Analysis on Nam Le

Inequality for Square-Summable Complex Series

Thu, 28 May 2026 00:00:00 +0000

Some inequalities look formidable until the right decomposition makes them transparent. The conjecture below, posed by Zoltan Retkes on the Open Problem Garden in 2012 with a £10 prize attached, is one such case: once the dyadic structure of the positive integers is made explicit, the proof reduces to two classical facts.

Conjecture (Retkes, 2012), now proved

For all $\alpha = (\alpha_1, \alpha_2, \ldots) \in \ell^2(\mathbb{C})$, $$\sum_{n \geq 1} |\alpha_n|^2 \geq \frac{6}{\pi^2} \sum_{k \geq 0} \left|, \sum_{l \geq 0} \frac{\alpha_{2^k(2l+1)}}{l+1} ,\right|^2.$$

The conjecture was confirmed by an anonymous comment on the problem page in November 2013. A self-contained proof and an extension to $\ell^p$ were subsequently published by Ibragimov and Salimova in Elemente der Mathematik 70 (2015), 79–81.

The Dyadic Decomposition #

The index $2^k(2l+1)$ running over $k \geq 0$ and $l \geq 0$ is not arbitrary: it encodes a canonical partition of the positive integers. Every $n \in \mathbb{N}^+$ factors uniquely as $$n = 2^k \cdot r, \qquad k \geq 0,\quad r \text{ odd positive},$$ where $k = v_2(n)$ is the 2-adic valuation of $n$ and $r = n/2^k$ is its odd part. Writing $r = 2l+1$ gives the bijection $\mathbb{N}_0 \times \mathbb{N}_0 \to \mathbb{N}^+$, $(k, l) \mapsto 2^k(2l+1)$. In particular the sets $$A_k = {2^k(2l+1) : l \geq 0} = {2^k, 3 \cdot 2^k, 5 \cdot 2^k, \ldots}$$ form a partition of $\mathbb{N}^+$. Explicitly: $A_0 = {1, 3, 5, 7, \ldots}$ (odd numbers), $A_1 = {2, 6, 10, 14, \ldots}$ (twice an odd number), and so on. This partition is the key structural fact behind the proof.

Proof #

The argument has two ingredients: the Basel sum $\sum_{l \geq 0}(l+1)^{-2} = \pi^2/6$, and the Cauchy–Schwarz inequality in $\ell^2(\mathbb{C})$.

Define two sequences in $\ell^2(\mathbb{C})$: $$x = \left(1,, \tfrac{1}{2},, \tfrac{1}{3},, \ldots\right), \qquad y_k = \left(\alpha_{2^k},, \alpha_{3 \cdot 2^k},, \alpha_{5 \cdot 2^k},, \ldots\right) \quad (k \geq 0).$$

The inner sum in the conjecture is exactly the $\ell^2$ inner product $\langle x, y_k \rangle$: $$\sum_{l \geq 0} \frac{\alpha_{2^k(2l+1)}}{l+1} = \langle x, y_k \rangle.$$

Step 1: Apply Cauchy–Schwarz. For each $k$,

$$|\langle x, y_k \rangle|^2 \leq |x|_2^2 \cdot |y_k|_2^2.$$

Summing over $k \geq 0$,

$$\sum _{k \geq 0} |\langle x, y _k \rangle|^2 \leq |x| _2^2 \sum _{k \geq 0} |y _k| _2^2.$$

Step 2: Evaluate using the Basel problem and the partition. The Basel problem gives $$|x| _2^2 = \sum _{l \geq 0} \frac{1}{(l+1)^2} = \frac{\pi^2}{6}.$$

Since the sets $A_k$ partition $\mathbb{N}^+$, $$\sum _{k \geq 0} |y_k|_2^2 = \sum _{k \geq 0} \sum _{l \geq 0} |\alpha _{2^k(2l+1)}|^2 = \sum _{n \geq 1} |\alpha_n|^2.$$

Combining both steps, $$\sum_{k \geq 0} \left|\sum_{l \geq 0} \frac{\alpha_{2^k(2l+1)}}{l+1}\right|^2 \leq \frac{\pi^2}{6} \sum_{n \geq 1} |\alpha_n|^2,$$ which is the inequality with the $\frac{6}{\pi^2}$ factor moved to the other side.

Sharpness of the Constant #

The constant $6/\pi^2$ is the best possible. To see this, consider the truncated sequence $\alpha^{(N)}$ defined by $\alpha^{(N)}_{2l+1} = 1/(l+1)$ for $l = 0, 1, \ldots, N-1$ and $\alpha^{(N)}_n = 0$ otherwise. Then:

The left-hand side equals $\displaystyle\sum_{l=0}^{N-1} \frac{1}{(l+1)^2} \to \frac{\pi^2}{6}$.
The only non-zero contribution to the right-hand side comes from $k = 0$ (since all non-zero indices are odd, i.e. in $A_0$), giving $\displaystyle\frac{6}{\pi^2}\left(\sum_{l=0}^{N-1} \frac{1}{(l+1)^2}\right)^2 \to \frac{6}{\pi^2} \cdot \frac{\pi^4}{36} = \frac{\pi^2}{6}$.

The ratio of the right-hand side to the left-hand side therefore tends to $1$ as $N \to \infty$, so no larger constant than $6/\pi^2$ can hold universally. Equality is never achieved for $\alpha \in \ell^2(\mathbb{C})\setminus{0}$ with finite norm since the limiting sequence does not belong to $\ell^2(\mathbb{C})$.

Extension to $\ell^p$ #

The Cauchy–Schwarz inequality used above is a special case of Hölder’s inequality, and the proof generalises immediately.

Theorem (Ibragimov–Salimova, 2015)

Let $p, q \in (1,\infty)$ with $\tfrac{1}{p} + \tfrac{1}{q} = 1$. For all $\alpha = (\alpha_1, \alpha_2, \ldots) \in \ell^p(\mathbb{C})$ and $x = (x_0, x_1, \ldots) \in \ell^q(\mathbb{C})$, $$\sum_{n \geq 1} |\alpha_n|^p \geq \left(\sum_{l \geq 0} |x_l|^q\right)^{-p/q} \sum_{k \geq 0} \left|\sum_{l \geq 0} x_l, \alpha_{2^k(2l+1)}\right|^p.$$

Retkes’s original inequality is the case $p = q = 2$ and $x_l = 1/(l+1)$, where $(\sum_{l\geq 0}|x_l|^2)^{-1} = 6/\pi^2$ by the Basel problem.

Remarks on Structure #

The role of the dyadic partition. The sets $A_k$ are the dyadic layers of $\mathbb{N}^+$: each integer sits in exactly one layer determined by its 2-adic valuation. This structure also appears in the theory of Hardy spaces, where the dyadic martingale decomposition underpins the $H^1$–BMO duality, and in wavelets, where the dyadic scaling of the real line organises the multiresolution analysis. The inequality can be read as a norm comparison between the $\ell^2$ norm and a weighted sum over dyadic layers.

Relation to the Basel problem. The constant $6/\pi^2$, the reciprocal of $\zeta(2)$, appears here because the weight sequence $1/(l+1)$ used in the inner sum is precisely the harmonic sequence, whose $\ell^2$ norm squared is $\zeta(2)$. Any other weight sequence $x \in \ell^2(\mathbb{C})$ would produce the analogous inequality with $|x|_2^{-2}$ in place of $6/\pi^2$.

The inequality as a rearrangement estimate. The right-hand side reorganises the entries of $\alpha$ by their dyadic layer and applies a weighted average within each layer. The inequality says the total $\ell^2$ energy cannot be less than $6/\pi^2$ times the energy of this rearranged, averaged version of the sequence, a quantitative statement about how averaging destroys energy.

Further Questions #

While the original conjecture is settled, several natural variants remain.

Question 1

What is the sharp constant in the inequality if the dyadic partition is replaced by the partition induced by a prime $p \neq 2$, i.e. by the sets $A_k^{(p)} = {p^k m : \gcd(m, p) = 1}$? The same argument applies with $x_l = w_l$ for any weight sequence $w \in \ell^2(\mathbb{C})$, but the resulting constant depends on $|w|_2$ and the choice of weight, not on $\pi$.

Question 2

The inner sum $\sum_{l \geq 0} \alpha_{2^k(2l+1)}/(l+1)$ averages the entries in layer $A_k$ with the harmonic weights. What happens if the harmonic weight $1/(l+1)$ is replaced by a weight $w(l)$ depending on the position $l$ within the layer in a more general way, for instance $w(l) = l^{-s}$ for $s > 1/2$? The sharp constant would then involve $\zeta(2s)$ instead of $\zeta(2) = \pi^2/6$.

Question 3

For $p = 1$ the Ibragimov–Salimova theorem requires $q = \infty$, and the Hölder inequality takes a different form. Does an analogue of Retkes’s inequality hold for $\alpha \in \ell^1(\mathbb{C})$, and if so, what is the sharp constant?

References #

Ibragimov, Z. O. & Salimova, D. F. (2015). On an inequality in $\ell_p(\mathbb{C})$ involving Basel problem. Elemente der Mathematik, 70(2), 79–81. https://ems.press/content/serial-article-files/45532
Retkes, Z. (2012). Inequality for square summable complex series. Open Problem Garden. http://www.openproblemgarden.org/op/inequality_for_square_summable_complex_series
Benko, D. & Molokach, J. (2013). The Basel problem as a rearrangement of series. College Mathematics Journal, 44(3), 171–176.
Ritelli, D. (2013). Another proof of $\zeta(2) = \pi^2/6$ using double integrals. American Mathematical Monthly, 120(7), 642–645.

Something Like Picard for 1-Forms

Wed, 27 May 2026 00:00:00 +0000

Picard’s great theorem is a statement about how wildly a holomorphic function can behave near an essential singularity. The conjecture below asks whether injectivity of local primitives of a 1-form is enough to rule out such wild behaviour at the origin, forcing the 1-form to extend meromorphically across the puncture.

Conjecture (Elsner, 2010)

Let $D$ be the open unit disk and let $U_1,\dots,U_n$ be open sets with $\bigcup_{j=1}^n U_j = D\setminus{0}$. Suppose there are injective holomorphic functions $f_j : U_j \to \mathbb{C}$ such that $$\mathrm{d}f_j = \mathrm{d}f_k \quad \text{on every connected component of } U_j \cap U_k.$$ Then the $\mathrm{d}f_j$ glue together to a meromorphic 1-form on $D$.

The problem is rated medium importance on the Open Problem Garden and is not recommended for undergraduates, reflecting the depth of the tools involved. It arises from Elsner’s study of hyperelliptic action integrals in the context of the exact WKB method for Schrödinger equations with polynomial potential (Elsner, Ann. Inst. Fourier 49(1), 1999).

Setup and Interpretation #

The compatibility condition $\mathrm{d}f_j = \mathrm{d}f_k$ on each connected component of $U_j \cap U_k$ is equivalent to saying $f_j - f_k$ is locally constant there. The local differentials therefore glue together unambiguously to a global holomorphic 1-form $$\omega \in \Omega^1(D\setminus{0})$$ whose restriction to each $U_j$ equals $\mathrm{d}f_j$. The conjecture asserts that $\omega$ does not have an essential singularity at the origin: it extends to a meromorphic 1-form on all of $D$, meaning near $0$ it looks like $$\omega = \left(\frac{c_{-m}}{z^m} + \cdots + \frac{c_{-1}}{z} + c_0 + c_1 z + \cdots\right)dz$$ for some $m \ge 0$.

The injectivity of each $f_j$ is the crucial hypothesis. Without it the statement is false: any holomorphic 1-form $\omega$ on $D\setminus{0}$ with an essential singularity at $0$ is locally $\mathrm{d}f_j$ for some holomorphic $f_j$, and these $f_j$ can be chosen on contractible pieces of the cover; injectivity is what prohibits essential singularities from arising.

What Is Already Known #

Partial Result

Under the hypotheses of the conjecture:

The 1-form $\omega$ is holomorphic on $D\setminus{0}$.
If the residue of $\omega$ at the origin vanishes, Picard’s big theorem can be applied to conclude that $\omega$ extends meromorphically across $0$.

Point (1) is straightforward: each $\mathrm{d}f_j$ is holomorphic on $U_j$ and the local forms agree on overlaps, so $\omega$ is holomorphic wherever it is defined, i.e. on $D\setminus{0}$.

Point (2) is the key partial result recorded by Elsner. If $\operatorname{Res}_0\omega = 0$, then $\omega$ has trivial monodromy around the origin and admits a single-valued holomorphic primitive $F$ on the punctured disk: $\omega = \mathrm{d}F$. The injectivity of each local branch $f_j$ then forces $F$ itself to be injective on some punctured neighbourhood of $0$ (since $f_j = F + c$ locally). An injective holomorphic function on a punctured disk cannot have an essential singularity there, and this is where Picard enters: at an essential singularity, by Picard’s big theorem, every value is taken infinitely often in any punctured neighbourhood, contradicting injectivity. Hence $F$ has at most a pole at $0$, and $\omega = \mathrm{d}F$ is meromorphic.

The open case is when $\operatorname{Res}_0\omega \ne 0$, so that $\omega$ has non-trivial monodromy and no single-valued global primitive exists. The local primitives $f_j$ then experience monodromy as one loops around the origin, and the injectivity constraint must be leveraged in this more delicate multi-valued setting.

Connection to Picard’s Theorem #

The title of the conjecture reflects a precise structural analogy.

Theorem (Picard's Great Theorem)

If $f$ has an essential singularity at $z_0$, then in every punctured neighbourhood of $z_0$ the function $f$ takes every value in $\mathbb{C}$, with at most one exception, infinitely many times.

In particular, a function with an essential singularity is far from injective near that point. The conjecture elevates this observation to the level of 1-forms: an injective holomorphic primitive should preclude essential singularities in the 1-form itself, even when the primitive is only locally and multi-valuedly defined.

Standard Picard covers the zero-residue case by reducing to a single-valued primitive. The conjecture asks for an analogue that works when the monodromy is non-trivial, a genuinely new statement about multi-valued functions and their differential geometry.

Origin: Hyperelliptic Action Integrals #

The problem arises from the exact WKB method applied to the stationary Schrödinger equation $-\psi’’ + V(x)\psi = E\psi$ with polynomial potential $V$. The formal WKB ansatz $\psi \sim e^{S/\hbar}$ produces a multivalued action integral $$\mathcal{I}(E) = \int_\gamma \sqrt{V(x) - E}\mathrm{d}x$$ defined on a hyperelliptic Riemann surface whose branch structure depends on the energy parameter $E$. Elsner’s 1999 paper constructs the Riemann surface of $\mathcal{I}$ explicitly and shows its branch points accumulate densely in the value plane, a phenomenon that obstructs Borel–Laplace resummation of the WKB symbols.

In this setting the local inverses of $\mathcal{I}$ play the role of the $f_j$: they are locally injective holomorphic functions whose differentials agree on overlaps. The conjecture asks whether the obstruction to global meromorphic extension can arise only from a pole, a controlled singularity, rather than an essential one.

Research Directions #

1. The Non-Zero Residue Case #

The open heart of the problem is the case $\operatorname{Res}_0\omega \ne 0$. Here $\omega$ is not exact near $0$, the monodromy of the primitive is a non-trivial translation $f_j \mapsto f_j + 2\pi i, \operatorname{Res}_0\omega$, and no single injective function encompasses the full behaviour near the singularity.

A natural approach is to pass to a cyclic cover $\tilde D \to D$ that trivialises the monodromy, construct a single-valued primitive on $\tilde D\setminus{0}$, and then appeal to the zero-residue argument there. The key difficulty is that the injectivity of each $f_j$ on $U_j$ does not immediately imply injectivity of the lifted primitive on $\tilde D$, since different sheets can collide. Making this argument precise, or finding a counterexample, is the main open problem.

2. Quantitative Control via Nevanlinna Theory #

An alternative strategy replaces Picard’s theorem by its quantitative form. If $F$ is a meromorphic function on the punctured disk with an essential singularity, the Nevanlinna characteristic $T(r,F)$ grows faster than any power of $\log(1/r)$ as $r\to 0$. For an injective function the counting functions $N(r,a,F)$, recording how often $F = a$ in the punctured disk, satisfy strong constraints.

Nevanlinna-theoretic methods might give a direct bound on $T(r,f_j)$ in terms of the geometry of the cover ${U_j}$ and the injectivity of $f_j$, ruling out essential singularities of $\omega$ without passing through the monodromy argument. This would require adapting the standard Nevanlinna machinery to functions that are only locally defined on an open cover.

3. Replacing Injectivity by Finite Valence #

One can ask whether the conjecture remains true if “injective” is weakened to “at most $d$-to-one” for some fixed integer $d$. Finite-valence holomorphic functions cannot have essential singularities either, by a Picard-type argument (a function of valence at most $d$ takes each value at most $d$ times, so in any neighbourhood of an essential singularity it must omit a set of positive capacity, contradicting Picard).

If the conjecture extends to finite valence, the proof strategy will likely yield a valence-independent argument that illuminates the zero-residue case more transparently. If it fails for finite valence, the counterexample geometry would clarify what role injectivity plays beyond the mere avoidance of essential singularities.

4. Several Complex Variables #

In $\mathbb{C}^n$ for $n \ge 2$ the theory of isolated singularities of holomorphic functions changes dramatically: by Hartogs’ extension theorem, isolated singularities of holomorphic functions are always removable. One would expect the analogous conjecture for holomorphic 1-forms in $\mathbb{C}^n$ to be more tractable, or even to follow from known extension results.

Formulating the precise analogue, replacing the punctured disk by a domain $\Omega\setminus{0}$ in $\mathbb{C}^n$, and specifying what “meromorphic 1-form” means on a higher-dimensional domain, and checking whether Hartogs-type arguments already resolve it would clarify which features of the problem are genuinely one-dimensional.

5. Geometric Formulation on Riemann Surfaces #

The disk $D$ and the puncture at $0$ are not special: the same question can be posed on any Riemann surface $X$ with a marked point $p$. Given an open cover of $X\setminus{p}$ and injective holomorphic functions $f_j$ on each piece with compatible differentials, does $\omega = \mathrm{d}f_j$ extend meromorphically across $p$?

The answer may depend on the genus and the function theory of $X$. For the disk (simply connected, genus 0) the monodromy is a simple translation; for a torus or higher-genus surface the monodromy group is richer and the argument structure should change. Comparing these cases may isolate the essential input from the topology versus the analysis.

References #

Elsner, B. (1999). Hyperelliptic action integral. Annales de l’Institut Fourier, 49(1), 303–331. https://www.numdam.org/item/AIF_1999__49_1_303_0/
Ahlfors, L. V. (1979). Complex Analysis (3rd ed.). McGraw-Hill.
Conway, J. B. (1978). Functions of One Complex Variable (2nd ed.). Springer.
Nevanlinna, R. (1970). Analytic Functions. Springer.
Forster, O. (1981). Lectures on Riemann Surfaces. Springer.
Delabaere, E., Dillinger, H., & Pham, F. (1993). Résurgence de Voros et périodes des courbes hyperelliptiques. Annales de l’Institut Fourier, 43(1), 163–199.

Criterion for Boundedness of Power Series

Tue, 26 May 2026 00:00:00 +0000

Introduction & Problem Statement #

Power series constitute one of the most ubiquitous objects in analysis. A power series $\sum_{n=0}^{\infty}a_n x^n$ with infinite radius of convergence defines a real-entire function $f:\mathbb{R}\to\mathbb{R}$. Whereas the question of convergence is completely settled by Cauchy–Hadamard theory, the question of boundedness of the sum function is far subtler and, as of this writing, remains open.

Question 1 (Rüdinger, 2009)

Let $(a_n) _{n\ge 0}$ be a sequence of real numbers such that the power series $\sum _{n=0}^{\infty}a_n x^n$ converges for every $x\in\mathbb{R}$, thereby defining a smooth function $f:\mathbb{R}\to\mathbb{R}$. Give a necessary and sufficient criterion on $(a_n)$ for $f$ to be bounded on $\mathbb{R}$.

The problem is rated low importance on the Open Problem Garden and is recommended as accessible to undergraduates; nevertheless, a complete answer appears to be unknown.

Motivating examples.

Function	Power series	Bounded?
$\cos x$	$\displaystyle\sum_{k=0}^{\infty}\frac{(-1)^k}{(2k)!}x^{2k}$	$\|\cos x\|\le 1$
$\sin x$	$\displaystyle\sum_{k=0}^{\infty}\frac{(-1)^k}{(2k+1)!}x^{2k+1}$	$\|\sin x\|\le 1$
$e^x$	$\displaystyle\sum_{n=0}^{\infty}\frac{x^n}{n!}$	$e^x\to+\infty$
$p(x)=a_0+\cdots+a_Nx^N,\ N\ge 1$	(polynomial)	unbounded

Background & Prerequisites #

This section collects the core mathematical tools needed to engage seriously with Question 1.

Power Series and Entire Functions #

Definition 1 (Power Series & Radius of Convergence)

A power series centred at the origin is a formal series $\sum_{n=0}^{\infty}a_n x^n$ with $a_n\in\mathbb{R}$. Its radius of convergence is $$ R = \frac{1}{\limsup_{n\to\infty}|a_n|^{1/n}} \in [0,+\infty]. $$

Throughout this note we always assume $R=+\infty$, i.e., $\limsup_{n\to\infty}|a_n|^{1/n}=0$.

Definition 2 (Entire Function)

A function $f:\mathbb{C}\to\mathbb{C}$ is called entire if it is holomorphic on all of $\mathbb{C}$. Every power series with $R=+\infty$ defines a real-entire function, and by the identity theorem its complex extension is entire.

Theorem 1 (Cauchy–Hadamard)

The radius of convergence of $\sum a_n z^n$ equals $$ R = \Bigl(\limsup_{n\to\infty}|a_n|^{1/n}\Bigr)^{-1}. $$

Remark 1

The condition $R=+\infty$ is equivalent to $a_n = O(r^n/n!)$ for every $r>0$, i.e., the coefficients decay faster than any geometric sequence. This is the Paley–Wiener type condition for entire functions of order $1$.

Order and Type of Entire Functions #

Definition 3 (Order and Type)

The order of an entire function $f$ is $$ \rho = \limsup_{r\to\infty}\frac{\log\log M(r)}{\log r}, \qquad M(r)=\max_{|z|=r}|f(z)|. $$ The type $\sigma$ (for $0<\rho<\infty$) is $$ \sigma = \limsup_{r\to\infty}\frac{\log M(r)}{r^{\rho}}. $$

A bounded complex entire function has order $\rho=0$ (by Liouville’s theorem it must be constant), while a bounded real-valued entire function can be non-constant. Boundedness is therefore a genuinely real-variable phenomenon.

Liouville’s Theorem and Its Limitations #

Theorem 2 (Liouville)

Every bounded entire function $f:\mathbb{C}\to\mathbb{C}$ is constant.

Remark 2 (Why Liouville does not solve the problem)

Question 1 concerns real-valued functions $f:\mathbb{R}\to\mathbb{R}$. A function may be bounded on $\mathbb{R}$ while its complex extension is unbounded. For instance, $\cos z$ satisfies $|\cos z|\to\infty$ along the imaginary axis (since $\cos(iy)=\cosh y\to+\infty$). Liouville’s theorem therefore does not apply, and the problem is genuinely non-trivial.

Algebraic Structure of the Relevant Function Space #

Definition 4 (Space of Bounded Power Series)

Let $\mathcal{B}$ denote the set of all functions $f:\mathbb{R}\to\mathbb{R}$ that can be represented as a convergent power series $\sum_{n\ge 0}a_n x^n$ (with $R=+\infty$) and that are bounded on $\mathbb{R}$.

Proposition 1, Algebraic Properties of $\mathcal{B}$ (Rüdinger, 2009)

$\mathcal{B}$ is a linear subspace of $C^\infty(\mathbb{R})$: if $f,g\in\mathcal{B}$ and $\lambda\in\mathbb{R}$ then $f+\lambda g\in\mathcal{B}$.
$\mathcal{B}$ is closed under pointwise multiplication: if $f,g\in\mathcal{B}$ then $fg\in\mathcal{B}$.
$\mathcal{B}$ contains all functions of the form $c\cos(h(x))$, where $c\in\mathbb{R}$ and $h:\mathbb{R}\to\mathbb{R}$ is any entire function.

Remark 3

Part (3) follows from $\cos(h(x)) = \operatorname{Re}(e^{ih(x)})$ together with $|\cos(h(x))|\le 1$. The class is strictly larger than ${c\cos(bx):c,b\in\mathbb{R}}$; for example, $\cos(x^3-x)\in\mathcal{B}$.

Known Partial Results #

Necessary Conditions #

Proposition 2, Necessary Condition for Boundedness (Rüdinger, 2009)

Suppose $f(x)=\sum_{n=0}^{\infty}a_n x^n$ is bounded on $\mathbb{R}$. Then either:

$a_0$ is the only non-zero coefficient (i.e., $f$ is the constant function $f\equiv a_0$), or
there are infinitely many indices $n$ with $a_n\neq 0$, and the signs of the non-zero $a_n$ change infinitely often.

Remark 4

The sign-change condition is necessary: if the non-zero coefficients are eventually of one sign, the dominant-term comparison shows $f(x)\to\pm\infty$ as $x\to+\infty$ or $x\to-\infty$.

Corollary 1

Every non-constant polynomial is unbounded on $\mathbb{R}$.

Proof.

A polynomial has only finitely many non-zero coefficients. By Proposition 2 (1), the only bounded polynomial is the constant function. Any non-constant polynomial satisfies $|p(x)|\to\infty$ as $|x|\to\infty$.

The Sign-Change Condition Is Not Sufficient #

The condition of Proposition 2 is not sufficient, as the following examples show.

Example 1

Consider the geometric series $$ f(x) = \sum_{n=0}^{\infty}(-1)^n x^{2n} = \frac{1}{1+x^2}, \qquad |x|<1. $$ The coefficients alternate in sign, yet $R=1\neq+\infty$. One must first require $R=+\infty$ before the sign-change condition becomes meaningful.

For a subtler case with $R=+\infty$: take $a_n=(-1)^n/n!$, so $$ f(x) = \sum_{n=0}^{\infty}\frac{(-1)^n}{n!}x^n = e^{-x}. $$ The signs alternate, yet $e^{-x}\to+\infty$ as $x\to-\infty$.

Remark 5

The $e^{-x}$ example reveals the key gap: sign alternation of the coefficients does not prevent the function from growing in one direction, because the series for $e^{-x}$ reconstructs exponential growth in the negative half-line. A complete criterion must capture cancellation in both directions.

Connections to Entire Function Theory #

Theorem 3 (Borel–Carathéodory)

Let $f$ be holomorphic in $|z|\le R$. Then for $0<r<R$, $$ M(r) \le \frac{2r}{R-r}\sup_{|z|=R}\operatorname{Re}f(z) + \frac{R+r}{R-r},|f(0)|. $$

Remark 6

Borel–Carathéodory shows that the real part of a complex-valued entire function controls its modulus. For a real-valued function on $\mathbb{R}$ the analogous control is more delicate, since we only observe the function on a line, not on a disk.

Theorem 4 (Hadamard Factorisation)

Every entire function of finite order $\rho$ can be written as $$ f(z) = z^m e^{g(z)}\prod_{n=1}^{\infty} E_p!\left(\frac{z}{z_n}\right), $$ where $m\ge 0$, $p=\lfloor\rho\rfloor$, $g$ is a polynomial of degree $\le\rho$, and the $E_p$ are Weierstrass elementary factors.

Remark 7

A bounded real entire function of infinite order (if one exists) would not be directly covered by the Hadamard factorisation. Understanding the zero set and the exponential factor in $e^{g(z)}$ may be key to classifying all $f\in\mathcal{B}$.

The Open Sub-Question on the Generators of $\mathcal{B}$ #

Question 2 (Rüdinger, 2009)

Does $\mathcal{B}$ consist precisely of functions of the form $c\cos(h(x))$ and their linear combinations and products, where $h:\mathbb{R}\to\mathbb{R}$ is entire and $c\in\mathbb{R}$?

A positive answer would give an implicit characterisation via algebraic generators. A negative answer would require producing a bounded entire function on $\mathbb{R}$ that does not lie in the $\mathbb{R}$-algebra generated by ${\cos\circ, h : h\text{ entire}}$.

Remark 8

By Proposition 1 (3), every $c\cos(h(x))$ belongs to $\mathcal{B}$, and $\mathcal{B}$ is an algebra, so all products and sums remain in $\mathcal{B}$. What is unknown is whether every element of $\mathcal{B}$ arises this way. Note that $\sin x = \cos(x-\pi/2) \in \mathcal{B}$, so sine is already covered.

Research Directions and Conjectures #

Direction 1: Coefficient Growth Rate #

A promising approach is to examine the rate of decay of $|a_n|$, not just the sign pattern.

Question 3

Is there a decay condition on $|a_n|$, combined with the sign-change condition, that gives a sufficient criterion for $f\in\mathcal{B}$?

Approach. The Cauchy estimates give $|a_n| = |f^{(n)}(0)|/n!\le M(r)/r^n$ for all $r>0$. If $f\in\mathcal{B}$ with $|f|\le B$, the bound $|a_n|\le B/r^n$ holds for every $r>0$, but this recovers only the $R=+\infty$ condition. Is there a sharper constraint?

Direction 2: Fourier-Analytic Approach #

Every $f\in L^\infty(\mathbb{R})\cap L^2(\mathbb{R})$ possesses a square-integrable Fourier transform. If $f$ is also entire, Paley–Wiener forces the transform to be compactly supported. However, a generic $f\in\mathcal{B}$ may not lie in $L^2$ (e.g., $\cos x\notin L^2(\mathbb{R})$).

Question 4

Can the Fourier theory for tempered distributions give a necessary and sufficient condition for $f\in\mathcal{B}$ in terms of the spectral support of $f$?

Direction 3: Differential Equation Characterisation #

Bounded entire functions often arise as solutions to ODEs. For instance $y’’+y=0$ has bounded solutions $A\cos x + B\sin x$. More generally, $y’’+\omega(x)y=0$ with $\omega$ entire and bounded can produce bounded solutions.

Question 5

Characterise those linear differential operators $L$ with entire coefficients whose full solution space lies within $\mathcal{B}$.

Direction 4: Even/Odd Decomposition and Reduction #

Every $f\in\mathcal{B}$ splits as $f=f_e+f_o$ where $$ f_e(x)=\tfrac{1}{2}(f(x)+f(-x))=\sum_{k\ge 0}a_{2k}x^{2k} \quad\text{and}\quad f_o(x)=\tfrac{1}{2}(f(x)-f(-x))=\sum_{k\ge 0}a_{2k+1}x^{2k+1}. $$ Since $f_e(x)=g(x^2)$ for the entire function $g(t)=\sum_{k\ge 0}a_{2k}t^k$, boundedness of $f_e$ reduces to: is $g$ bounded on $[0,+\infty)$? This reduction may make the even and odd parts easier to study separately.

Direction 5: Polynomial Approximation and Numerics #

Question 6

If the partial sums $S_N(x)=\sum_{n=0}^{N}a_n x^n$ are uniformly bounded on growing intervals $[-R_N,R_N]$ (with $R_N\to\infty$), does it follow that $f\in\mathcal{B}$? Conversely, if $f\in\mathcal{B}$, how fast must $R_N$ grow relative to $N$ for the bound to hold?

Summary of Open Problems #

#	Statement
Q1	Give a necessary and sufficient condition on $(a_n)$ for $f=\sum a_n x^n$ to be bounded on $\mathbb{R}$.
Q2	Is $\mathcal{B}$ generated (as an algebra) precisely by ${c\cos(h(x)):h\text{ entire}}$?
Q3	Does a sharper decay condition on $
Q4	Can spectral-support (Paley–Wiener / distribution) theory characterise $\mathcal{B}$?
Q5	Which linear ODEs with entire coefficients have solution space $\subseteq\mathcal{B}$?
Q6	What is the precise relationship between truncation bounds on $[-R_N,R_N]$ and $f\in\mathcal{B}$?

References #

Ahlfors, L. V. (1979). Complex Analysis, 3rd ed. McGraw-Hill.
Boas, R. P. (1954). Entire Functions. Academic Press.
Conway, J. B. (1978). Functions of One Complex Variable, 2nd ed. Springer.
Levin, B. Ya. (1996). Lectures on Entire Functions. AMS Translations of Mathematical Monographs, vol. 150.
Rudin, W. (1976). Principles of Mathematical Analysis, 3rd ed. McGraw-Hill.
Rudin, W. (1987). Real and Complex Analysis, 3rd ed. McGraw-Hill.
Rüdinger, A. (2009). Criterion for boundedness of power series. Open Problem Garden. http://www.openproblemgarden.org/op/criterion_for_boundedness_of_power_series
Stein, E. M. and Shakarchi, R. (2003). Fourier Analysis: An Introduction. Princeton University Press.
Stein, E. M. and Shakarchi, R. (2010). Complex Analysis. Princeton University Press.
Titchmarsh, E. C. (1939). The Theory of Functions, 2nd ed. Oxford University Press.