[Paper Review] The ABC Theorem for Meromorphic Functions

Machiel van Frankenhuijsen|ArXiv.org|May 12, 2008

Meromorphic and Entire Functions10 references3 citations

TL;DR

This paper presents a new proof of the abc theorem for meromorphic functions using a formal height-to-radical identity and an analytic bound on the archimedean contribution to the radical. It establishes that the logarithmic height is bounded by a completed radical, with the archimedean term controlled via the logarithmic derivative lemma, yielding an effective error term of order $2\log h + O(\log\log h)$ outside a set of finite measure.

ABSTRACT

Using a `height-to-radical' identity, we define the archimedean contribution to the radical, $r_\arch$, and we give a new proof of the abc theorem for the field of meromorphic functions. The first step of the proof is completely formal and yields that the height is bounded by the radical, $h\leq r$, where $r=r_ a+r_\arch$ is the radical completed with the archimedean contribution. The second step is analytic in nature and uses the lemma on the logarithmic derivative to derive a bound for $r_\arch$.

Motivation & Objective

• To reprove the abc theorem for meromorphic functions using a two-step approach that separates formal and analytic components.
• To define and analyze the archimedean contribution to the radical, which is missing in classical formulations.
• To provide a new interpretation of the error term in the abc conjecture as arising from archimedean valuations.
• To bridge Nevanlinna theory and number-theoretic abc conjectures by introducing a completed radical including archimedean contributions.
• To suggest that the error term $\psi(h)$ in the abc conjecture should be viewed as a bound on the archimedean radical contribution, not as the contribution itself.

Proposed method

• Introduce a formal height-to-radical identity that yields $h(P,\rho) \leq r_{\text{na}}(P,\rho) + r_{\text{arch}}(P,\rho)$ for all $\rho \geq 1$.
• Define the archimedean contribution to the radical as $r_{\text{arch}}(P,\rho) = \int_{|z|=\rho} h_z((b/c)^\ell, (c/a)^\ell, (a/b)^\ell) \frac{dz}{2\pi i z} - h_\infty((b/c)^\ell, (c/a)^\ell, (a/b)^\ell)$.
• Apply the logarithmic derivative lemma to bound $r_{\text{arch}}(P,\rho) \leq 2\log h(P,\rho) + O(\log\log h(P,\rho))$ outside a set $E \subset (1,\infty)$ of finite total length.
• Establish the function-theoretic abc inequality: $h(P,\rho) \leq r_{\text{na}}(P,\rho) + 2\log h(P,\rho) + O(\log\log h(P,\rho))$ for $\rho \geq 1$, $\rho \notin E$.
• Propose a reinterpretation of the abc conjecture's error term $\psi(h)$ as an upper bound on the archimedean radical contribution, not the contribution itself.
• Suggest a program to define a completed radical $r(P) = r_{\text{na}}(P) + r_{\text{arch}}(P)$ and prove $h(P) \leq r(P)$, with $r_{\text{arch}}(P) \leq \psi(h(P))$ as the key analytic challenge.

Experimental results

Research questions

• RQ1 How can the abc theorem for meromorphic functions be re-proven using a formal identity that separates height and radical components?
• RQ2 What is the correct definition of the archimedean contribution to the radical in the context of Nevanlinna theory?
• RQ3 Can the error term in the abc conjecture be interpreted as a bound on the archimedean radical contribution rather than the contribution itself?
• RQ4 What is the sharp analytic bound for the archimedean contribution to the radical in terms of the height function?
• RQ5 How does the completed radical, including archimedean contributions, relate to the classical abc conjecture in number theory?

Key findings

• The height-to-radical identity yields $h(P,\rho) \leq r(P,\rho)$ for all $\rho \geq 1$, where $r(P,\rho) = r_{\text{na}}(P,\rho) + r_{\text{arch}}(P,\rho)$, establishing a formal abc inequality.
• The archimedean contribution to the radical is defined via an integral of logarithmic derivatives, ensuring coordinate independence and analytic tractability.
• Using the logarithmic derivative lemma, the archimedean contribution is bounded by $r_{\text{arch}}(P,\rho) \leq 2\log h(P,\rho) + O(\log\log h(P,\rho))$ outside a set $E$ of finite total length.
• The resulting function-theoretic abc theorem states $h(P,\rho) \leq r_{\text{na}}(P,\rho) + 2\log h(P,\rho) + O(\log\log h(P,\rho))$ for $\rho \geq 1$, $\rho \notin E$.
• The paper proposes that the error term $\psi(h)$ in the number-theoretic abc conjecture should be interpreted as an upper bound on the archimedean radical contribution, not as the contribution itself.
• A stronger result with $\psi(h) = O(\sqrt{h}/\log h)$ would imply $h \leq C r$, potentially settling Fermat’s Last Theorem for large exponents.