Barsegian and Yang (2004) introduced jumping a-points, a generalization of multiple a-points, and proved that a-points at which ∣w′∣|w'| is small (or at which the inverse meromorphic derivative is large) are exceptional in the Nevanlinna sense: the associated counting functions satisfy a deficiency relation summing to at most 4. Their Theorem 2 is proved only for meromorphic functions of lower order λ>2\lambda>2, and the exponential function (λ=1\lambda=1) refutes it at λ=1\lambda=1. The open problem is to determine for which lower orders 0≤λ≤20\le\lambda\le2 these deficiency relations hold.

Background

For a meromorphic function ww in the complex plane, Nevanlinna theory measures growth by the characteristic function T(r,w)T(r,w) and measures how often ww takes a value aa by the integrated counting function N(r,a,w)N(r,a,w) of the aa-points (the solutions of w(z)=aw(z)=a). The deficiency δ(a)=1−lim sup⁡r→∞N‾(r,a,w)/T(r,w)\delta(a)=1-\limsup_{r\to\infty}\overline{N}(r,a,w)/T(r,w) quantifies how rarely aa is taken, and the classical deficiency relation states ∑aδ(a)≤2\sum_a \delta(a)\le 2. Multiple aa-points, i.e. aa-points where w′(z)=0w'(z)=0, are also known to be exceptional in this sense: writing N(r,w=a,w′=0)N(r,w=a,w'=0) for the counting function of aa-points at which the derivative vanishes, the classical relation recorded as inequality (1.6) in the source is ∑ν=1qlim inf⁡r→∞N(r,w=aν,w′=0)/T(r,w)≤4\sum_{\nu=1}^q \liminf_{r\to\infty} N(r,w=a_\nu,w'=0)/T(r,w)\le 4 for any distinct values a1,…,aqa_1,\dots,a_q.

G. A. Barsegian and C. C. Yang, in a 2004 book chapter, generalized this picture. Given distinct values a1,…,aqa_1,\dots,a_q with mutual separation encoded by ρ(w)=min⁡ν≠μ∣aν−aμ∣/3\rho^{(w)}=\min_{\nu\ne\mu}|a_\nu-a_\mu|/3, they call an aνa_\nu-point z0z_0, with Taylor expansion w(z)=aν+c1(z−z0)+c2(z−z0)2+⋯w(z)=a_\nu+c_1(z-z_0)+c_2(z-z_0)^2+\cdots, a jumping aνa_\nu-point if for some integer k≥2k\ge 2 the coefficient ckc_k dominates a fixed multiple of ∣c1∣k|c_1|^k, the multiple depending only on kk and ρ(w)\rho^{(w)}. This notion generalizes multiple points (which correspond to c1=0c_1=0) and captures points where the first derivative is very small relative to the higher Taylor coefficients. Their Theorem 1 proves the same deficiency relation for jumping aa-points: ∑ν=1qlim inf⁡r→∞N(r,aν,↑,w)/T(r,w)≤4\sum_{\nu=1}^q \liminf_{r\to\infty} N(r,a_\nu,\uparrow,w)/T(r,w)\le 4, where the arrow marks the counting function restricted to jumping aa-points (counted with multiplicity).

The subject of the present problem is their Theorem 2. The lower order of ww is λ=lim inf⁡r→∞log⁡T(r,w)/log⁡r\lambda=\liminf_{r\to\infty}\log T(r,w)/\log r. For a fixed R>0R>0, let N(r,w=a,∣w′∣≤R)N(r,w=a,|w'|\le R) be the integrated counting function of aa-points (counted with multiplicity) at which ∣w′(z)∣≤R|w'(z)|\le R; and for an integer U≥1U\ge 1 let F(U)(w(z))F^{(U)}(w(z)) denote the UU-th inverse meromorphic derivative, defined as in the second precursor paper (Barsegian–Yang, Complex Variables 44 (2001) 13–27): for an aa-point ziz_i of ww, let FiF_i be the branch of the local inverse function of ww normalized by Fi(w(zi))=ziF_i(w(z_i))=z_i, and let Fi(U)F_i^{(U)} be its UU-th derivative with respect to ww; the composition F(U)(w(z))=Fi(U)(w(z))F^{(U)}(w(z))=F_i^{(U)}(w(z)) is then a meromorphic function of zz, single-valued near each aa-point, having at a multiple aa-point of multiplicity kk a pole of order Uk−1Uk-1 (in particular F(1)(w(z))=1/w′(z)F^{(1)}(w(z))=1/w'(z)), so the condition ∣F(U)(w(z))∣≥R|F^{(U)}(w(z))|\ge R is well defined at every aa-point independently of any choice of domains in the proofs. Theorem 2 states that if λ>2\lambda>2, then for every R>0R>0: (1) for every value aa outside a countable set depending on ww, lim inf⁡r→∞N(r,w=a,∣w′∣≤R)/T(r,w)=0\liminf_{r\to\infty} N(r,w=a,|w'|\le R)/T(r,w)=0, and ∑a∈Clim inf⁡r→∞N(r,w=a,∣w′∣≤R)/T(r,w)≤4\sum_{a\in\mathbb{C}} \liminf_{r\to\infty} N(r,w=a,|w'|\le R)/T(r,w)\le 4; and (2) the same pair of conclusions holds for N(r,w=a,∣F(U)∣≥R)N(r,w=a,|F^{(U)}|\ge R). Thus, for sufficiently fast-growing functions, aa-points with small derivative (or with large inverse derivative) are as exceptional as multiple points.

The proof rests on the machinery of two earlier Barsegian–Yang papers: Ahlfors-type sets of simple aa-points grouped into pairwise disjoint simply connected domains Ei(r,w)E_i(r,w), of diameter at most φ(r)r/A1/2(r,w)\varphi(r) r/A^{1/2}(r,w), on which ww is univalent, where A(r,w)A(r,w) is the Ahlfors–Shimizu characteristic (the spherical area of the image of ∣z∣<r|z|<r, which is comparable to T(r,w)T(r,w)) and φ(r)\varphi(r) is an auxiliary function growing arbitrarily slowly to infinity. On these domains the derivative at the captured aa-points is comparable to A1/2(r,w)/(φ(r)r)A^{1/2}(r,w)/(\varphi(r) r). Since A1/2(r,w)/rA^{1/2}(r,w)/r grows like rλ/2−1r^{\lambda/2-1}, this lower bound outruns any fixed RR only when λ>2\lambda>2, which is exactly where the method stops. The restriction is not cosmetic: for w(z)=ezw(z)=e^z, whose lower order is λ=1\lambda=1, every aa-point with a≠0a\ne 0 satisfies ∣w′∣=∣a∣|w'|=|a|, so for any R>0R>0 all values 0<∣a∣≤R0<|a|\le R have N(r,w=a,∣w′∣≤R)/T(r,w)→1N(r,w=a,|w'|\le R)/T(r,w)\to 1, and five such values already violate the sum bound. So Theorem 2 genuinely fails at λ=1\lambda=1, while the case 0≤λ≤20\le\lambda\le 2 is not covered by the source's theorem.

Problem Statement

Determine the range of validity, in lower order 0≤λ≤20\le\lambda\le 2, of the Barsegian–Yang Theorem 2 for meromorphic functions ww in the complex plane. Since w=ezw=e^z (lower order λ=1\lambda=1) verifiably refutes the statement at λ=1\lambda=1 (see Background), the question is: for which λ∈[0,2]\lambda\in[0,2] is it true that, for every fixed R>0R>0, the aa-points with ∣w′∣≤R|w'|\le R are exceptional in the sense that

lim inf⁡r→∞N(r,w=a,∣w′∣≤R)T(r,w)=0\liminf_{r\to\infty}\frac{N(r,w=a,|w'|\le R)}{T(r,w)}=0

for every value aa outside a countable set depending on ww, and

∑a∈Clim inf⁡r→∞N(r,w=a,∣w′∣≤R)T(r,w)≤4,\sum_{a\in\mathbb{C}}\liminf_{r\to\infty}\frac{N(r,w=a,|w'|\le R)}{T(r,w)}\le 4,

and likewise that the aa-points with ∣F(U)(w(z))∣≥R|F^{(U)}(w(z))|\ge R (for each integer U≥1U\ge 1) are exceptional, with the same individual-liminf and sum statements for N(r,w=a,∣F(U)∣≥R)N(r,w=a,|F^{(U)}|\ge R)? Here N(r,w=a,∣w′∣≤R)N(r,w=a,|w'|\le R) counts aa-points with multiplicity, T(r,w)T(r,w) is the Nevanlinna characteristic, F(U)(w(z))F^{(U)}(w(z)) is the UU-th inverse meromorphic derivative defined in the Background (the branch of the local inverse normalized by Fi(w(zi))=ziF_i(w(z_i))=z_i, differentiated UU times with respect to ww and composed with ww; at a multiple aa-point of multiplicity kk it has a pole of order Uk−1Uk-1), and the lower order is λ=lim inf⁡r→∞log⁡T(r,w)/log⁡r\lambda=\liminf_{r\to\infty}\log T(r,w)/\log r. A complete answer determines the set S⊆[0,2]S\subseteq[0,2] of lower orders for which both parts of Theorem 2 hold, proves the relations for every λ∈S\lambda\in S, and exhibits meromorphic counterexamples of lower order λ\lambda for every λ∈[0,2]∖S\lambda\in[0,2]\setminus S; no interval structure of SS is presumed.

References

  1. G. A. Barsegian and C. C. Yang, On Some New Concept of Exceptional Values, in: Value Distribution Theory and Related Topics (Advances in Complex Analysis and Its Applications 3), Kluwer Academic Publishers / Springer, Boston, MA, 2004, pp. 105–116. DOI: 10.1007/1-4020-7951-6_4, https://link.springer.com/chapter/10.1007/1-4020-7951-6_4; zbMATH review Zbl 1090.30033, https://zbmath.org/?q=an:1090.30033
  2. G. A. Barsegian and C. C. Yang, Some new generalizations in the theory of meromorphic functions and their applications, I: On the magnitudes of functions on the sets of a-points of derivatives, Complex Variables, Theory and Application 41 (2000), 293–313. DOI: 10.1080/17476930008815256, https://doi.org/10.1080/17476930008815256
  3. G. A. Barsegian and C. C. Yang, Some new generalizations in the theory of meromorphic functions and their applications, II: On the derivatives of meromorphic and inverse functions, Complex Variables, Theory and Application 44 (2001), 13–27. DOI: 10.1080/17476930108815340, https://doi.org/10.1080/17476930108815340
  4. G. A. Barsegian, I. Laine and C. C. Yang, On a method of estimating derivatives in complex differential equations, Journal of the Mathematical Society of Japan 54 (2002), 923–935. DOI: 10.2969/jmsj/1191591998, https://doi.org/10.2969/jmsj/1191591998
  5. J. Lin, W. Xiong and W. Yuan, Meromorphic Solutions of Some Algebraic Differential Equations, Abstract and Applied Analysis 2014 (2014), 796312. DOI: 10.1155/2014/796312, https://doi.org/10.1155/2014/796312
  6. Grigor Barsegian, Universal Version of Value Distribution Theory for Functions in a Given Domain, Lobachevskii Journal of Mathematics 45 (2024), 3479–3487. DOI: 10.1134/S1995080224604648, https://doi.org/10.1134/S1995080224604648