(1-|\zeta|)^2 = \frac{13}{14}\left(\frac{1}{|\zeta|}+1\right)^2\epsilon^2\cos^2t -\left(\sqrt{\frac{14}{13}}(1-|\zeta|)-\sqrt{\frac{13}{14}}\left(\frac{1}{|\zeta|}+1\right)\epsilon\cos t\right)^2

can be used to deduce via (28) and (29) that

1+((1∣ζ∣+∣ζ∣)2−∣ζ∣27)ε2+2(1∣ζ∣−∣ζ∣)εcos⁡t+(2∣ζ∣27−4)ε2cos⁡2t−1413(1−∣ζ∣)21+\left(\left(\frac{1}{|\zeta|}+|\zeta|\right)^2-\frac{|\zeta|^2}{7}\right)\varepsilon^2+2\left(\frac{1}{|\zeta|}-|\zeta|\right)\varepsilon\cos t+\left(\frac{2|\zeta|^2}{7}-4\right)\varepsilon^2\cos^2 t-\frac{14}{13}(1-|\zeta|)^2
≤1+((1∣ζ∣+∣ζ∣)2−∣ζ∣27)ε2+(2∣ζ∣27−4+1314(1∣ζ∣+1)2)ε2cos⁡2t\le1+\left(\left(\frac{1}{|\zeta|}+|\zeta|\right)^2-\frac{|\zeta|^2}{7}\right)\varepsilon^2+\left(\frac{2|\zeta|^2}{7}-4+\frac{13}{14}\left(\frac{1}{|\zeta|}+1\right)^2\right)\varepsilon^2\cos^2 t
≤1+272((1314)26+17)ε2.\le1+\frac{27}{2}\left(\left(\frac{13}{14}\right)^{26}+\frac{1}{7}\right)\varepsilon^2.

Combining this with (24) (recall that ∣1−∣ζ∣∣=O(ε)\lvert1-|\zeta|\rvert=O(\varepsilon) in this range) and (30), we conclude that

∣gε(ζ)∣2(1−1314∣ζ∣13)2≤1+272((1314)26+17)ε2+O(ε3).|g_{\varepsilon}(\zeta)|^2\left(1-\frac{13}{14}\sqrt[13]{|\zeta|}\right)^2\le1+\frac{27}{2}\left(\left(\frac{13}{14}\right)^{26}+\frac{1}{7}\right)\varepsilon^2+O(\varepsilon^3).

In any case, since 272(1314)26+17<27(1314)26\frac{27}{2}\left(\frac{13}{14}\right)^{26}+\frac{1}{7}<27\left(\frac{13}{14}\right)^{26}, we may choose ε0\varepsilon_0 depending only on R1R_1 and R2R_2 such that (23) holds for all 0<ε<ε00<\varepsilon<\varepsilon_0. □

Remark 3. Using the change of variables ζ=(1413z2)13\zeta=\left(\frac{14}{13}z^2\right)^{13}, the above lemma shows that the function fεf_\varepsilon defined in (10) satisfies that

max⁡r1≤∣z∣≤r2{∣fε′(z)∣2(1−∣z∣2)2}<1+27(1314)26ε2\max_{r_1\le|z|\le r_2}\{|f_\varepsilon'(z)|^2(1-|z|^2)^2\}<1+27\left(\frac{13}{14}\right)^{26}\varepsilon^2

if 0<ε<ε00<\varepsilon<\varepsilon_0 for some 0<r1<1314<r2<10<r_1<\sqrt{\frac{13}{14}}<r_2<1.

Lemma 13. Let x∈Rx\in\mathbb{R}, C≥0C\ge0 and 0<β<10<\beta<1. Then

max⁡{0,x−C}+βC≥βmax⁡{x,0}.\max\{0,x-C\}+\beta C\ge\beta\max\{x,0\}.

Proof. If x≤0x\le0 the right hand side is 0. If 0<x<C0<x<C, the inequality is reduced to βC≥βx\beta C\ge\beta x. Finally, if x≥Cx\ge C, then x−C+βC≥β(x−C)+βC=βxx-C+\beta C\ge\beta(x-C)+\beta C=\beta x. □

References

Departamento de Matemáticas, Universidad Autónoma de Madrid, 28049 Madrid, Spain Email address: [email protected]

Departamento de Matemáticas, Universidad Autónoma de Madrid, 28049 Madrid, Spain Email address: [email protected]

Departamento de Matemáticas, Universidad de Alicante, 03690 Sant Vicent del Raspeig, Spain Email address: [email protected]

References

  1. [1]J. M. Anderson, J. Clunie, and Ch. Pommerenke, On Bloch functions and normal functions, J. Reine Angew. Math. 270 (1974), 12–37. MR 361090DOI
  2. [2]J. Arazy and S. D. Fisher, The uniqueness of the Dirichlet space among Möbius-invariant Hilbert spaces, Illinois J. Math. 29 (1985), no. 3, 449–462. MR 786732DOI
  3. [3]I. Efraimidis, A. Mas, and D. Vukotić, Estimates for truncated area functionals on the Bloch space, Proc. Amer. Math. Soc. 151 (2023), no. 9, 3845–3854. MR 4607629arxiv.org/abs/2208.10626
  4. [4]O. El-Fallah, K. Kellay, J. Mashreghi, and T. Ransford, A primer on the Dirichlet space, Cambridge Tracts in Mathematics, vol. 203, Cambridge University Press, Cambridge, 2014. MR 3185375
  5. [5]I. R. Kayumov, A note on an area-type functional of Bloch functions, Lobachevskii J. Math. 38 (2017), no. 3, 466–468. MR 3659227
  6. [6]I. R. Kayumov and K.-J. Wirths, Coefficient inequalities for Bloch functions, Lobachevskii J. Math. 40 (2019), no. 9, 1319–1323. MR 4021521DOI
  7. [7]————, Coefficients problems for Bloch functions, Anal. Math. Phys. 9 (2019), no. 3, 1069–1085. MR 4014857DOI
  8. [8]————, On the sum of squares of the coefficients of Bloch functions, Monatsh. Math. 190 (2019), no. 1, 123–135. MR 3998335arxiv.org/abs/1904.10846
  9. [9]______, Inequalities of Carlson type for α-Bloch functions, Mediterr. J. Math. 17 (2020), no. 3, Paper No. 83, 9. MR 4099644
  10. [10]R. Khasianov, On the weighted sum of squares of the coefficients of Bloch functions, Anal. Math. Phys. 16 (2026), no. 1, Paper No. 13, 17. MR 5013332
  11. [11]L. A. Rubel and R. M. Timoney, An extremal property of the Bloch space, Proc. Amer. Math. Soc. 75 (1979), no. 1, 45–49. MR 0529210DOI
  12. [12]K.-J. Wirths, Über holomorphe Funktionen, die einer Wachstumsbeschränkung unterliegen, Arch. Math. (Basel) 30 (1978), no. 6, 606–612. MR 492223

Paper details

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

views · 7

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