betaFeedback

WorksMathematics

Uniform cubic bounds for irrational values of the dilogarithm

Abstract

We study explicit denominator conditions ensuring the irrationality of the dilogarithm at positive and negative rational points. We prove that Li⁡2(p/q)\operatorname{Li}_2(p/q) is irrational whenever q>(4e4/27)p3q>(4e^4/27)p^3, and obtain an explicit, slightly smaller threshold for Li⁡2(−p/q)\operatorname{Li}_2(-p/q). The positive argument specializes the Rhin–Viola integral family and uses a uniform estimate for a restricted prime divisor. The negative argument combines the same arithmetic construction with three adjacent integral moments and an explicit maximization. We also give finite-numerator refinements by interpolating certified partition bounds and by introducing a comparison measure with a quadratic tail. In particular, Li⁡2(5/q)\operatorname{Li}_2(5/q) is irrational for every integer q≥418q\ge418. The proof separates the established integral and determinant constructions from the parameter estimates and finite certificates used here.

Introduction

The aim of this paper is to give explicit uniform denominator bounds for irrational values of the dilogarithm, and to refine them when the numerator is small. For ∣x∣≤1|x| \le1, we use Li⁡2(x)=∑m≥1xm/m2\operatorname{Li}_2(x) = \sum_{m \ge1} x^m/m^2. The classical values at 11 and −1-1 are π2/6\pi^2/6 and −π2/12-\pi^2/12. Away from these special points, a convergent defining series does not itself provide an effective irrationality argument: one needs integer linear forms whose arithmetic denominators grow more slowly than their analytic decay.

Integral constructions have been central to this problem since Beukers’s proof for ζ(2)\zeta(2) and ζ(3)\zeta(3) [1]. Hata [5] developed rational approximations to the dilogarithm and proved, in particular, irrationality at 1/s1/s for integers s≥7s \ge7 and s≤−5s \le-5. Rhin and Viola [6] introduced a five-parameter family of double integrals together with a permutation group that supplies additional common divisors of their coefficients. Their results give computable thresholds for arbitrary fixed positive numerators; the first two include Li⁡2(1/s)∉Q\operatorname{Li}_2(1/s) \notin\mathbb{Q} for s≥6s \ge6 and Li⁡2(2/s)∉Q\operatorname{Li}_2(2/s) \notin\mathbb{Q} for s≥51s \ge51. Rhin and Viola subsequently extended the construction to negative arguments and to linear independence of 11, Li⁡1\operatorname{Li}_1, and Li⁡2\operatorname{Li}_2 [7]. We use their polynomial identities and divisibility results as established arithmetic input.

Recent work on individual small rational points provides further useful input. Prellberg [8] proves irrationality at −1/4-1/4, 1/51/5 and −1/3-1/3. His negative-argument proof uses three bounded parameter shifts and a four-term moment recurrence to resolve nonvanishing. We apply that strategy to the symmetric Legendre family, with a variable negative argument, and obtain the explicit uniform threshold in Theorem 1.2.

The fixed-numerator arguments follow a separate recent line of work. Fauzan’s proof of the irrationality of ζ(5)\zeta(5) [3] uses normalized Hankel determinants, moment representations and logarithmic-energy estimates. Qian Tang adapted this approach to the dilogarithm, including the value Li⁡2(1/2)\operatorname{Li}_2(1/2) [10], and Si-Qi Liu extended the method to further rational arguments in Li2Rational [11]. This mathematical lineage is recorded in the latter project’s README and NOTICE. Our small-numerator refinements, including the bound q≥418q \ge418 for Li⁡2(5/q)\operatorname{Li}_2(5/q), build on the integer polynomials, arithmetic estimates, nonvanishing theorem and finite-circle comparison inequality in the pinned version of [11]. We retain its normalization and supply the interpolation estimates and quadratic comparison measure used below.

Our uniform statement has a simple cubic scale. Put C0=4e4/27=8.088614819725…C_0=4e^4/27=8.088614819725\ldots All numerators and denominators in the following statements are positive integers; coprimality is not required.

Theorem 1.1. If q>C0p3q>C_0p^3, then Li⁡2(p/q)\operatorname{Li}_2(p/q) is irrational.

Theorem 1.2. Define

B−(p)=2e4p327[1+(1+3e4p2)3/2]−2p3.(1)B_-(p)=\frac{2e^4p^3}{27}\left[1+\left(1+\frac{3}{e^4p^2}\right)^{3/2}\right]-\frac{2p}{3}. \tag*{(1)}

If q>B−(p)q>B_-(p), then Li⁡2(−p/q)\operatorname{Li}_2(-p/q) is irrational. In particular, each of the conditions q>C0p3−p/3+1/(4e4p)q> C_0p^3-p/3+1/(4e^4p) and q>C0p3q>C_0p^3 is sufficient.

The negative threshold satisfies B−(p)=C0p3−p/3+1/(4e4p)−1/(8e8p3)+O(p−5)B_-(p)=C_0p^3-p/3+1/(4e^4p)-1/(8e^8p^3)+O(p^{-5}). This expansion describes the size of the improvement; the sufficient conditions follow from a global inequality, proved in § 4, rather than from the expansion alone.

Theorem 1.3. For each row of Table 1 and Table 4, Li⁡2(p/q)\operatorname{Li}_2(p/q) is irrational for every integer q≥Q+q\ge Q_+, and Li⁡2(−p/q)\operatorname{Li}_2(-p/q) is irrational for every integer q≥Q−q\ge Q_-.

pp1234567
Q+Q_+1125416841810382233
Q−Q_-1166017843410582256

Table 1. Uniform sufficient thresholds for fixed numerators.

ppQ+Q_+Q−Q_-ppQ+Q_+Q−Q_-
84 1424 1381739 74039 734
95 8975 8941847 17247 166
108 0888 0841955 48055 474
1110 76610 7632064 70864 702
1213 97613 97425126 385126 377
1317 77117 767501 011 0741 011 060
1422 19622 1901008 088 6148 088 582
1527 30027 2941 0008 088 614 8208 088 614 484
1633 13033 126

Table 4. Additional sufficient thresholds, completing the 24-row list.

These thresholds apply to entire denominator ranges. Exact cancellation also gives additional isolated cases below them and includes the first denominator in several rows of the table.

Proof strategy and contributions

The proof has three parts. For positive arguments we fix the parameter family (2K,K−1,K,K,K)(2K,K-1,K,K,K) in the Rhin–Viola construction. A restricted first-period prime divisor suffices: its logarithmic gain is at least (32log⁡K−1)/K(\frac{3}{2}\log K-1)/K for K≥32K\ge32. Taking K=2pK=2p proves the large-numerator range, while four rational certificates cover 3≤p≤793\le p\le79. This gives one explicit constant for all numerators.

For negative arguments, we use three adjacent moments of a Legendre-type function. A four-term recurrence and uniqueness of polynomial moments ensure that these three moments cannot all vanish. Their common exponential majorant reduces the argument to maximizing Ax(1−x)2−xAx(1-x)^2-x on [0,1][0,1], where A=e4p2A=e^4p^2. The resulting maximum is the expression in (1.1) divided by pp.

For small numerators, positive coefficients make the logarithm of the absolute partition sum convex in the logarithm of its argument. This allows interpolation between existing certified profiles. A new comparison density, constant on an initial interval and quadratic on its remaining support, gives the rate needed for the last three rows of Table 1. We provide the approximation argument connecting this continuous density to the existing finite-circle theorem, with the comparison family fixed before the determinant order tends to infinity.

Arithmetic input and irrationality criteria

This section states the established arithmetic constructions in the normalizations used below and records the elementary final contradiction. Write dm=lcm⁡(1,…,m)d_m=\operatorname{lcm}(1,\ldots,m), with d0=1d_0=1. The prime number theorem gives log⁡dm=m+o(m)\log d_m=m+o(m); see, for example, [2 pp. 111–114].

Lemma 2.1. Let ξ∈R\xi\in\mathbb{R}. Suppose that, for each sufficiently large nn, finitely many numbers Ln,rL_{n,r} belong to Z+Zξ\mathbb{Z}+\mathbb{Z}\xi, their maximum absolute value tends to zero, and at least one is nonzero. Then ξ\xi is irrational.

Proof. If ξ=a/b\xi=a/b with a∈Za\in\mathbb{Z} and b∈Z>0b\in\mathbb{Z}_{>0}, then each bLn,rbL_{n,r} is an integer. A nonzero one has absolute value at least 1, contradicting their common convergence to zero.

For nonnegative integer parameters satisfying the usual four nonnegativity conditions, let Iz(h,j,k,l,m)I_z(h,j,k,l,m) denote the Rhin–Viola linear combination of a real double integral and its mixed residue integral; we follow [6 Section 2] for z>1z>1 and [7 Section 2] for z<0z<0. Their polynomial identity has the form

dHdKzα(z−1)βIz=P(z)−Q(z)Li⁡2(1/z),P,Q∈Z[z],(2)d_Hd_Kz^\alpha(z-1)^\beta I_z=P(z)-Q(z)\operatorname{Li}_2(1/z),\qquad P,Q\in\mathbb{Z}[z], \tag*{(2)}

with explicit degree bounds. We use the notation for HH, KK, α\alpha, β\beta, δ\delta from those sources only when specializing this identity; the letter KK in Section 3 is our freely chosen family parameter. The permutation action gives further divisibility by products of primes detected by floor-function inequalities [6 Lemma 4.1 and (4.4)].

The fixed-numerator argument uses a different established construction. Denote its positive absolute partition sum by Zn(t)Z_n(t), where 0<t≤1/20<t\le1/2, and put

Rσ(p,q)=83log⁡q−94log⁡(q−σp)−6log⁡p+7225,σ∈{−1,1}.(3)R_\sigma(p,q)=\frac{8}{3}\log q-\frac{9}{4}\log(q-\sigma p)-6\log p+\frac{72}{25},\qquad\sigma\in\{-1,1\}. \tag*{(3)}

The determinant and arithmetic results of [11] yield the following useful formulation.

Proposition 2.2 (Partition criterion). Suppose 0<t≤1/20<t\le1/2 and Zn(t)≤exp⁡(Bn2)Z_n(t)\le\exp(Bn^2) for all sufficiently large nn. If p/q≤tp/q\le t and

B+log⁡pqt<Rσ(p,q),(4)B+\log\frac{p}{qt}<R_\sigma(p,q), \tag*{(4)}

then Li⁡2(σp/q)\operatorname{Li}_2(\sigma p/q) is irrational.

Proof. The node-scaling inequality is Zn(p/q)≤(p/(qt))n(n+1)Zn(t)Z_n(p/q)\le(p/(qt))^{n(n+1)}Z_n(t). The integer polynomials in the cited construction have degree at most 2n2n, and their arithmetic growth is governed by (2.2). The strict gap in (4) therefore gives integer polynomials PnP_n with ∣Pn(Li⁡2(σp/q))∣≤e−cn2\lvert P_n(\operatorname{Li}_2(\sigma p/q))\rvert\le e^{-cn^2} for some c>0c>0, and infinitely many of these values are nonzero. If Li⁡2(σp/q)=u/v\operatorname{Li}_2(\sigma p/q)=u/v, then v2nPn(u/v)v^{2n}P_n(u/v) is an integer and its absolute value tends to zero along that infinite subsequence, a contradiction. For unreduced p/qp/q, cancelling a common factor hh increases the arithmetic rate by (67/12)log⁡h(67/12)\log h, so the stated raw rate remains sufficient.

The polynomial construction, arithmetic growth estimate and nonvanishing theorem in this proposition are inputs from [11]. The interpolation and comparison estimates supplying BB are proved below.

The positive cubic bound

The aim of this section is to prove Theorem 1.1 by choosing one family of Rhin–Viola parameters and estimating its prime gain uniformly. Fix pp, qq and put z=q/p>1z=q/p>1. Choose an integer K≥2K\ge2, write ε=1/K\varepsilon=1/K, and set N=KnN=Kn. The parameter KK is fixed while odd positive integers nn tend to infinity. Specialize the five parameters to (2N,N−n,N,N,N)(2N,N-n,N,N,N).

Integral forms and their sign

Removing the factor z−2Nz^{-2N} from the normalization in [6], the real integral is

J0,n=∫01∫01xN−n(1−x)2NyN(1−y)N[x+(z−x)y]N−n+1 dy dx.(5)J_{0,n}=\int_0^1\int_0^1\frac{x^{N-n}(1-x)^{2N}y^N(1-y)^N}{[x+(z-x)y]^{N-n+1}}\,\mathrm{d}y\,\mathrm{d}x. \tag*{(5)}

Let J1,nJ_{1,n} be obtained by replacing the inner integration by the normalized, positively oriented residue around its unique pole, and put Fn=J0,n−(log⁡z)J1,nF_n=J_{0,n}-(\log z)J_{1,n}. If r=N−nr=N-n, expansion at the pole gives

J1,n=(−1)n∫01xN(1−x)2N(z−x)N+1∑i=0N(Ni)(N+ir)(xz−x)i dx.(6)J_{1,n}=(-1)^n\int_0^1\frac{x^N(1-x)^{2N}}{(z-x)^{N+1}}\sum_{i=0}^{N}\binom{N}{i}\binom{N+i}{r}\left(\frac{x}{z-x}\right)^i\,\mathrm{d}x. \tag*{(6)}

Every factor inside the integral is nonnegative, and the integrand is positive on (0,1)(0,1). Hence J1,n<0J_{1,n}<0 for odd nn, and Fn>0F_n>0.

For these parameters, the invariants in the divisibility theorem are M=NRRV=α=δ=2KM=NR_{\mathrm{RV}}=\alpha=\delta=2K and β=0\beta=0 before multiplication by nn. Let ΩK⊂(0,1)\Omega_K\subset(0,1) consist of the points satisfying at least one of

⌊(2K−1)ω⌋+⌊(K+1)ω⌋<⌊2Kω⌋+⌊Kω⌋,(7)\lfloor(2K-1)\omega\rfloor+\lfloor(K+1)\omega\rfloor<\lfloor2K\omega\rfloor+\lfloor K\omega\rfloor, \tag*{(7)}
⌊(K−1)ω⌋+⌊(K+1)ω⌋<2⌊Kω⌋.(8)\lfloor(K-1)\omega\rfloor+\lfloor(K+1)\omega\rfloor<2\lfloor K\omega\rfloor. \tag*{(8)}

Let Δn\Delta_n be the product of the primes ℓ>2Kn\ell>\sqrt{2Kn} with n/ℓ∈ΩKn/\ell\in\Omega_K. The divisibility result in [6] and clearing the denominator of z=q/pz=q/p give

Ln=dn2Δnp2NFn∈Z+ZLi⁡2(p/q).(9)L_n=\frac{d_n^2}{\Delta_n}p^{2N}F_n\in\mathbb{Z}+\mathbb{Z}\operatorname{Li}_2(p/q). \tag*{(9)}

Only the indicated first-period primes are used. Since ΩK\Omega_K is a finite union of rational intervals bounded away from zero, the prime number theorem applied to each interval gives

lim⁡n→∞log⁡ΔnKn=GK,GK=1K∫ΩKdωω2.(10)\lim_{n\to\infty}\frac{\log\Delta_n}{Kn}=G_K,\qquad G_K=\frac{1}{K}\int_{\Omega_K}\frac{\mathrm{d}\omega}{\omega^2}. \tag*{(10)}

Changing any of the finitely many interval endpoints does not change this limit.

A uniform lower bound for the prime gain

Lemma 3.1. For every integer K≥32K\ge32, KGK>32log⁡K−1KG_K>\frac{3}{2}\log K-1.

Proof. Put m=⌈K/4⌉m=\lceil K/4\rceil and change variables to u=Kωu=K\omega. Direct substitution into (7)–(8) shows that the following disjoint half-open intervals are admissible:

[k,k1−ε)(1≤k≤m),[k+12,k+1/21−ε/2)(0≤k<m).(11)\left[k,\frac{k}{1-\varepsilon}\right)\quad(1\le k\le m),\qquad \left[k+\frac{1}{2},\frac{k+1/2}{1-\varepsilon/2}\right)\quad(0\le k<m). \tag*{(11)}

Indeed, throughout these intervals εu<1/3\varepsilon u < 1/3. In the first family uu has integer part kk while (1−ε)u<k(1-\varepsilon)u < k and (1+ε)u<k+1(1+\varepsilon)u < k+1, which proves (8). In the second family 2u2u has integer part 2k+12k+1, whereas (2−ε)u<2k+1(2-\varepsilon)u < 2k+1 and (1+ε)u<k+1(1+\varepsilon)u < k+1, which proves (7). The same inequalities ensure that an interval in the first family ends before the next half-integer and an interval in the second family ends before the next integer.

Writing Hm=∑j=1m1/jH_m = \sum_{j=1}^{m} 1/j, integration over (11) yields KGK≥12Hm+H2mKG_K \ge\frac{1}{2}H_m + H_{2m}. The elementary estimate Hj≥log⁡(j+1)+1−log⁡2H_j \ge\log(j+1) + 1 - \log2 gives

KGK≥12log⁡(m+1)+log⁡(2m+1)+32(1−log⁡2)≥32log⁡K+32−72log⁡2>32log⁡K−1.\begin{aligned} KG_K \ge\frac{1}{2}\log(m+1) + \log(2m+1) + \frac{3}{2}(1-\log2) \\ &\ge\frac{3}{2}\log K + \frac{3}{2} - \frac{7}{2}\log2 > \frac{3}{2}\log K - 1. \end{aligned}

Here m+1≥K/4m+1 \ge K/4 and 2m+1≥K/22m+1 \ge K/2 for K≥32K \ge32; the last inequality follows, for example, from log⁡2<5/7\log2 < 5/7.

Analytic rates

Set aε=4(1−ε)1−ε/(3−ε)3−εa_\varepsilon= 4(1-\varepsilon)^{1-\varepsilon}/(3-\varepsilon)^{3-\varepsilon}, bε=εε/(1+ε)1+εb_\varepsilon= \varepsilon^\varepsilon/(1+\varepsilon)^{1+\varepsilon} and h(ε)=−εlog⁡ε−(1−ε)log⁡(1−ε)h(\varepsilon) = -\varepsilon\log\varepsilon- (1-\varepsilon)\log(1-\varepsilon). Since x+(z−x)y≥zyx+(z-x)y \ge zy, the exponential part of (5) is bounded by z−1+εx1−ε(1−x)2yε(1−y)z^{-1+\varepsilon}x^{1-\varepsilon}(1-x)^2y^\varepsilon(1-y). Its two one-variable maxima are aεa_\varepsilon and bεb_\varepsilon. The remaining weight [x+(z−x)y]−1[x+(z-x)y]^{-1} is integrable on the square: near the only possible singular corner its denominator is comparable with x+yx+y. Consequently

lim sup⁡n→∞J0,n1/N≤z−1+εaεbε.(12)\limsup_{n\to\infty} J_{0,n}^{1/N} \le z^{-1+\varepsilon}a_\varepsilon b_\varepsilon. \tag*{(12)}

For the residue integral, (N+iN−n)/(NN−n)=∏j=1i(N+j)/(n+j)≤Ki\binom{N+i}{N-n} / \binom{N}{N-n} = \prod_{j=1}^{i}(N+j)/(n+j) \le K^i. Applying the binomial theorem in (6), using max⁡[0,1]x(1−x)2=4/27\max_{[0,1]}x(1-x)^2 = 4/27, and using the finite entropy bound (NN−n)≤exp⁡(Nh(ε))\binom{N}{N-n} \le\exp(Nh(\varepsilon)) gives

lim sup⁡n→∞∣J1,n∣1/N≤427(z−1)exp⁡(h(ε))(1+Kz−1).(13)\limsup_{n\to\infty}|J_{1,n}|^{1/N} \le\frac{4}{27(z-1)}\exp(h(\varepsilon))\left(1+\frac{K}{z-1}\right). \tag*{(13)}

Together with (10) and log⁡d22Kn/(Kn)→4\log d_{2}^{2}K_n/(K n) \to4, these estimates show that Ln→0L_n \to0 provided both

E0=4−GK+2log⁡p−(1−ε)log⁡z+log⁡aε+log⁡bε,(14)E_0 = 4-G_K+2\log p-(1-\varepsilon)\log z+\log a_\varepsilon+\log b_\varepsilon, \tag*{(14)}
E1=4−GK+2log⁡p+log⁡(4/27)−log⁡(z−1)+h(ε)+log⁡(1+Kz−1)(15)E_1 = 4-G_K+2\log p+\log(4/27)-\log(z-1)+h(\varepsilon)+\log\left(1+\frac{K}{z-1}\right) \tag*{(15)}

are negative. For fixed pp, KK, both expressions decrease as zz increases. It therefore suffices to check z=C0p2z=C_0p^2.

Large and small numerators

For p≥80p \ge80, take K=2pK=2p. The elementary logarithmic estimates log⁡aε≤log⁡(4/27)+εlog⁡4\log a_\varepsilon\le\log(4/27)+\varepsilon\log4, log⁡bε≤−εlog⁡K−ε\log b_\varepsilon\le-\varepsilon\log K-\varepsilon, and h(ε)≤εlog⁡K+εh(\varepsilon) \le\varepsilon\log K+\varepsilon apply. The second follows from (1+ε)log⁡(1+ε)≥ε(1+\varepsilon)\log(1+\varepsilon)\ge\varepsilon, and the third from −(1−ε)log⁡(1−ε)≤ε-(1-\varepsilon)\log(1-\varepsilon)\le\varepsilon. For the first, differentiate the logarithm of aεa_\varepsilon: its derivative is log⁡((3−ε)/(1−ε))≤log⁡4\log((3-\varepsilon)/(1-\varepsilon))\le\log4 when 0≤ε≤1/30\le\varepsilon\le1/3. Lemma 3.1 now gives

E0≤−12log⁡p+log⁡(C0/2)2p<0,(16)E_0 \le\frac{-\frac{1}{2}\log p+\log(C_0/\sqrt{2})}{2p}<0, \tag*{(16)}
E1≤−12log⁡(2p)+2+2p(1+2p)8p2−12p<−12log⁡160+2.512p<0.(17)E_1 \le\frac{-\frac{1}{2}\log(2p)+2+\frac{2p(1+2p)}{8p^2-1}}{2p}<\frac{-\frac{1}{2}\log160+2.51}{2p}<0. \tag*{(17)}

The first strict inequality uses C0<9C_0 < 9, and the second uses C0>8C_0 > 8, log⁡(1+s)≤s\log(1+s) \le s, and p≥80p \ge80. All constants have substantial rational margins; the accompanying exact checks use 8.08<C0<8.098.08 < C_0 < 8.09.

For 3≤p≤793 \le p \le79, enumerate the rational breakpoints of the floor functions in (3.3)–(3.4), and integrate ω−2\omega^{-2} on the accepted intervals. Table 2 gives the resulting lower bounds and negative upper bounds at z=C0p2z=C_0p^2. After substituting the boundary z=C0p2z=C_0p^2, the reduced

ppKKlower bound for GKG_Kupper bound for E0E_0upper bound for E1E_1
3–460.547222−0.0220−0.0024
5–12140.331074−0.0089−0.0014
13–58470.137880−0.00050−0.00030
59–791140.068567−0.0142−0.0141

Table 2. Four exact rational certificates for the finite positive range.

expression for E0E_0 increases with pp, whereas that for E1E_1 decreases. Thus only the upper and lower endpoint of each numerator block are required. This is distinct from the decrease in zz with pp fixed used above.

For p=1,2p=1,2, the inequalities q>C0p3q>C_0p^3 imply respectively q≥9,65q\ge9,65, which lie in the ranges already covered by [6]. In every remaining case, (9) gives positive linear forms tending to zero along odd nn. Lemma 2.1 completes the proof of Theorem 1.1.

The negative threshold

The aim of this section is to establish a common decay estimate for three adjacent negative-argument forms and optimize the resulting condition. Put Z=q/p>1Z=q/p>1 and z=−Zz=-Z. Let PnP_n be the Legendre polynomial normalized by Pn(1)=1P_n(1)=1, and define Wn−1(u)=∑k=1nPk−1(u)Pn−k(u)/kW_{n-1}(u)=\sum_{k=1}^{n}P_{k-1}(u)P_{n-k}(u)/k and u=(z+x)/(z−x)=(Z−x)/(Z+x)u=(z+x)/(z-x)=(Z-x)/(Z+x). For n≥1n\ge1 set

gn(x)=xn(z−x)n+1[−Pn(u)log⁡x−2Wn−1(u)],Fn,r=∫01(1−x)2n+rgn(x) dx,(18)g_n(x)=\frac{x^n}{(z-x)^{n+1}}\left[-P_n(u)\log x-2W_{n-1}(u)\right],\qquad F_{n,r}=\int_0^1(1-x)^{2n+r}g_n(x)\,\mathrm{d}x, \tag*{(18)}

where r=0,1,2r=0,1,2 and gn(0)=0g_n(0)=0 by continuity.

Arithmetic and a common majorant

For z>1z>1, the inner real integral of the Rhin–Viola family is 2(z−x)−n−1Qn(u)2(z-x)^{-n-1}Q_n(u), and its normalized residue is (z−x)−n−1Pn(u)(z-x)^{-n-1}P_n(u). These are the classical integral formulas for the Legendre functions [9 (14.7.4), (14.7.6)–(14.7.7), and (14.12.11)], obtained by nn integrations by parts from Rodrigues’s formula. Since 2Qn(u)=Pn(u)log⁡((u+1)/(u−1))−2Wn−1(u)2Q_n(u)=P_n(u)\log((u+1)/(u-1))-2W_{n-1}(u) and (u+1)/(u−1)=z/x(u+1)/(u-1)=z/x, subtracting the log⁡z\log z residue term gives Fn,r=z2nIz(2n+r,n,n,n,n)F_{n,r}=z^{2n}I_z(2n+r,n,n,n,n).

In this corrected Legendre representation the logarithm in the integrand is log⁡x\log x, independent of zz. Set t=1/zt=1/z. The denominator is a power of 1−tx1-tx, whose reciprocal is analytic as a continuous-function-valued map for ∣t∣<1\lvert t\rvert<1. The factor xnlog⁡xx^n\log x is continuous at zero. Multiplying the polynomial identity by t2n+rt^{2n+r} clears its apparent pole at t=0t=0, and both sides are real analytic throughout (−1,1)(-1,1). Their agreement for 0<t<10<t<1 extends to −1<t<0-1<t<0 by the identity theorem. This also matches the negative-argument normalization of [7 Theorem 2.1], without a separate log⁡(−Z)\log(-Z) term in (18).

The corresponding invariants in (2) are H=K=δ=2n+rH=K=\delta=2n+r, α=2n\alpha=2n and β=0\beta=0. Since r≤2r\le2, a common denominator gives

d2n+22+p2n+2Fn,r∈Z+ZLi⁡2(−p/q),r=0,1,2.(19)d_{2n+2}^2+p^{2n+2}F_{n,r}\in\mathbb{Z}+Z\operatorname{Li}_2(-p/q),\qquad r=0,1,2. \tag*{(19)}

Here the degree bound in (2.1) is what makes the factor p2n+2p^{2n+2} sufficient.

For 0≤x≤10 \le x \le1, we have u∈[−1,1]u \in[-1,1]. The classical bound ∣Pk(u)∣≤1|P_k(u)| \le1 gives ∣Wn−1(u)∣≤Hn|W_{n-1}(u)| \le H_n. Therefore, if M−(Z)=max⁡0≤x≤1x(1−x)2/(Z+x)M_-(Z) = \max_{0 \le x \le1} x(1-x)^2/(Z+x), then

∣Fn,r∣≤M−(Z)nZ∫01[−log⁡x+2Hn] dx=1+2HnZM−(Z)n≤1+2nZM−(Z)n.(20)|F_{n,r}| \le\frac{M_-(Z)^n}{Z}\int_0^1[-\log x+2H_n]\,\mathrm{d}x=\frac{1+2H_n}{Z}M_-(Z)^n\le\frac{1+2n}{Z}M_-(Z)^n. \tag*{(20)}

The estimate is uniform in the three shifts; the coarser factor 1+2n1+2n suffices for the exponential argument.

Nonvanishing of three adjacent moments

Lemma 4.1. For each n≥3n \ge3 and z<−1z<-1, the three numbers Fn,0,Fn,1,Fn,2F_{n,0},F_{n,1},F_{n,2} do not all vanish.

Proof. Fix nn, zz and write μH=∫01(1−x)Hgn(x) dx\mu_H=\int_0^1(1-x)^H g_n(x)\,\mathrm{d}x. With J=x−ngnJ=x^{-n}g_n, the Legendre equation and the change of variable u=(z+x)/(z−x)u=(z+x)/(z-x) give x(z−x)J′′+[z−(2n+3)x]J′−(n+1)2J=0x(z-x)J''+[z-(2n+3)x]J'-(n+1)^2J=0. To obtain this equation, substitute the logarithmic representation of QnQ_n into its ordinary Legendre equation and use the equation for PnP_n. The resulting inhomogeneous equation for Wn−1W_{n-1} is a polynomial identity, so it extends from u>1u>1 to all real uu. In the expression for JJ, the logarithmic cross terms then cancel. Conjugating by xnx^n gives

x2(z−x)gn′′+[(1−2n)zx−3x2]gn′+(n2z−x)gn=0.(21)x^2(z-x)g_n''+\bigl[(1-2n)zx-3x^2\bigr]g_n'+(n^2z-x)g_n=0. \tag*{(21)}

For H≥2H \ge2, integrate (4.4) against (1−x)H(1-x)^H and transfer the derivatives to the polynomial coefficients. The resulting recurrence is

0=H(H−1)(z−1)μH−2+H[3H−(2H+2n+1)z]μH−1+[z(H+n+1)2−3H(H+1)−1]μH+(H+1)2μH+1.(22)\begin{aligned} 0={}&H(H-1)(z-1)\mu_{H-2}+H[3H-(2H+2n+1)z]\mu_{H-1}\\ &+[z(H+n+1)^2-3H(H+1)-1]\mu_H+(H+1)^2\mu_{H+1}. \tag*{(22)} \end{aligned}

For n≥3n \ge3, the continuous extension of gng_n at zero is twice continuously differentiable: its only potentially singular factor is xnlog⁡xx^n\log x. At zero the polynomial coefficients and the required first-order boundary combinations vanish; at one the weight and its first derivative vanish. Thus every boundary term in the two integrations by parts is zero.

If μ2n=μ2n+1=μ2n+2=0\mu_{2n}=\mu_{2n+1}=\mu_{2n+2}=0, the nonzero coefficient (H+1)2(H+1)^2 in (4.5) propagates the vanishing to all H≥2nH \ge2n. Hence the continuous function t2ngn(1−t)t^{2n}g_n(1-t) has zero integral against every polynomial. Uniform polynomial approximation, applied to this function itself, implies that the integral of its square is zero, so it vanishes identically. On the other hand, Pn(1)=1P_n(1)=1 and Wn−1(1)=HnW_{n-1}(1)=H_n give gn(x)=z−n−1xn[−log⁡x−2Hn+O(x∣log⁡x∣)]g_n(x)=z^{-n-1}x^n[-\log x-2H_n+O(x|\log x|)] as x↓0x\downarrow0. This function is nonzero for all sufficiently small positive xx, a contradiction. □\square

Optimization and completion

The nnth root of the common factor d2n+2p2n+2d_{2n+2}p^{2n+2} tends to e4p2e^4p^2. Thus (4.3), Lemma 4.1 and Lemma 2.1 prove irrationality whenever e4p2M−(Z)<1e^{4p^2}M_-(Z)<1. Put A=e4p2A=e^{4p^2}. This condition is equivalent to Z>max⁡[0,1][Ax(1−x)2−x]Z>\max_{[0,1]}[Ax(1-x)^2-x]. Since A>9A>9, the maximum is attained at x∗=(2−1+3/A)/3x_*=(2-\sqrt{1+3/A})/3 and equals

2A27[1+(1+3A)3/2]−23.(23)\frac{2A}{27}\left[1+\left(1+\frac{3}{A}\right)^{3/2}\right]-\frac{2}{3}. \tag*{(23)}

To check global maximality, differentiate the cubic, compare its two critical points with [0,1][0,1], and compare the interior value with the endpoint values 00 and −1-1. Substitution gives (4.6); it is greater than 11 for A≥e4A \ge e^4. Consequently the condition itself ensures Z>1Z>1.

Multiplying (4.6) by pp gives (1.1). Finally, (1+t)3/2<1+3t/2+3t2/8(1+t)^{3/2}<1+3t/2+3t^2/8 for t>0t>0: the third derivative of (1+t)3/2(1+t)^{3/2} is negative, so the quadratic Taylor polynomial at zero is a strict upper bound. This proves the first simpler bound. The correction −p/3+1/(4e4p)-p/3+1/(4e^{4p}) is negative for p≥1p\ge1, proving the second. Theorem 1.2 follows.

Interpolation for fixed numerators

This section derives the first four rows of Table 1 from existing certified profiles and the convexity of positive partition sums. For each nn, the partition in [11] can be written as a summable series Zn(t)=∑λan,λtwn,λZ_n(t)=\sum_\lambda a_{n,\lambda}t^{w_{n,\lambda}}, with an,λ≥0a_{n,\lambda}\ge0 and wn,λ≥0w_{n,\lambda}\ge0. Hölder’s inequality therefore gives

Zn(t01−θt1θ)≤Zn(t0)1−θZn(t1)θ,0≤θ≤1.(24)Z_n\left(t_0^{1-\theta}t_1^\theta\right)\le Z_n(t_0)^{1-\theta}Z_n(t_1)^\theta,\qquad0\le\theta\le1. \tag*{(24)}

Together with monotonicity in tt, this remains an upper bound whenever t≤t01−θt1θt\le t_0^{1-\theta}t_1^\theta. Endpoint rates B0B_0, B1B_1 consequently give the rate (1−θ)B0+θB1(1-\theta)B_0+\theta B_1 at tt.

The four rational choices in Table 3 satisfy the required logarithmic inequality. The endpoint rates are the exact rational bounds extracted from the corresponding certificates in [11]. Their source identifiers and exact inequalities are retained in the ancillary files. For

t0t_0t1t_1θ\thetattusable rate BBheight cutoff CC
3/563/562/172/1723/100023/10003/553/55−193/100-193/100197/200197/200
3/623/623/563/564/254/253/613/61−1057/500-1057/500451/500451/500
4/1834/1833/623/62101/1000101/10004/1694/169−3251/1000-3251/100099/20099/200
4/1834/1833/623/627/2007/2004/1784/178−3333/1000-3333/100093/20093/200

Table 3. Interpolation data. The second and fourth rows are used for negative values of the dilogarithm.

the indicated sign and target t=p/Qt=p/Q, exact logarithmic bounds prove B+log⁡(Q/p)<C<Rσ(p,Q)+log⁡(Q/p)B+\log(Q/p)<C<R_\sigma(p,Q)+\log(Q/p). For q≥Q≥3pq\ge Q\ge3p, the function Rσ(p,q)+log⁡(q/p)R_\sigma(p,q)+\log(q/p) is increasing, so Proposition 2.2 gives the entire denominator range. Its derivative is (17q−44p)/(12q(q−p))(17q-44p)/(12q(q-p)) for the positive sign and (17q+44p)/(12q(q+p))(17q+44p)/(12q(q+p)) for the negative sign.

The preceding endpoints are improved to 54, 60, 168 by 3/54=1/183/54=1/18, −3/60=−1/20-3/60=-1/20 and 4/168=1/424/168=1/42. Similarly, the source thresholds 13 and 17 for numerators 2 and −2-2 include 12 and 16 after 2/12=1/62/12=1/6 and −2/16=−1/8-2/16=-1/8. The unit-fraction results in the pinned source [11], together with the classical values at ±1\pm1, supply the first row. This establishes the first four rows of Table 1.

A quadratic comparison measure

The aim of this section is to obtain an analytic rate below −4.2361-4.2361 at t=5/418t=5/418 and transfer it to the last three rows of Table 1. Set a=12/25a=12/25, L=121/50L=121/50, D=L−a=97/50D=L-a=97/50, and w=9/20w=9/20. Define k1=2(1−a)(1−w)/D2=1430/9409k_1=2(1-a)(1-w)/D^2=1430/9409 and k2=3(1−a)w/D3=87750/912673k_2=3(1-a)w/D^3=87750/912673, and let

ρ(x)={1,0≤x≤a,k1(L−x)+k2(L−x)2,a<x≤L,0,otherwise.(25)\rho(x)= \begin{cases} 1, & 0\le x\le a,\\ k_1(L-x)+k_2(L-x)^2, & a<x\le L,\\ 0, & \text{otherwise}. \end{cases} \tag*{(25)}

The tail mass is 1−a1-a, and its maximum is k1D+k2D2<1k_1D+k_2D^2<1. Thus ρ\rho is a probability density bounded by 1. Define its logarithmic potential and energy by U(x)=∫ρ(t)log⁡∣x−t∣ dtU(x)=\int\rho(t)\log|x-t|\,dt and E=∫U(x)ρ(x) dxE = \int U(x)\rho(x)\,\mathrm{d}x. Put

H(x)=3(x+1)log⁡(x+1)−2xlog⁡x−(x+4)log⁡(x+4)+1,H(x) = 3(x+1)\log(x+1) - 2x\log x - (x+4)\log(x+4) + 1,
f(x)=2U(x)+12H(x)−12xlog⁡(418/5),d=−539387/100000,f(x) = 2U(x) + \frac{1}{2}H(x) - \frac{1}{2}x\log(418/5), \qquad d = -539387/100000,
M=d+∫0∞(f(x)−d)+ dx,(26)M = d + \int_{0}^{\infty} (f(x)-d)_{+}\,\mathrm{d}x, \tag*{(26)}

with the usual continuous convention 0log⁡0=00\log0 = 0.

Exact potential and energy certificates

Lemma 6.1. For the density (6.1), E>−1.02905E > -1.02905 and M<−5.16753M < -5.16753. In particular,

12log⁡2+4+4(M−E)<−4.2361.(27)12\log2 + 4 + 4(M-E) < -4.2361. \tag*{(27)}

Proof. The potential, its primitive, and the energy are finite rational linear combinations of powers and logarithms at rational arguments; the formulas and certification method are given in Appendix § 7. The positive part in (6.2) is certified by five rational root boxes and a complete sign cover of their complement. On each root box we use the coarse bound (f−d)+<32(f-d)_{+} < 32. Their total length is 15⋅10−815 \cdot10^{-8}, so their combined contribution is less than 4.8⋅10−64.8 \cdot10^{-6}. Integrating the primitive on the three certified positive components and adding this error gives the asserted bound for MM. The energy formula gives the stated lower bound for EE. All logarithmic bounds are obtained from rational series with explicit remainders. Finally 12log⁡2+4+4(M−E)<−4.236112\log2 + 4 + 4(M-E) < -4.2361 follows by rational interval arithmetic.

Passage to finite circles

The finite-circle comparison theorem of [11] applies to disjoint circles with positive rational centres. The next argument supplies such a fixed family from (6.1).

Proposition 6.2. For all sufficiently large nn, Zn(5/418)≤exp⁡(−4.236n2)Z_n(5/418) \le\exp(-4.236n^2).

Proof. For a positive integer NN, take the NN midpoint quantiles xjx_j of ρ(x) dx\rho(x)\,\mathrm{d}x. Since ρ≤1\rho\le1, consecutive quantiles are separated by at least 1/N1/N and x1≥1/(2N)x_1 \ge1/(2N). Choose rational centres cjc_j with ∣cj−xj∣<1/(100N)\lvert c_j-x_j\rvert< 1/(100N) and a common radius rN=1/(4N)r_N = 1/(4N). The centres are positive and bounded by 16, the circles are disjoint, and their minimum separation is at least 0.98/N0.98/N. Assign equal mass to the circles. Their potential on the real axis is

UN(x)=1N∑j=1Nlog⁡max⁡(rN,∣x−cj∣).(28)U_N(x) = \frac{1}{N}\sum_{j=1}^{N}\log\max(r_N,\lvert x-c_j\rvert). \tag*{(28)}

We use the subsequence N=m2N = m^2 and put δ=1/m\delta= 1/m. The kernel log⁡max⁡(δ,∣x−t∣)\log\max(\delta,\lvert x-t\rvert) is 1/δ1/\delta-Lipschitz in tt, uniformly for real xx. Midpoint-quantile quadrature and the rational perturbations therefore give a uniform error at most (L+0.01)/(Nδ)(L+0.01)/(N\delta). Removing the truncation changes the continuous potential by at most 2δ2\delta, because ρ≤1\rho\le1 and ∫−δδlog⁡(δ/∣t∣) dt=2δ\int_{-\delta}^{\delta}\log(\delta/\lvert t\rvert)\,\mathrm{d}t = 2\delta.

For the discrete potential, choose a centre nearest to xx. Every other centre, at index distance jj, is at distance at least 0.49j/N0.49j/N from xx. The inequality log⁡+t≤t\log_{+} t \le t gives the uniform truncation error bound

log⁡(4Nδ)N+2δ0.49HN.(29)\frac{\log(4N\delta)}{N} + \frac{2\delta}{0.49}H_N. \tag*{(29)}

Since HN≤1+log⁡NH_N \le1+\log N, all three errors tend to zero for N=m2N = m^2. Thus UN→UU_N \to U uniformly on the entire real axis. By the mean-value identity for logarithmic potentials of disjoint circles, their energy is EN=N−1∑jUN(cj)E_N=N^{-1}\sum_j U_N(c_j). Uniform convergence and quantile quadrature for the continuous function UU give EN→EE_N\to E.

The exact inequalities 2/(4−L)−12log⁡(418/5)<02/(4-L)-\frac{1}{2}\log(418/5)<0 and f(4)<df(4)<d imply that ff decreases on [4,∞)[4,\infty). Indeed, HH is decreasing there, and for x≥y≥4x\ge y\ge4 the logarithmic inequality log⁡u−log⁡v≤(u−v)/v\log u-\log v\le(u-v)/v gives U(x)−U(y)≤(x−y)/(4−L)U(x)-U(y)\le(x-y)/(4-L). It follows that g=(f−d)+g=(f-d)_{+} is continuous and supported in [0,4][0,4] on the nonnegative axis.

Fix ε>0\varepsilon>0 small enough to preserve the strict gap between (27) and −4.236-4.236. Choose and fix one sufficiently large N=m2N=m^2 for which both the potential and energy errors are smaller than ε\varepsilon. The corresponding profile satisfies fN≤d+g+2εf_N\le d+g+2\varepsilon on the entire nonnegative axis. Since gg is continuous and compactly supported, its equally spaced Riemann sums converge to its integral. Each positive integer node occurs at most twice among the 2n2n effective nodes of the determinant construction. Hence their total profile contribution is at most 2n(M+3ε)2n(M+3\varepsilon) for all sufficiently large nn.

Substitute this estimate in the finite-circle comparison inequality of [11]. Its remaining errors are O(nlog⁡n)O(n\log n) for the fixed family of circles. The choice of ε\varepsilon and the strict rate gap give the asserted eventual bound. The order of choices is essential: NN is fixed before the determinant order nn tends to infinity.

Endpoint inequalities and complete denominator ranges

With B=−4.236B=-4.236, rational logarithmic bounds show B+log⁡(418/5)<1901/10000<Rσ(p,Q)+log⁡(Q/p)B+\log(418/5)<1901/10000<R_{\sigma}(p,Q)+\log(Q/p) at the six triples

(p,Q,σ)=(5,418,1),(5,434,−1),(6,1039,1),(6,1058,−1),(7,2234,1),(7,2256,−1).(30)(p,Q,\sigma)=(5,418,1),(5,434,-1),(6,1039,1),(6,1058,-1),(7,2234,1),(7,2256,-1). \tag*{(30)}

Since p/Q≤5/418p/Q\le5/418, Proposition 2.2 applies. The monotonicity used in § 5 then gives every denominator beyond each endpoint. The equalities 6/1038=1/1736/1038=1/173 and 7/2233=1/3197/2233=1/319 add the two preceding positive endpoints. This proves the remaining rows of Table 1. The additional rows of Theorem 1.3 follow from the cubic bounds and the exact reductions described in Appendix § 7.

Exact certificates and reproducibility

The aim of this appendix is to describe the finite data and exact arithmetic supporting the numerical inequalities in the proofs. All displayed terminating decimals in inequalities are interpreted as rational numbers. To bound a logarithm, reduce its argument to a fixed compact interval by extracting powers of 22, and use log⁡x=2∑j=0m−1v2j+12j+1+Rm\log x=2\sum_{j=0}^{m-1}\frac{v^{2j+1}}{2j+1}+R_m, where v=(x−1)/(x+1)v=(x-1)/(x+1) and ∣Rm∣≤2∣v∣2m+1(2m+1)(1−v2)\lvert R_m\rvert\le\frac{2\lvert v\rvert^{2m+1}}{(2m+1)(1-v^2)}. This gives explicit rational upper and lower bounds. The generators output exact rational inequalities and complete interval covers for independent checking.

Quadratic potential formulas

For v≠0v\ne0, put T(v)=vlog⁡∣v∣−vT(v)=v\log\lvert v\rvert-v, J(v)=12v2log⁡∣v∣−34v2J(v)=\frac{1}{2}v^2\log\lvert v\rvert-\frac{3}{4}v^2, N(v)=16v3log⁡∣v∣−1136v3N(v)=\frac{1}{6}v^3\log\lvert v\rvert-\frac{11}{36}v^3, and O(v)=124v4log⁡∣v∣−25288v4O(v)=\frac{1}{24}v^4\log\lvert v\rvert-\frac{25}{288}v^4, extending all four continuously at zero. These satisfy J′=TJ'=T, N′=JN'=J and O′=NO'=N. With c=k1D+k2D2−1c=k_1D+k_2D^2-1, integration of (25) gives

U(x)=T(x)+cT(x−a)−(k1+2k2D)J(x−a)+2k2N(x−a)+k1J(x−L)−2k2N(x−L),\begin{aligned} U(x)&=T(x)+cT(x-a)-(k_1+2k_2D)J(x-a)+2k_2N(x-a)\\ &\quad+k_1J(x-L)-2k_2N(x-L), \end{aligned}
V(x)=J(x)+cJ(x−a)−(k1+2k2D)N(x−a)+2k2O(x−a)+k1N(x−L)−2k2O(x−L).\begin{aligned} V(x)&=J(x)+cJ(x-a)-(k_1+2k_2D)N(x-a)+2k_2O(x-a)\\ &\quad+k_1N(x-L)-2k_2O(x-L). \end{aligned}

where V′=UV' = U. If W(v)=12v2log⁡∣v∣−14v2W(v) = \frac{1}{2}v^2\log|v| - \frac{1}{4}v^2, a primitive of ff is 2V(x)+12[3W(x+1)−2W(x)−W(x+4)+x]−14x2log⁡(418/5)2V(x) + \frac{1}{2}[3W(x+1) - 2W(x) - W(x+4) + x] - \frac{1}{4}x^2\log(418/5). These formulas follow by integrating the piecewise quadratic density; the continuous values at aa and LL fix the constants.

For the energy, put A(v)=13v3log⁡∣v∣−49v3A(v) = \frac{1}{3}v^3\log|v| - \frac{4}{9}v^3 and P(v)=14v4log⁡∣v∣−516v4P(v) = \frac{1}{4}v^4\log|v| - \frac{5}{16}v^4, with continuous extensions at zero. Thus A′(v)=vT(v)A'(v) = vT(v) and P′(v)=v2T(v)P'(v) = v^2T(v). Set

C1=L[J(L)−J(a)]−[A(L)−A(a)]−DJ(D)+A(D),C_1 = L[J(L)-J(a)] - [A(L)-A(a)] - DJ(D) + A(D),
C2=L2[J(L)−J(a)]−2L[A(L)−A(a)]+P(L)−P(a)−D2J(D)+2DA(D)−P(D).C_2 = L^2[J(L)-J(a)] - 2L[A(L)-A(a)] + P(L)-P(a) - D^2J(D) + 2DA(D) - P(D).

The initial interval, the two cross-interactions and the tail self-interaction give

E=a2log⁡a−32a2+2k1C1+2k2C2+(1−a)2[log⁡D−74−w6−w212].E = a^2\log a - \frac{3}{2}a^2 + 2k_1C_1 + 2k_2C_2 + (1-a)^2\left[\log D - \frac{7}{4} - \frac{w}{6} - \frac{w^2}{12}\right].

For example, the normalized linear and quadratic tail densities on [0,1][0,1] are 2(1−t)2(1-t) and 3(1−t)23(1-t)^2. Their self-energies are −7/4-7/4 and −2-2, and their mutual energy is −11/6-11/6; weighting them by 1−w1-w and ww gives the last bracket. These values follow by splitting the square into s<ts<t and s>ts>t and integrating powers times logarithms. All arguments in the displayed energy formula are positive rationals, so the stated energy interval follows directly from rational logarithmic bounds.

The five root boxes for f−df-d are 10−810^{-8} times [41820077,41820080][41820077,41820080], [47875826,47875829][47875826,47875829], [79183885,79183888][79183885,79183888], [139988153,139988156][139988153,139988156], and [166842325,166842328][166842325,166842328]. Away from these boxes the positive components are [0,41820077/108][0,41820077/10^8], [47875829/108,79183885/108][47875829/10^8,79183885/10^8] and [139988156/108,166842325/108][139988156/10^8,166842325/10^8]. The complementary sign decisions are certified on 167 rational intervals, including the changes in the explicit formulas at aa and LL. The global cover and the tail argument ensure that no omitted component contributes to MM.

Additional threshold rows

The aim of this table is to record the remaining uniform tails from the complete threshold list. Each entry follows from Theorems 1.1 and 1.2, together with exact cancellation at the finitely many preceding denominators. The exact reduction identities are listed in the ancillary data.

Contents of the ancillary package

The package (available at https://zenodo.org/records/23247332) contains the exact positive floor-interval and logarithmic certificates, the polynomial verification of (22), the scalar and global sign certificates for (25), and the endpoint-rate data used in Table 3. It also contains a pinned copy of the source construction with its original license and attribution. Running reference/verify_all.py checks the original finite certificates. The versioned source and the explicit parameter tables make the dependence of each numerical claim reproducible.

Acknowledgments and use of artificial intelligence

We acknowledge the arithmetic integral constructions of Georges Rhin and Carlo Viola, Aabir Fauzan’s determinant approach to ζ(5)\zeta(5), Qian Tang’s adaptation to the dilogarithm, and Si-Qi Liu’s extension to rational arguments. The formal development underlying our fixed-numerator results comes through Liu’s Li2Rational, which derives from Tang’s Li2NegHalf (version of 27 September 2026). It also includes code adapted by Liu from Moritz Firsching’s Zeta5 formalization [4], notably for Andréief and Gram determinant identities, logarithmic potentials and energies, and arithmetic divisibility and nonvanishing. The original licenses, source headers and detailed attributions are retained in the ancillary package. We also acknowledge the Lean and Mathlib contributors whose work supports these developments.

The development of the arguments and estimates specific to this manuscript was carried out primarily by ChatGPT (Codex), working from the cited mathematical results and supplied source material. Its contributions included developing proof strategies, deriving intermediate lemmas, exploring and selecting parameters, constructing and checking exact certificates, writing and debugging the Lean formalization, and drafting and revising the manuscript. The authors’ role was to initiate and direct this process: they supplied the problem and source material, set the objectives, and requested successive mathematical developments, Lean verification and revisions. Thus the AI contribution includes substantive mathematical reasoning as well as computation, formalization and exposition. The authors take responsibility for the manuscript and for accurately reporting the provenance and verification status of its results.

The new estimates and deductions have been checked in Lean 4 with Mathlib and the pinned development of [11]. The general theorems retain seven explicit published or classical inputs: the least-common-multiple and finite-prime-window consequences of the prime number theorem; two specialized Rhin–Viola polynomial identities, one including their established small-numerator corollaries; the integral and logarithmic formulas for the Legendre function of the second kind; the unit-interval Legendre bound; and the Legendre differential equation. The ancillary dependency document states these inputs precisely. They are theorem parameters, and their proofs are not re-formalized here. The first seven positive and negative threshold rows, including the 418 partition estimate, are checked using the pinned source without these external parameters. The complete project compiles, and the public-theorem audit reports only the standard foundational axioms of Lean; the new source contains no admitted proofs or additional axioms.

References

  1. [1]F. Beukers, A note on the irrationality of ζ(2) and ζ(3), Bull. London Math. Soc. 11 (1979), 268–272.
  2. [2]H. Davenport, Multiplicative Number Theory, 2nd ed., Graduate Texts in Mathematics 74, Springer, New York, 1980.DOI
  3. [3]A. Fauzan, ζ(5) is irrational, preprint, Zenodo, version 1, 17 September 2026. doi:10.5281/zenodo.22826419.DOI
  4. [4]M. Firsching, Zeta5, software source (2026), commit f19a1960609f. https://github.com/mo271/Zeta5.
  5. [5]M. Hata, Rational approximations to the dilogarithm, Trans. Amer. Math. Soc. 336 (1993), 363–387.DOI
  6. [6]G. Rhin and C. Viola, The permutation group method for the dilogarithm, Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) 4 (2005), 389–437.
  7. [7]G. Rhin and C. Viola, Linear independence of 1, Li1 and Li2, Moscow J. Comb. Number Theory 8 (2019), 81–96.DOI
  8. [8]T. Prellberg, Three irrationality results for the dilogarithm: Rhin–Viola and Viola–Zudilin constructions at −1/4, 1/5, and −1/3, preprint (2026), arXiv:2609.06895.arxiv.org/abs/2609.06895
  9. [9]NIST Digital Library of Mathematical Functions, Chapters 14 and 18, in particular Sections 14.2, 14.7, 14.12 and 18.14. https://dlmf.nist.gov/14.
  10. [10]Q. Tang, Irrationality of Li2(1/2), software source (2026). https://github.com/dtq1997/li2-half-irrationality.
  11. [11]S.-Q. Liu, Li2Rational, software source, commit b2a61c46c157. https://github.com/siqiliu-tsinghua/Li2Rational.

Paper details

vlnc:W_qvuam8sd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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