Irrationality exponents of logarithms of positive rational numbers

arXiv:2610.10192 · math.NT · Submitted 2026-10-07 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: We're looking at the paper: "Irrationality exponents of logarithms of positive rational numbers".

Emmy: Here's the gist of the paper. We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for µ(log r) = 2 whenever r ∈ Q>0 !

Ada: The main result asserts that this lower bound is attained throughout the family considered here!

Emmy: The Core Result on Irrationality Exponents Theorem 1 states that for every r ∈ Q>0 !

Ada: and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p/q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)>!

Title and authors: Ada: So we're looking at a paper called "Irrationality exponents of logarithms of positive rational numbers." It’s by Jingwen Liu, Kai Jiang, and Pingwen Zhang. It’s about how well you can approximate the logarithm of a positive fraction with a rational number.

Emmy: Exactly. The authors are combining approximation theory with some geometric ideas to figure out something about these logarithms that we don't always expect. It sounds like they're setting a lower bound on the irrationality exponent for these values, suggesting it’s exactly two for any positive rational number except one.

Kolya: When you say "irrationality exponent," what does that mean in this context? Is it about how irrational the number is, or something else related to how well we can approximate it?

Ada: It’s about the approximation quality. They show that for any positive rational number r not equal to one, if you have a real number nu greater than two, you can always find a denominator q, and an integer p, such that the difference between your approximation of the logarithm of that fraction and the actual logarithm is smaller than something related to that denominator, specifically less than q minus nu.

Emmy: That's a lower bound result. It’s not just saying these logarithms are irrational; it’s giving us a specific rate at which they are approximated. The authors claim this lower bound is attained throughout the whole family of rational numbers they're considering.

Kolya: So, what does this actually change for us? Does it give us a better handle on how hard it is to approximate r ?

Ada: It means that the lower bound suggested by Dirichlet's theorem, which is two, is not just an asymptotic result; they show it’s achieved. They establish that for every positive rational number r, the exponent mu(r) equals two. This confirms a long-standing conjecture for this specific class of numbers >

The paper's summary: Emmy: Now, looking at the details of "Irrationality exponents of logarithms of positive rational numbers," they use a moving-centre polynomial interpolation theorem. This is the geometric input they rely on.

Ada: That geometric part is key because it allows them to set up a structure where they can construct a nonzero rational determinant. This determinant argument is their main tool for proving the result about mu(r) = two <ref:2610.10192#pg1>.

Kolya: So, how does that construction work? Is it some kind of clever way to use the approximation errors to build something non-zero? What kind of mathematical object are they comparing?

Ada: They construct a nonzero rational determinant. They then show that the arithmetic lower bound for this determinant is incompatible with an analytic upper bound they derive from repeated Taylor expansions and row translations. This contradiction forces the result, showing that no excessively good approximations can exist simultaneously.

Emmy: The paper also gives them some improvements to the method itself, specifically a bounded-fibre interpolation extension. This lets them show that even when the interpolation centres aren't all distinct points in space, you can still make the map surjective under certain conditions.

Kolya: That sounds like they’re making the geometric part more robust. If you lose that distinct centre structure, does the whole argument fall apart? What are those specific conditions for it to still work?

Ada: The authors prove that for a fixed rational argument, they can find an explicit degree threshold. This threshold depends on the chosen approximation centers, but once those centers are fixed, the final degree bound is determined by the structure of that configuration. They also show that this result extends to every real nu greater than two.

The paper's improvements: Emmy: Beyond just proving the main theorem, they offer several structural improvements to how you approach these problems. One of the key ones is the idea of successive weight thresholds for the interpolation weights.

Ada: They show that these successive weight thresholds are independent of where you pick your interpolation centers. That independence is a big deal because it means the choice of centers doesn't mess up the fundamental structure they've built up for finding these bounds.

Kolya: That’s good because usually, when you change the setup—like moving those centres around—you have to re-derive everything. Does this independence simplify things for applying these results to different kinds of rational numbers?

Emmy: It simplifies the application significantly. They show that for a fixed rational argument, they can find an explicit bound on the final analytic degree required to get the contradiction. This is a concrete number instead of just saying "it exists."

Ada: And they also have this bounded-fibre interpolation extension. This shows that even if you don't have perfectly distinct centres, as long as you meet a certain condition on the number of points used, the map can still be surjective. It’s a more practical tool for when your setup isn't perfectly ideal.

Kolya: So, to summarize these improvements: they got rid of circular dependence in the weight choices and gave us an explicit degree bound for the analytic part. Does that mean we have a more reliable way to prove irrationality exponents?

Ada: It means yes. They've replaced vague existence claims with concrete bounds derived from geometric constraints, which is exactly what you need when you’re trying to establish sharp results on these approximation problems >

Conclusion: Emmy: So, to wrap up the paper "Irrationality exponents of logarithms of positive rational numbers," they’ve shown that the irrationality exponent for the logarithms of positive rational numbers is exactly two. This result comes from a combination of a moving-centre interpolation theorem and a determinant argument comparing arithmetic and analytic bounds.

Ada: And what this means for us is that we now have a rigorous way to establish this lower bound, showing it's attained everywhere, not just in some limit case. It connects the analytic side—the growth of coefficients—with the arithmetic side—the denominators of approximations >

Kolya: From a probabilistic standpoint, I see this as confirming that the tail behavior governed by these approximation errors is very strong. The constants they derive survive when you apply them to various rational combinations or affine changes >

Emmy: And they go further with consequences. They show that you can form nonzero rational linear combinations of these logarithms, and also logs of positive numbers raised to a positive integer power, and even rational affine changes that still maintain the exponent of two >

Ada: It’s a nice structure. The whole thing hinges on the fact that r is rational. They also provided a counterexample showing that you can't rely only on volume conditions for interpolation; you need the weight separation hypothesis >

Kolya: That's important context because it shows where the methods are actually limited. The geometric proof in Section two requires that weight separation, and without it, even if the volume ratio is small, the map fails to be onto > <ref:2610.10192#pg1>

Emmy: So they’ve given us a complete picture: a sharp result for rational logarithms, structural improvements for their method and how we set up those problems. It’s a solid piece of work >

School of Mathematics and Statistics, Wuhan University · Wuhan Institute for Math & AI

math.NT

Submitted: 2026-10-07

Updated: 2026-11-05

Comments: 18 pages. Revised manuscript and abstract, with expanded historical context, explanations of the parameter choices, and a discussion of the connection with the argument for pi through the exponential map

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 86/100

The gist: We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for µ(log r) = 2 whenever r ∈ Q>0 ! The main result asserts that this…

Key concepts

Irrationality Exponent ($\mu(x)$)
This measures how well a number $x$ can be approximated by rational numbers. If $\mu(x) = 2$, it means that for any rational approximation $p/q$, the error $|x - p/q|$ is bounded below by $1/q^2$. The paper aims to prove this specific bound for logarithms of positive rationals.
Separated-Weight Determinant Argument
This is a technique used to establish irrationality measures. It involves constructing a rational determinant where the arithmetic constraints (lower bounds) derived from its structure are shown to be incompatible with the analytic constraints (upper bounds) derived from an identity involving the logarithm.
Dirichlet's Theorem Lower Bound
Dirichlet's theorem provides a general lower bound for irrational numbers, stating that $\mu(x) \ge 2$. The paper proves that this minimum bound is actually achieved for the logarithms of positive rational numbers, meaning they are as well-approximated as possible according to this fundamental theorem.

Terminology

Summary

We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for µ(log r) = 2 whenever r ∈ Q>0 ! p. 0

Where the paper says it

Irrationality exponents of logarithms of positive rational numbers Jingwen Liu1 Kai Jiang2 Pingwen Zhang3 7 October 2026 Abstract We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for µ(log r) = 2 whenever r ∈ Q>0 \ {1}. The geometric input is an interpolation theorem for logarithmic jets at points with distinct multiplicative coordinates.

The main result asserts that this lower bound is attained throughout the family considered here! p. 1
Where the paper says it

Dirichlet’s theorem gives µ(x) ≥ 2 for every irrational real x. The main result asserts that this lower bound is attained throughout the family considered here.

The Core Result on Irrationality Exponents

Theorem 1 states that for every r ∈ Q>0 ! and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p/q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)>! p. 1

Where the paper says it

Theorem 1 (Logarithms of positive rational numbers). For every r ∈ Q>0 \ {1} and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)).

Consequently log r is irrational and µ(log r) = 2! p. 1
Where the paper says it

For every r ∈ Q>0 \ {1} and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)). Consequently log r is irrational and µ(log r) = 2.

This result establishes that the lower bound provided by Dirichlet's theorem is attained for logarithms of positive rational numbers!

Geometric Input and Interpolation

The argument combines moving-centre polynomial interpolation with arithmetic and analytic estimates for a single determinant! The geometric input depends on the centre configuration, and the present proof adapts the separated-weight method of [1, Section 2]! p. 1

Where the paper says it

The geometric input depends on the centre configuration. The present proof adapts the separated-weight method of [1, Section 2].

Specifically, Theorem 2 proves that there are successive lower thresholds for positive rational weights w1,. p. 3
Where the paper says it

Fix positive integers m, K, positive rational numbers w0, v0, and a rational number 0 < θ < 1, such that K(w0/v0)θ m < 1. (1) There are successive lower thresholds for the positive rational weights w1, .

.., wm depending only on these fixed data and earlier weights! p. 3
Where the paper says it

(1) There are successive lower thresholds for the positive rational weights w1, . , wm, depending only on these fixed data and earlier weights, with the following property.

The Determinant Argument

The proof outline constructs a nonzero rational determinant and shows that its arithmetic lower bound is incompatible with its analytic upper bound! The steps involve assuming excessively good approximations, using interpolation to obtain a nonzero square minor ∆H, bounding it from below by arithmetic means, and bounding it from above by analysis via the identity e j log r = r j! p. 2

Where the paper says it

The steps, and their order of dependence, are as follows. Assume excessively good approximations and fix the parameters.

The comparison of these bounds leads to a contradiction for arbitrarily large exceptional denominators! p. 2
Where the paper says it

The remaining errors tend to zero, and the upper bound becomes smaller than the lower bound. This contradiction excludes arbitrarily large exceptional denominators.

Consequences and Extensions

The consequences include nonzero rational linear combinations of such logarithms, rational affine changes, and logarithms of positive numbers with a rational positive integer power! Corollary 7 shows that if α > 0, α ≠ 1, and α D ∈ Q for some positive integer D, then µ(log α) = 2! p. 1

Where the paper says it

Theorem 1 (Logarithms of positive rational numbers). For every r ∈ Q>0 \ {1} and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)).

Furthermore, the paper provides a bounded-fibre interpolation extension and a counterexample to interpolation under the volume condition alone! p. 0
Where the paper says it

The consequences include nonzero rational linear combinations of such logarithms, rational affine changes, and logarithms of positive numbers with a rational positive integer power. We also give a bounded-fibre interpolation extension, an explicit bound for the final analytic degree, and a counterexample to interpolation under the volume condition alone.

Scope and Effectivity

The integer determinant argument uses that r is rational, which avoids the need for estimates on conjugate determinants! The proof is qualitative in the denominator threshold Q(r, ν), and Appendix A gives an explicit degree bound for the analytic contradiction once all numerical parameters are fixed! p. 0

Where the paper says it

The consequences include nonzero rational linear combinations of such logarithms, rational affine changes, and logarithms of positive numbers with a rational positive integer power. We also give a bounded-fibre interpolation extension, an explicit bound for the final analytic degree, and a counterexample to interpolation under the volume condition alone.

The final limit requires only the existence of an interpolation threshold for that fixed centre configuration! p. 3
Where the paper says it

All asymptotic statements in a degree variable keep the dimension and weights fixed; all centres are fixed as well before the final degree limit is taken. Theorem 2 (MI: moving-centre interpolation).

Rational Combinations and Affine Changes

Lemma 5 proves that if x is irrational, u ∈ Q ! and v ∈ Q, then µ(ux + v) = µ(x)! Corollary 6 shows that for any nonzero logarithmic combination ξ = Xs k=1 ck log rk ≠ 0, the irrationality exponent is µ(ξ) = 2! p. 0

Where the paper says it

Irrationality exponents of logarithms of positive rational numbers Jingwen Liu1 Kai Jiang2 Pingwen Zhang3 7 October 2026 Abstract We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for µ(log r) = 2 whenever r ∈ Q>0 \ {1}. The geometric input is an interpolation theorem for logarithmic jets at points with distinct multiplicative coordinates.

Corollary 7 also covers positive rational powers, stating that if α > 0, α ≠ 1, and α D ∈ Q for some positive integer D, then µ(log α) = 2! p. 1
Where the paper says it

Theorem 1 (Logarithms of positive rational numbers). For every r ∈ Q>0 \ {1} and every real ν > 2, there is a positive integer Q(r, ν) such that log r − p q > q−ν (p ∈ Z, q ∈ Z, q ≥ Q(r, ν)).

The Insufficiency of Volume Condition

There is a concrete counterexample to dropping weight separation from MI! This example shows that the volume ratio (w0/v0)θ = 8/9 < 1, yet the interpolation map fails to be onto for every such H, despite the strict volume inequality!

Interpolation with Bounded Fibre Multiplicity

Theorem 9 recovers Theorem 2 by taking yj = 1 for all j and κ = K! This extension allows the map (2) to be onto for every H = nR with n ≥ n0, even when the centres are not distinct as full points! p. 3

Where the paper says it

For any pairwise distinct y0, . , yK−1 ∈ C ∗ and any cji ∈ C, the map PW (H) −→ K M−1 j=0 JV (H), P 7−→ P yj (1 + t),(cji + ui + log(1 + t))m i=1 j (2) is onto for every H = nR with n ≥ n0.

Why the Volume Condition Alone is Insufficient

There is a concrete counterexample to dropping weight separation from MI! This example shows that the volume ratio (w0/v0)θ = 8/9 < 1, yet the interpolation map fails to be onto for every such H, despite the strict volume inequality!

The Geometric Feature of Distinct Values

The distinct values rj are the geometric feature that permits the direct logarithmic argument! Theorem 2 allows arbitrary additive coordinates because a curve with constant multiplicative coordinate meets at most one centre! p. 2

Where the paper says it

curve with constant Y meets at most one centre, allowing arbitrary additive coordinates in Theorem 2. For π, all centres lie in Y = 1; interpolation requires additive-coordinate separation and Kθm < 1, as in the case κ = K of Appendix B.

The Role of Weight Separation

Appendix C shows that the volume inequality alone cannot replace the separation hypotheses! This demonstrates why weight separation is necessary for the geometric proof in Section 2! p. 1

Where the paper says it

The geometric input depends on the centre configuration. The present proof adapts the separated-weight method of [1, Section 2].

The Final Contradiction

The nonzero minor supplied by Theorem 2 therefore leads to a contradiction! The interpolation threshold may depend on the centres; all of them were fixed before taking this limit! p. 3

Where the paper says it

All asymptotic statements in a degree variable keep the dimension and weights fixed; all centres are fixed as well before the final degree limit is taken. Theorem 2 (MI: moving-centre interpolation).

This proves the asserted threshold Q(r, ν), and the reduction at the start extends it to every real ν > 2! p. 2
Where the paper says it

This contradiction excludes arbitrarily large exceptional denominators. Choosing a rational exponent between 2 and any given real exponent extends the conclusion to every real ν > 2.

The Final Conclusion

This completes the proof of Theorem 1! p. 8

Where the paper says it

Taking H = nR and undoing the filtered coordinate change gives (2). This completes the proof of Theorem 2.

Dirichlet’s theorem gives µ(ξ) ≥ 2, and the proved inequalities give µ(ξ) ≤ 2! p. 1
Where the paper says it

For an irrational real number x, its irrationality exponent is µ(x) = sup ν > 0 : x − p q < q−ν for infinitely many p ∈ Z, q ∈ Z>0 . Dirichlet’s theorem gives µ(x) ≥ 2 for every irrational real x.

This completes the proof of Theorem 1!

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved irrationality measure computation for logarithms of positive rational numbers by providing an explicit denominator threshold Q(r, nu) based on a geometric interpolation theorem and a comparison of arithmetic lower bounds with analytic upper bounds.

  2. Enhanced ability to derive consequences for rational linear combinations, affine changes, and logarithms of positive numbers with a rational positive integer power from the main result regarding the irrationality exponent of logarithms.

  3. Capability to verify if the volume condition alone is sufficient for interpolation by identifying concrete counterexamples showing that the volume inequality alone cannot replace the geometric proof in Section 2.

  4. Implementation of a method to construct nonzero rational determinants and compare their arithmetic lower bounds with analytic upper bounds to establish irrationality measures, as outlined in the proof outline.

  5. Functionality to select specific parameters for approximation arguments, such as setting theta = (5nu/2 - 4nu + 4)(3nu - 2) squared to control the growth of coefficients and balancing polynomial weight against centre growth.

Abstract

We study rational approximation to logarithms of positive rational numbers and give a separated-weight determinant argument for μ(r)=2 whenever r in Q>0 1. The geometric input is an interpolation theorem for logarithmic jets at points with distinct multiplicative coordinates. Its successive weight thresholds are independent of the interpolation centres; the eventual degree threshold may depend on them. A local intersection-length bound and a curve inequality lead to surjectivity through a blow-up and Serre vanishing. For a fixed rational argument, this produces a nonzero rational determinant. Clearing denominators gives an arithmetic lower bound, while an exact row translation and repeated Taylor indices yield a contradictory analytic upper bound. The consequences include nonzero rational linear combinations of such logarithms, rational affine changes, and logarithms of positive numbers with a rational positive integer power. We also give a bounded-fibre interpolation extension, an explicit bound for the final analytic degree, and a counterexample to interpolation under the volume condition alone.

Sources

Related papers