Let η>1 be a unit of a real quadratic order and Un=(ηn−ηˉn)/(η−ηˉ) its Lucas sequence. We prove that ∑n≥1Un−3 is irrational. Equivalently, ∑n≥1Un(P,Q)−3∈/Q for the Lucas sequences with Q=−1, P≥1, and with Q=1, P≥3. In particular the sums of the reciprocal cubes of the Pell numbers and of the numbers Fkn, for every k≥1, are irrational. The case of the Fibonacci numbers was treated in earlier work of the author; the proof here is the same Hankel-determinant construction, carried out for a general unit and for both signs of its norm. The constants of the argument do not depend on η once they are measured in units of logη. The whole argument is formalised in Lean 4 with Mathlib.
Introduction
For integers P,Q let Un=Un(P,Q) be the Lucas sequence U0=0, U1=1, Un+1=PUn−QUn−1. When Q=±1 and D=P2−4Q>0 is not a square, Un=(ηn−ηˉn)/D with η=(P+D)/2 a unit of the order Z[η] of norm ηηˉ=Q. Conversely, every unit η>1 of a real quadratic order arises in this way, with P=η+ηˉ≥1 if Q=−1 and P≥3 if Q=1. The Fibonacci numbers are Un(1,−1), the Pell numbers Pn are Un(2,−1), and Fkn/Fk=Un(Lk,(−1)k), where Lk is the k-th Lucas number.
Theorem 1.1. Let Q=−1 and P≥1, or Q=1 and P≥3. Then ∑n≥1Un(P,Q)31 is irrational.
Corollary 1.2. The numbers ∑n≥1Pn−3 and, for every k≥1, ∑n≥1Fkn−3 are irrational.
Proof. Theorem 1.1 with (P,Q)=(2,−1), and with (P,Q)=(Lk,(−1)k), which is admissible: Lk≥1, and Lk≥3 for k≥2. Then Un=Fkn/Fk by Binet’s formula, so ∑nFkn−3=Fk−3∑nUn−3. □
The case (P,Q)=(1,−1), ∑nFn−3∈/Q, is the main result of the preprint [7]; we refer there for the history of the Fibonacci zeta function. Briefly:
Key words and phrases. Lucas sequences, Pell numbers, Fibonacci numbers, irrationality, q-series, Hankel determinants, Lean. André-Jeannin proved that ∑nUn−1 is irrational for a class of binary recurrences that includes every Un(P,±1), in particular ∑nFn−1∈/Q[1]. For both signs of Q, Duverney, Ke. Nishioka, Ku. Nishioka and Shiokawa proved that the even powers ∑nUn−2s (s≥1) are transcendental, and for Q=−1 also the odd-index sums ∑nU2n−1−s[2]; Elsner, Shimomura and Shiokawa determined the algebraic relations among such numbers [3, 4, 5]. These sums are values of quasi-modular forms at algebraic points, to which Nesterenko’s theorem [8] applies (see [13, §3.4] for the Fibonacci zeta function). The odd powers ∑nUn−(2s+1) of the full sums are not known to be of this kind. For the Pell numbers, Xu and Wang [14] evaluated the integer part of (∑k≥nPk−3)−1; we know of no irrationality result for any full sum ∑nUn−3 before the preprint [7].
The method
Put t=η−1 and w=Qt2. Since ηˉ/η=Qt2, we have Un=t−n(1−wn)/D, and so
Ξ:=DD1n≥1∑Un31=N≥1∑(1−wN)3t3N.(1)
We follow [7], which adapted to Z[φ] a construction of an unrefereed, AI-assisted research draft on the q-zeta value ∑nn2/(pn−1), p≥2 an integer [11, 10]; the construction is a moving-denominator Hankel determinant in the line of Uvarov [12], Krattenthaler [6], Prévost [9] and Zudilin [15, 16]. A linear functional L, a second Euler derivative summed against the weights (Qt)n at the nodes wn, has moments L(sk)∈Q(D), rational functions of t, and satisfies L(1/(1−was))=λa(Ξ−Sa) with explicit λa and partial sums Sa of (1.1). The Hankel determinants ΔK(X) of L against DK(s)=∏a≤K(1−was) satisfy ΔK(Ξ)=det[L(si+j/DK)], and four estimates are combined:
size (per K3logη)
real decay ∣ΔK(Ξ)∣
−2/9
Proposition 3.1
conjugate ∣σΔK(−Ξ)∣
−1
Proposition 5.1
denominators log∣NCK(t)∣
≤c∗=1.1954…<11/9
Propositions 4.4, 4.6
nonvanishing
infinitely many K
Proposition 6.1
Table 1.
If ∑nUn−3 were rational, CK(t)(mDK2)ΔK(Ξ), where m is the denominator of the sum, would be a nonzero element of Z[η] of norm <1 for large $K.
Compared with [7] there are three changes. First, the sign: for Q=1 the nodes wn=t2n are positive and the roots of unity at which the coefficients have poles are of a different type (§ 4); the local arguments are unchanged, only the classification of the roots changes, and the elementary count of the α-roots used in the proof is the same for both signs (the difference appears only in the sharp constant of Remark 4.7). Second, the unit: all analytic estimates use only 0<t≤φ−1, which holds for every admissible (P,Q), and every constant is a nondecreasing function of t on that interval.
Third, the order Z[η] need not be maximal; the argument uses only that the norm of a nonzero element of Z[η] is a nonzero rational integer. Sections 2–7 follow [7] closely, often verbatim with φ−1 replaced by t and −1 by Q; we repeat the arguments so that the present paper is self-contained.
The same argument gives ∑nUkn−3∈/Q for every admissible (P,Q) and k≥1, since Ukn=UkUn′ with U′ the Lucas sequence of the unit ηk. The normalisation η>1 does not cover the alternating sums ∑n(−1)n−1Un(P,−1)−3=∑nUn(−P,−1)−3.
Formal verification
The theorem and every step of its proof are formalised in Lean 4 with Mathlib (https://github.com/telaviv/lucas-zeta3-irrationality; Section 8). The analytic ingredients are formalised for every real t with 0<t≤φ−1 and both signs, and the integrality statement for every real τ=0 that is not a root of unity, with one integer polynomial for all such τ; only the final assembly uses that t is the inverse of a quadratic unit.
The functional and the determinants
Notation
Throughout, (P,Q) is admissible: Q=−1 and P≥1, or Q=1 and P≥3. Put D=P2−4Q, η=(P+D)/2, ηˉ=(P−D)/2=Q/η, t=η−1, O=Z[η], and let σ be the nontrivial automorphism of Q(D), so σt=1/ηˉ=Qη. We write φ=(1+5)/2, Euler’s function φE(d) and the Möbius function μ(d). Put
ρ=Qt,w=Qt2,αk=ρwk(k≥0),
ψN=(1−wN)3t3N,Sa=N=1∑aψN,
AK=K(K+1)/2 and θ=sdsd. Then ρ2=Qw, ∣w∣=t2, Ξ=∑NψN, and w/ρ=t. We write hm for the complete homogeneous symmetric polynomial of degree m (hm=0 for m<0), and K≥2.
Lemma 2.1. (i) D is not a square, so D∈/Q. (ii) η≥φ, so 0<t≤φ−1. (iii) t and σt are the roots of τ2−PQτ+Q, and tσt=Q. (iv) D=P−2Qt and P−2Qσt=−D. (v) Un=(ηn−ηˉn)/D=t−n(1−wn)/D, so 1/Un3=DDψn and (1.1) holds.
Proof. (i) For Q=−1, P2<D<(P+2)2 and D=(P+1)2 would give 2P=3; for Q=1, P≥3, (P−2)2<D<P2 and D=(P−1)2 would give 2P=5. (ii) D≥5 in both cases, and P≥1. (iii) From η2=Pη−Q we get 1=Pt−Qt2, i.e. t2−PQt+Q=0 (as Q2=1); the other root is Q/t=Qη=σt. (iv) 1/t=P−Qt, so D=η−ηˉ=1/t−Qt=P−2Qt; and P−2Q2η=P−2η=−D. (v) Binet’s formula, by induction from η2=Pη−Q and ηˉ2=Pηˉ−Q; then use ηn=t−n, ηˉn=(Qt)n=wnt−n and (D)3=DD.
Definition 2.2. For g analytic on a neighbourhood of ∣s∣≤t2 put L(g)=
n≥1∑ρn[θ(θ+1)g](wn).
The series converges absolutely since θ(θ+1)g=O(s) and ∣ρnwn∣=t3n.
Lemma 2.3. (i) L(sk)=νk:=k(k+1)αk/(1−αk) for k≥0. (ii) For a≥1, L(1/(1−was))=λa(Ξ−Sa) with λa=2ρ−a.
Proof. (i) θ(θ+1)sk=k(k+1)sk; sum the geometric series. (ii) θ(θ+1)(1−cs)−1=2cs/(1−cs)3, so the left side is ∑nρnwa+n/(1−wa+n)3=2ρ−a∑N>aρNwN/(1−wN)3, and ρNwN=Q2Nt3N=t3N.
Let DK(s)=∏a=1K(1−was), with poles xa=w−a, ∣xa∣=ηa. For a polynomial P0 of degree <2K−1,
Every quantity above is the value at τ=t of a rational function of an indeterminate τ with rational coefficients, which depends on (P,Q) only through Q: ρ=Qτ, w=Qτ2, αk=Qk+1τ2k+1, ψN=τ3N/(1−(Qτ2)N)3. Write ΔK,τ(X)∈Q(τ)[X] for the resulting polynomial, and w(ζ), αk(ζ) for the values at τ=ζ. For Q=−1 it is the polynomial of [7]. Since σt=Q/t, the coefficientwise conjugate σΔK is the value of ΔK,τ at τ=Q/t. Both values are defined because no denominator vanishes at a real number that is neither 0 nor a root of unity.
Real decay
Proposition 3.1.log∣ΔK(Ξ)∣≤−92(K3+3K2−4)logη+O(K2), with an absolute implied constant.
Proof. Let Vj(s)=∏u=1j(s−w⌈u/3⌉), Rj=⌈j/3⌉ and aj=∑u≤j⌈u/3⌉. The change of basis from 1,s,…,sK−1 to V0,…,VK−1 is unitriangular, so ΔK(Ξ)=det[L(ViVj/DK)].
Since θ(θ+1)g(s0) depends only on the 2-jet of g at s0, the m-th term of L(ViVjW), W=1/DK, vanishes for m<R:=max(Ri,Rj): then ViVj has a triple zero at wm. For m≥R and s=wm every root β of Vj satisfies ∣s∣≤∣β∣, so ∣s−β∣≤2∣β∣ and θ(s−β)=s; the signs of the nodes play no role. Hence ∣Vj(s)∣≤2j∣w∣aj and ∣θuVj(s)∣≤ju2j∣w∣aj∣s∣/∣w∣Rj for u=1,2. On ∣s∣≤t2 we have ∣was∣≤t2a+2≤φ−2a−2, so ∣W∣, ∣θW∣/∣s∣ and ∣θ2W∣/∣s∣ are bounded by an absolute constant. Since θ(θ+1)=θ2+θ kills constants, by Leibniz every term of θ(θ+1)(ViVjW) contains at least one θ on some factor, hence a factor ≤∣s∣/∣w∣R. With ∣ρ∣m≤1 and ∑m≥R∣w∣m−R≤(1−φ−2)−1,
∣L(ViVjW)∣≤c1(i+j+1)24i+j∣w∣ai+aj
for an absolute constant c1. Pull ∣w∣ai out of row i and column i and apply Hadamard’s inequality: ∣ΔK(Ξ)∣≤∣w∣2∑j<Kaj(c1(2K)2424⋅2KK)K. Finally 2∑j<Kaj=(K3+3K2−ιK)/9 with ιK∈{0,4,2} according to K mod 3, and ∣w∣=η−2.
Denominators
The integrality ring and the root types
Lemma 4.1. The coefficients of ΔK,τ(X) lie in Z[τ,τ−1,Φd(τ)−1:d∈DK], where DK is the set of orders of roots of unity ζ such that (W) w(ζ)=Qζ2 has order e≤K, or (α)αk(ζ)=1 for some 1≤k≤K−2.
Proof. The inputs are the Laurent monomials wa and λa=2Qaτ−a; the factors 1/(1−wb−a) and (1−wN)−3 (N≤K), where 1−wn=1−Qnτ2n is ± a product of cyclotomic polynomials Φd(τ) with w(ζ)n=1; and νk=k(k+1)αk/(1−αk), of which only k≤K−2 occur (the quotient has degree ≤2K−2−K), with 1−αk=1−Qk+1τ2k+1=±∏Φd(τ). Division by DK is integral because its leading coefficient ∏a(−wa)=±τK(K+1) is a unit. No rational constant is inverted.
For d∈DK the type depends only on d: every primitive d-th root is a power ζj with j coprime to 2d, and since Qj=Q, Qζ2j=(Qζ2)j and αk(ζj)=αk(ζ)j; powers coprime to the order preserve the order. If ζ has order d, the order e(d) of w(ζ) and the least k≥1 with αk(ζ)=1 are as follows.
Q=−1: e(d)=2d (d odd), d (d≡2(mod4)), d/4 (d≡4(mod8)), d/2 (d≡0(mod8)); kmin(d)=(d−1)/2 (d≡3(mod4)), (3d−1)/2 (d≡1(mod4)), (d−2)/4 (d≡2(mod8), d≥10), (3d−2)/4 (d≡6(mod8)), kmin(2)=2, and none if 4∣d.
Q=1: e(d)=d (d odd), d/2 (d even); αk(ζ)=ζ2k+1=1 forces d odd, and then kmin(d)=(d−1)/2 for d≥3, kmin(1)=1.
In both cases ∑e(d)=eϕE(d)=2ϕE(e): for each primitive e-th root ω the equation Qτ2=ω has two roots, and these 2ϕE(e) roots of unity are exactly those with e(τ)=e. All d∈DK satisfy d≤4K: w(ζ)e=1 gives ζ4e=1, and αk(ζ)=1 gives ζ4k+2=1.
Local setting. Fix ζ of order d∈DK, let ϖ=τ−ζ and let v=ordϖ on Q(ζ)((ϖ)), extended to polynomials in X as the Gauss valuation (the minimum over the coefficients; it is multiplicative). Then: (a) if e(d)=e≤K, then xa(ζ)=xb(ζ) iff e∣a−b, and then v(xa−xb)=1: indeed xa−xb=xa(1−wb−a), and 1−wn has only simple zeros, since dτdwn=0 at τ=0;
(b) ψN has a pole of order exactly 3 if e∣N and is regular otherwise, by (a); so Sa has a pole of order ≤3, and is regular if a<e;
(c) 1−αk=1−Qk+1τ2k+1 is separable, so it has at most a simple zero at ζ;
(d) λa and xa are Laurent monomials, hence units.
We call ζ a W-root if e(d)≤K and an α-root if αk(ζ)=1 for some 1≤k≤K−2; a pure W-root is a W-root that is not an α-root.
Local bounds
Put c(m)=2∑j=0m−2max(0,2m+2−j) and c′(m)=2∑j=0m−1max(0,2m+2−j); thus c(1)=0, c(m)=4(m−1) for m≤6, c(m)≤(m+3)2/4, c′(1)=3 and 0≤c′−c≤3.
Proposition 4.2. Let e=e(d)≤K and partition {1,…,K} into the clusters Ir={a≡rmode} of sizes mr, 1≤r≤e. Then
with δα=1 if ζ is also an α-root and δα=0 otherwise.
Proof. The proof of [7 Proposition 4.2] uses only the facts (a)–(d) and the integrality of the quotient by DK, which hold for both signs; we repeat it. Let F(s)=∏a(s−xa) and Fr=∏a∈Ir(s−xa). Fix a cluster r, write m=mr and Ir={a1<⋯<am}.
Newton cluster basis. For 0≤j<m put Nr,j=(F/Fr)∏l=m−j+1m(s−xal); it vanishes at the top j nodes of the cluster. At ϖ=0 these reduce to (F/Fr)(s−yr)⋅(unit), yr=xr(ζ), which form a basis of the polynomials of degree <K by the Chinese remainder theorem (distinct residues r give distinct points yr). So the change of basis from 1,…,sK−1 has unit determinant, and v(ΔK,τ) is the valuation of G=[LX(NN′/DK)] in this basis.
Entries.G(r,j),(r′,k)=Ppol+∑aresa(NN′)λa(X−Sa), where Ppol is LX of the polynomial part. (i) If r=r′ then F∣Nr,jNr′,k and all resa=0. (ii) If r=r′, then resai=0 unless i≤m−max(j,k), and for such i
Here (F/Fr)(xai) is a unit, each displayed difference has valuation exactly 1 by (a), and the denominator has valuation exactly m−1. So v(resai)=j+k−(m−1). (iii) v(λa)=0, and by (b) the Gauss valuation of X−Sa is ≥−3 if a≥e and ≥0 if a<e. (iv) For r<e the node a1=r is regular and ai is singular for i≥2; for r=e every node is singular. So the proper part of the (j,k) entry has v≥j+k−(m−1)−3 if some singular node has i≤m−max(j,k), and v≥j+k−(m−1) otherwise. The second case occurs exactly when r<e and max(j,k)=m−1.
Weights. Let κj=max(0,2m+2−j) for j≤m−2 and κm−1=0 if r<e, and κj=max(0,2m+2−j) for all j<m if r=e. Then (A) if the (j,k) entry has a singular node, κj+κk≥m+2−j−k, and (B) if r<e and max(j,k)=m−1, κj+κk≥0≥m−1−j−k. Indeed, in case (A) if both weights are positive the inequality is an equality; if κk=0, then k≥2m+2, so κj≥2m+2−j≥m+2−j−k; if both vanish, the right side is ≤0. Case (B) holds because j+k≥m−1. Scale row (r,j) by ϖ⌈κj⌉ and column (r,j) by ϖ⌊κj⌋. Since a+b≥n∈Z implies ⌈a⌉+⌊b⌋≥n, every scaled proper entry is integral. Since κj∈21Z, ⌈κj⌉+⌊κj⌋=2κj, so the total scaling is 2∑κj, which is c(m) for r<e and c′(m) for r=e. All exponents are ≥0.
Polynomial part. Write (⋅)+ for the polynomial part (the quotient by DK). The quotient of NN′ has degree ≤K−2 and coefficients in Z[τ±1], so Ppol=∑k≤K−2[sk](NN′/DK)+νk (with ν0=0). At a pure W-root every νk, $1\le k\le K-2$, is regular, so the scaled matrix is integral; this gives the bound with δα=0. At an α-root, put Pα={1≤k≤K−2:αk(ζ)=1} and write νk=ϖ−1uk with uk regular for k∈Pα, by (c). Split Ppol=Apol+ϖ−1B, where B(f,g)=∑k∈Pαuk(0)[sk](fg/DK)+ and Apol collects ϖ−1(uk−uk(0))[sk](⋅)+ for k∈Pαand\nu_ks^k_+$ for k∈/Pα; both have regular entries. If degf≤kmin, then deg(fg)≤kmin+K−1, the quotient of fg has degree <kmin, and B(f,⋅)=0; so rankB≤max(0,K−1−kmin). After scaling the matrix is A+ϖ−1B′, where A (the scaled proper parts and the scaled Apol) is integral and B′ is B in the Newton basis, scaled; B′ has regular entries and rankB′=rankB, since the basis change and the scalings are invertible over the fraction field. Expanding the determinant multilinearly in the columns, a term taking more than rankB columns from ϖ−1B′ vanishes by linear dependence, and every other term has valuation ≥−max(0,K−1−kmin). □
Lemma 4.3. If e(d)>K and ζ is an α-root, then −v(ΔK,τ)≤max(0,K−1−kmin(d)).
Proof. Work in the monomial basis. Since e>K, no difference ∣a−b∣<K and no N≤K is divisible by e, so no factor 1−wb−a and no 1−wN vanishes at ζ; hence all resa, λa and Sa (a≤K) are regular at ζ, by (c), (d). Only the νk, k∈Pα, carry poles, and the rank argument of Proposition 4.2 applies with no scaling. □
Proposition 4.4. Let bK,d be the bound of Proposition 4.2 or Lemma 4.3 (0 if d∈/DK) and CK(τ)=∏dΦd(τ)bK,d. Then CK(τ)ΔK,τ(X)∈Z[τ±1][X]; hence CK(t)ΔK(X)∈Q[X], and CK(σt)σΔK(X)∈Q[X] is its conjugate.
Proof. By Lemma 4.1 a coefficient is P1/(τn∏Φdnd) with P1∈Z[τ]. The bound vζ≥−bK,d at every primitive d-th root forces Φdnd−bK,d∣P1 in Q[τ], hence in Z[τ] since Φd is monic. Specialise τ↦t, a unit of O (t=Qηˉ=Q(P−η)), and τ↦σt.
The size of CK(t)
Lemma 4.5. For d≥1, ∣log∣Φd(t)Φd(σt)∣−ϕE(d)logη∣≤2E1, where
E1=a≥1∑∣log(1−φ−a)∣=2.1136…
Proof.Φd(x)=∏a∣d(xa−1)μ(d/a) and ∑a∣dμ(d/a)a=ϕE(d). For ∣x∣=η write xa−1=xa(1−x−a) and use ∣log∣1−x−a∣∣≤∣log(1−η−a)∣≤∣log(1−φ−a)∣; for x=t use ∣1−ta∣ in the same way. □
Let β7=∏p≤7(1−p−2)=1225768 and c∗=811β7+31=36754393=1.19537…
Proposition 4.6.log∣NCK(t)∣≤(c∗K3+O(K2logK))logη, with an absolute implied constant.
Proof. By Lemma 4.5, log∣NCK(t)∣≤logη∑dbK,dϕE(d)+2E1∑dbK,d, and 2E1≤(2E1/logφ)logη. Moreover ∑dbK,d=O(K2logK): for each e≤K at most four orders d have e(d)=e (two if Q=1), each with a W part ≤4K+K2/(4e)+3 by the cluster bound below, and the at most 4K orders d≤4K have α parts ≤K; summing K2/(4e) over e≤K gives O(K2logK). Split bK,d into its W part and its α part.
W part. This part depends only on the cluster sizes and on the identity ∑e(d)=eϕE(d)=2ϕE(e), which are the same for both signs, so the estimate of [7 Proposition 4.6] applies verbatim. Since mr≥1, ∑rmr=K, mr<K/e+1 and c(m)=4(m−1)+δ(m) with 0≤δ(m)≤max(0,m−5)2/4,
r<e∑c(mr)+c′(me)≤4(K−e)+4emax(0,K−4e)2+3.
Bound ϕE(e)≤eg7(e) with g7(e)=∏p≤7p∣e(1−1/p), a function of period 210 with 0≤g7≤1 and mean β7. Summing over the orders d with e(d)=e gives the weight ∑e(d)=eϕE(d)=2ϕE(e)≤2eg7(e), so the W part is at most ∑e≤Kg7(e)fK(e) with fK(y)=8y(K−y)+21max(0,K−4y)2+6y. We use the elementary averaging estimate: if g has period ℓ, 0≤g≤1 and mean β, and f is a function on [0,K] with ∣f∣≤M and total variation V, then
1≤e≤K∑g(e)f(e)−β∫0Kf(y)dy≤2ℓ(M+V).
(Cut [0,K] into blocks of length ℓ. On a full block (e0,e0+ℓ] both ∑g(e)f(e) and β∫f differ from ℓβf(e0) by at most ℓ times the oscillation of f on the block; the oscillations add up to at most V, and the last, incomplete block costs at most 2ℓM.) For fK we have M,V=O(K2), so the W part is at most β7∫0KfK+O(K2)=811β7K3+O(K2).
α part. max(0,K−1−kmin) counts the k∈[kmin,K−2], so
since αj(ζ)=1 means ζ2j+1=Qj+1, which has 2j+1 solutions, for either sign. □
Remark 4.7. The following is a sketch and is not used in the proof. Using ∑n≤yϕE(n)=3y2/π2+O(ylogy) in residue classes and the exact kmin, the same bound holds with c∗ replaced by (12I+80/27)/π2≤1.130790 if Q=−1 and by (12I+8/3)/π2≤1.100769 if Q=1. Here I=∫01ℓ(x)xdx∈[0.6822834,0.6831236], ℓ is the piecewise linear function with ℓ(1/m)=c(m)/m, and 8/(3π2) is the contribution of the odd d with kmin=(d−1)/2.
Smallness at the conjugate
Proposition 5.1.log∣σΔK(−Ξ)∣≤−K2(K+1)logη+Klog(KMK), where MK=C0(K2/((1−φ−3)(1−φ−2))+φ2) and C0=∏i≥1(1−φ−2i)−1=2.0770906….
Proof. At τ=σt=Q/t the data become: w↦w−1, λa↦2(σρ)−a=2ta (as σρ=Qσt=η), Sa↦−Sa (as σψN=−ψN: UN is rational and σD=−D), and νk↦σνk=Qk(k+1)/(αk−Q) (as σαk=Q/αk). Let LXσ be the corresponding formal functional. The conjugate poles are ra=w−a, ∣ra∣=ta, and σDK(s)=∏a(1−w−as)=(−1)Kw−AKFσ(s) with Fσ(s)=∏a(s−ra). Hence
*The pole moments at X=−Ξ.∗ Since 1/(1−w−as)=−ra/(s−ra), LXσ(1/(s−ra))=−ra−12ta(X+Sa), and at X=−Ξ this is 2tara−1∑N>aψN=Γ(ra), using t4a=w2a=ra2 (for both signs) and wa+M=rawM, where
Bounds. As hm(r1,…,rK)=wmhm(1,w,…,wK−1), ∣hm(r)∣≤C0t2m, using ∏i≥1(1−t2i)−1≤C0 for t≤φ−1. For j≥1, ∣σνj∣≤j(j+1)/(1−t3), so the first sum is at most C0K2/((1−t3)(1−t2)). For the second, ∑nn(n+1)t2n+1=2t3/(1−t2)3≤2 and 1/(1−t3)≤φ2/2 for t≤φ−1, so ∑n∣γn∣≤φ2. All these quantities increase with t, so ∣Yk(−Ξ)∣≤MK for 0≤k≤2K−2. Hadamard’s inequality and ∣w∣AKK=η−K2(K+1) finish the proof. □
Nonvanishing
Proposition 6.1.ΔK(Ξ)=0for infinitely manyK.
Proof. The proof of [7 Proposition 6.1] uses only 0<∣w∣<1, ρ=0 and the distinctness of the nodes; we repeat it. Let C(z)=L(1/(z−s))=∑nρ2w2nz/(z−wn)3 for ∣z∣>t2. It continues to C∖{0} with a pole of order exactly 3 at each wn (leading coefficient 2ρnw2n=0), so it is not rational; it is analytic at ∞ and O(1/z). For degR≤n−1 put Hn,R(z)=R(s)/((z−s)Dn(s)), analytic for ∣z∣>t2 and in particular at every xa, and
Tn,R(z)=L((z−s)Dn(s)R(z)Dn(s)−Dn(z)R(s)),
a polynomial in z of degree ≤n−1. Then CR−Tn,R=DnHn,R.
If Δn(Ξ)=0, any kernel vector gives Rn=0 of degree ≤n−1 with L(sjRn/Dn)=0 for j<n. Put Tn=Tn,Rn and En=CRn−Tn. Expanding (z−s)−1=∑jsjz−j−1, uniformly with two s-derivatives on ∣s∣≤t2 for ∣z∣>t2+ϵ0 (ϵ0>0), gives Hn,Rn(z)=O(z−n−1), so En=O(z−1) and En(xa)=0 for a≤n. If Δn(Ξ)=Δn+1(Ξ)=0, then V0=Tn+1Rn−TnRn+1=Rn+1En−RnEn+1 is a polynomial, O(zn−1), vanishing at x1,…,xn; so V0=0 and Tn/Rn=Tn+1/Rn+1. If Δn(Ξ)=0 for all n≥N, all Tn/Rn equal one reduced fraction T/R0, so Rn=Un′R0, Tn=Un′T, and E:=CR0−T=0. Then Un′E=En=O(1/z) forces E=c0z−q(1+o(1)) with c0=0, q≥1, and degUn′≤q−1; but E(xa)=0 for all large a while En(xa)=0, so Un′(xa)=0 for all large a≤n, too many zeros once n is large. □
Proof of Theorem 1.1
Suppose ∑nUn−3=u/m with u∈Z, m∈Z>0. By (1.1) and Lemma 2.1(iv), Ξ=uD/(mD2)=u(P−2Qt)/(mD2) and σΞ=−Ξ. For K in the infinite set of Proposition 6.1 put yK=CK(t)(mD2)KΔK(Ξ). By Proposition 4.4, yK∈O, and yK=0 because ΔK(Ξ)=0 and CK(t)=0 (t is not a root of unity). Moreover σyK=CK(σt)(mD2)KσΔK(−Ξ). The norm yKσyK of the nonzero algebraic integer yK is a nonzero rational integer, so ∣yKσyK∣≥1, and by Propositions 3.1, 5.1 and 4.6
Since c∗−911−811β7−98=−0.02684…<0, this fails for large K. □
Remark 7.1. Not used in the proof. With the constants of Remark 4.7 the margin is at least 0.0914 (Q=−1) and 0.1214 (Q=1) per K3logη. At finite K, computed with exact determinants over Q(D) and interval arithmetic, the left side above without the term 2Klog(mD2), divided by K3logη, is about −0.12 for the Pell numbers at K=20 and K=25 and about −0.14 for (P,Q)=(3,1), i.e. ∑nF2n−3, at K=20. With the true denominator, the norm of the ideal generated by the coefficients of ΔK, in place of CK(t), these numbers become about −0.17 and −0.21. For Q=1 the local bounds of Section 4 agree with the exact multiplicities of the cyclotomic factors in Q(τ)[X] for K≤13, with equality in most cases. The scripts are available from the author.
all file names below refer to it. It (library LucasU, Lean 4.35.0, Mathlib v4.35.0, about 9200 lines) generalises the formalisation of [7] and proves
theorem LucasU.irrational_zetaU (hA : Admissible a E) : Irrational (zetaU a E),
where Admissible a E means E=1, a≥1 or E=−1, a≥3, and zetaU a E is the series ∑n≥01/Un+13 for the sequence Useq a E with Un+2=aUn+1+EUn, written with a tsum; so (a,E)=(P,−Q). The proof uses only the axioms propext, Classical.choice and Quot.sound; there is no sorry, no native_decide and no added axiom.
What has to be trusted. For the theorem itself, only the definitions of Useq, zetaU and Admissible (and Mathlib’s tsum and Irrational) have to be read. Since a divergent tsum has the value 0, which is rational, the statement cannot hold for a degenerate reason; moreover hasSum_zetaU shows that the series converges (to DDΞ). The corollaries are LucasU.irrational_pell, that the tsum of 1/Pn+13 over n≥0 is irrational, with pell = Useq 2 1, and LucasU.irrational_fib_mul, that for k≥1 the tsum of 1/Fk(n+1)3 over n≥0, written with Nat.fib, is irrational; the identity Fkn=FkUn(Lk,(−1)k) is LucasU.fib_mul_eq, stated as Fkn=FkUn for Useq (lucasNum k) ((-1)^(k+1)), i.e. (a,E)=(Lk,(−1)k+1), with the Lucas numbers lucasNum (all in LucasU/Corollaries.lean and LucasU/Final.lean).
Definitions and the chain
LucasU/Defs.lean defines the construction of Section 2 over an arbitrary field with a sign ε=Q and a parameter τ (wpar = ετ^2, alpha, nu, psi, S, node, F, res, mConst, mLin, nrm, Delta), the real functional Lfun and XiR as functions of (ε,t), and the clearing polynomial CKpoly with exponents nExp. LucasU/Statements.lean states the ingredients as propositions in (ε,t) (and Integrality in ε alone, for every real τ=0 that is not a root of unity); the proofs below assume Good εt, i.e. ε=±1 and 0<t≤φ−1. The conjugate point is τ=ε/t. LucasU/Unit.lean contains Lemma 2.1, and LucasU.irrational_zetaU of (in LucasU/Main.lean) derives the theorem from the ingredients at ε=−E, t=1/η, using that t and ε/t are the two roots of τ2+aEτ−E and that D is irrational (norm.ne_zero).
Paper
Lean statement
Lean proof
Lemma 2.3, entries
EntryIdentity
LucasU.Analytic.entryIdentity
Prop. 3.1
DecayBound
LucasU.Analytic.decayBound
Lemma 4.1–Prop. 4.4
Integrality
LucasU.Denom.integrality
Lemma 4.5, Prop. 4.6
CostBound
LucasU.Cost.costBound
Prop. 5.1
ConjBound
LucasU.Conj.conjBound
Prop. 6.1
Nonvanishing
LucasU.NonVan.nonvanishing
Section 7
LucasU.irrational_zetaU
Table 1 (PDF p. 12).
EntryIdentity is used only inside the proof of DecayBound, not by the assembly. The formal statements are the propositions in the form the assembly needs, with constants that may depend on (ε,t): the uniformity in (P,Q) claimed in Propositions 3.1, 4.6 and the explicit MK of Proposition 5.1 are not expressed by the formal statements and are not needed, since the theorem concerns one fixed (P,Q) (the formal proof of DecayBound does give the uniform constant C=4096). DecayBound is ∣ΔK(Ξ)∣≤CK2t2bK with bK=2∑j<Kaj (proved with C=4096), ConjBound is ∣σΔK(−Ξ)∣≤CK2t2(K+1), and CostBound is log∣CK(t)CK(ε/t)∣≤100121K3logt−1+CK2(1+logK)2; its proof follows Proposition 4.6 with the cubic constant c∗=36754393 for both signs, and c∗<100121<911. The proof of Integrality works in the local rings of C(τ) at the roots of unity, and uses the root types only through LucasU/Roots.lean; the proof of Nonvanishing works with the Cauchy transform C directly.
Provenance
The formal development generalises that of [7], parts of which are adapted from the Lean formalisation QZeta3 [10] of [11] (Apache License 2.0); every adapted file says so in its header.
Use of AI tools
This work was carried out with substantial assistance from AI systems (Anthropic’s Claude models), which were used to search the literature, to develop and check the proof, to write the Lean formalisation @page 12/13 starts_mid=0 ends_mid=1
Definitions and the chain
LucasU/Defs.lean defines the construction of Section 2 over an arbitrary field with a sign ε=Q and a parameter τ (wpar = ετ^2, alpha, nu, psi, S, node, F, res, mConst, mLin, nrm, Delta), the real functional Lfun and XiR as functions of (ε,t), and the clearing polynomial CKpoly with exponents nExp. LucasU/Statements.lean states the ingredients as propositions in (ε,t) (and Integrality in ε alone, for every real τ=0 that is not a root of unity); the proofs below assume Good εt, i.e. ε=±1 and 0<t≤φ−1. The conjugate point is τ=ε/t. LucasU/Unit.lean contains Lemma 2.1, and LucasU.irrational_zetaU of (in LucasU/Main.lean) derives the theorem from the ingredients at ε=−E, t=1/η, using that t and ε/t are the two roots of τ2+aEτ−E and that D is irrational (norm.ne_zero).
Paper
Lean statement
Lean proof
Lemma 2.3, entries
EntryIdentity
LucasU.Analytic.entryIdentity
Prop. 3.1
DecayBound
LucasU.Analytic.decayBound
Lemma 4.1–Prop. 4.4
Integrality
LucasU.Denom.integrality
Lemma 4.5, Prop. 4.6
CostBound
LucasU.Cost.costBound
Prop. 5.1
ConjBound
LucasU.Conj.conjBound
Prop. 6.1
Nonvanishing
LucasU.NonVan.nonvanishing
Section 7
LucasU.irrational_zetaU
Table 1 (PDF p. 12, second).
EntryIdentity is used only inside the proof of DecayBound, not by the assembly. The formal statements are the propositions in the form the assembly needs, with constants that may depend on (ε,t): the uniformity in (P,Q) claimed in Propositions 3.1, 4.6 and the explicit MK of Proposition 5.1 are not expressed by the formal statements and are not needed, since the theorem concerns one fixed (P,Q) (the formal proof of DecayBound does give the uniform constant C=4096). DecayBound is ∣ΔK(Ξ)∣≤CK2t2bK with bK=2∑j<Kaj (proved with C=4096), ConjBound is ∣σΔK(−Ξ)∣≤CK2t2(K+1), and CostBound is log∣CK(t)CK(ε/t)∣≤100121K3logt−1+CK2(1+logK)2; its proof follows Proposition 4.6 with the cubic constant c∗=36754393 for both signs, and c∗<100121<911. The proof of Integrality works in the local rings of C(τ) at the roots of unity, and uses the root types only through LucasU/Roots.lean; the proof of Nonvanishing works with the Cauchy transform C directly.
Provenance
The formal development generalises that of [7], parts of which are adapted from the Lean formalisation QZeta3 [10] of [11] (Apache License 2.0); every adapted file says so in its header.
Use of AI tools
This work was carried out with substantial assistance from AI systems (Anthropic’s Claude models), which were used to search the literature, to develop and check the proof, to write the Lean formalisation and to draft this text. The author directed the project and is responsible for its content. The formal proof is checked by the Lean kernel.
References
References
[1]R. André-Jeannin, Irrationalité de la somme des inverses de certaines suites récurrentes, C. R. Acad. Sci. Paris Sér. I Math. 308 (1989), no. 19, 539–541.
[2]D. Duverney, Ke. Nishioka, Ku. Nishioka, and I. Shiokawa, Transcendence of Rogers–Ramanujan continued fraction and reciprocal sums of Fibonacci numbers, Proc. Japan Acad. Ser. A Math. Sci. 73 (1997), no. 7, 140–142.DOI
[3]C. Elsner, S. Shimomura, and I. Shiokawa, Algebraic relations for reciprocal sums of Fibonacci numbers, Acta Arith. 130 (2007), 37–60.DOI
[4]———, Algebraic relations for reciprocal sums of odd terms in Fibonacci numbers, Ramanujan J. 17 (2008), 429–446.DOI
[5]———, Algebraic relations for reciprocal sums of even terms in Fibonacci numbers, J. Math. Sci. (N.Y.) 180 (2012), no. 5, 650–671, translated from Fundam. Prikl. Mat. 16 (2010), no. 5, 173–200.
[6]C. Krattenthaler, A determinant identity for moments of orthogonal polynomials that implies Uvarov’s formula for the orthogonal polynomials of rationally related densities, 2021, preprint, arXiv:2103.03969.arxiv.org/abs/2103.03969
[7]S. Krisman, The sum of the reciprocal cubes of the Fibonacci numbers is irrational, preprint, 2026; Lean formalisation at https://github.com/telaviv/fibonacci-zeta3-irrationality.
[8]Yu. V. Nesterenko, Modular functions and transcendence questions, Sb. Math. 187 (1996), 1319–1348.
[9]M. Pré vost, A new proof of the irrationality of ζ(2) and ζ(3) using Padé approximants, J. Comput. Appl. Math. 67 (1996), 219–235.
[10]Fikreab Solomon Admasu, QZeta3: Lean 4 formalisation of the irrationality of ζ_{1/p}(3) (moving-denominator Hankel argument), version 0.1.0, 2026-10-11, Apache License 2.0; directory lean/ of [11].
[11]uplift-universal-inc, Irrationality of the Lambert-series value ζ_{1/p}(3), unrefereed AI-assisted research draft with Lean formalisation, version 0.3.0-two-proofs, 2026-10-12, https://github.com/uplift-universal-inc/qzeta_3-irrationality (commit d12d56f).
[12]V. B. Uvarov, The connection between systems of polynomials orthogonal with respect to different distribution functions, U.S.S.R. Comput. Math. Math. Phys. 9 (1969), no. 6, 25–36, translated from Zh. Vychisl. Mat. Mat. Fiz. 9 (1969), 1253–1262.DOI
[14]Z. Xu and T. Wang, The infinite sum of the cubes of reciprocal Pell numbers, Adv. Difference Equ. 2013 (2013), Paper No. 18, 10 pp.DOI
[15]W. Zudilin, On the irrationality of generalized q-logarithm, Res. Number Theory 2 (2016), article 15.DOI
[16]———, A determinantal approach to irrationality, Constr. Approx. 45 (2017), 301–310.