The Stieltjes-parameter formula for generalized Rueppel sequences

About 5 years old · traced to

Let

rb,c(x)=1+cx+b(x3+x7+x15+x31+⋯ )r_{b,c}(x)=1+cx+b(x^3+x^7+x^{15}+x^{31}+\cdots)

and write its Stieltjes continued fraction as

rb,c(x)=S(c,−c,−bc2,bc2,−c,c,−1b,1b,c,−c,bc2,−bc2,−c,c,−1b,1b,c,−c,…).r_{b,c}(x)=\mathcal{S}\left(c,-c,-\frac{b}{c^2},\frac{b}{c^2},-c,c,-\frac{1}{b},\frac{1}{b},c,-c,\frac{b}{c^2},-\frac{b}{c^2},-c,c,-\frac{1}{b},\frac{1}{b},c,-c,\ldots\right).

Let sb,c(n)s_{b,c}(n) denote the corresponding Stieltjes parameters. Generalized Rueppel parameter conjecture. Then

sb,c(n)={c,n≡0,5(mod8),−c,n≡1,4(mod8),(−1)kb/c2,n=8k+3,−(−1)kb/c2,n=8k+2,(2P(k)−1)/b,n=8k+7,−(2P(k)−1)/b,n=8k+6,s_{b,c}(n)=\begin{cases}c,&n\equiv0,5\pmod 8,\\-c,&n\equiv1,4\pmod 8,\\(-1)^k b/c^2,&n=8k+3,\\-(-1)^k b/c^2,&n=8k+2,\\(2P(k)-1)/b,&n=8k+7,\\-(2P(k)-1)/b,&n=8k+6, \end{cases}

where P(k)P(k) is the paper-folding sequence OEIS A014577. The source also presents an equivalent closed form for sb,c(n)s_{b,c}(n); the conjecture is part of the generalization of Rueppel sequences and is followed by initial Hankel-transform data, but no proof or resolution is given.

References

Primary source

Paul Barry, “Conjectures and results on some generalized Rueppel sequences”, arXiv:2107.00442 (2021).

Progress summary

Refreshed
Claimed progress

A reader-submitted proof claims to settle the formula, but no independent source has verified it, so the problem remains open.

Barry’s paper records the proposed formula for the generalized Rueppel sequence as Conjecture 18, giving the Stieltjes parameters by residue classes modulo 88 and the paper-folding sequence P(k)P(k). The paper, dated July 2021 in the catalogue, provides no proof or resolution.

Community submission (unverified), August 25, 2026

A submitted proof claims the formula for all nonzero parameters b,cb,c over a characteristic-zero field. It introduces recurrences for ordinary, shifted, and twice-shifted Rueppel Hankel determinants and says these establish the continued-fraction parameters and nonvanishing minors; the argument is unverified.

Current status (as of August 2026): The formula remains unverified; the only new development is a reader-submitted purported proof, while the published source still records it as Conjecture 18.

Sources

Solutions 1

ProofThis solution needs a summarySee full solutionHide full solution

Proof of the generalized Rueppel Stieltjes-parameter conjecture

Let b,c≠0b,c\neq0 belong to a field of characteristic zero, and define

M(x)=rb,c(x)=1+cx+b∑j≥2x2j−1.(1)M(x)=r_{b,c}(x) =1+cx+b\sum_{j\geq2}x^{2^j-1}. \tag{1}

Write its Stieltjes continued fraction as

M(x)=11−s0x1−s1x1−s2x⋱.(2)M(x) =\cfrac1{1-\cfrac{s_0x}{1-\cfrac{s_1x}{1-\cfrac{s_2x}{\ddots}}}}. \tag{2}

Let P(k)∈{0,1}P(k)\in\{0,1\} be the regular paperfolding sequence, characterized by

P(2k)=1+(−1)k2,P(2k+1)=P(k).(3)P(2k)=\frac{1+(-1)^k}{2}, \qquad P(2k+1)=P(k). \tag{3}

We prove, for every k≥0k\geq0,

s8k=s8k+5=c,s8k+1=s8k+4=−c,s8k+3=−s8k+2=(−1)kbc2,s8k+7=−s8k+6=2P(k)−1b.(4)\boxed{ \begin{aligned} s_{8k}=s_{8k+5}&=c,\\ s_{8k+1}=s_{8k+4}&=-c,\\ s_{8k+3}=-s_{8k+2}&=(-1)^k\frac{b}{c^2},\\ s_{8k+7}=-s_{8k+6}&=\frac{2P(k)-1}{b}. \end{aligned}} \tag{4}

This proves Conjecture 18 of Paul Barry, Conjectures and results on some generalized Rueppel sequences uniformly for all nonzero parameters b,cb,c. In particular, all the Hankel minors required to define the continued fraction are nonzero.

Ordinary and shifted Rueppel determinants

Let

r(x)=∑j≥0rjxj=∑j≥0x2j−1,r(x)=\sum_{j\geq0}r_jx^j =\sum_{j\geq0}x^{2^j-1},

and put

Rm=det⁡(ri+j)0≤i,j<m,Jm=det⁡(ri+j+1)0≤i,j<m,R0=J0=1.(5)R_m=\det(r_{i+j})_{0\leq i,j<m}, \qquad J_m=\det(r_{i+j+1})_{0\leq i,j<m}, \qquad R_0=J_0=1. \tag{5}

The ordinary Rueppel determinant evaluation we will derive is

Rm=(−1)(m2).(6)R_m=(-1)^{\binom m2}. \tag{6}

Furthermore,

r2j={1,j=0,0,j≥1,r2j+1=rj.(7)r_{2j}=\begin{cases} 1,&j=0,\\ 0,&j\geq1, \end{cases} \qquad r_{2j+1}=r_j. \tag{7}

Separating even and odd row and column indices in the ordinary and once-shifted Rueppel Hankel matrices gives

R2t=(−1)tRt2,R2t+1=(−1)tJt2,J2t=RtJt,J2t+1=Rt+1Jt.(8)\begin{aligned} R_{2t}&=(-1)^tR_t^2, &R_{2t+1}&=(-1)^tJ_t^2,\\ J_{2t}&=R_tJ_t, &J_{2t+1}&=R_{t+1}J_t. \end{aligned} \tag{8}

Starting from R0=J0=1R_0=J_0=1, simultaneous induction in (8) proves (6) and shows that every JtJ_t equals 11 or −1-1.

For the twice-shifted determinants

Lt=det⁡(ri+j+2)0≤i,j<t,(9)L_t=\det(r_{i+j+2})_{0\leq i,j<t}, \tag{9}

both same-parity blocks vanish. The off-diagonal blocks are the once-shifted Rueppel matrices, so

L2u=(−1)uJu2=(−1)u,L2u+1=0.(10)L_{2u}=(-1)^uJ_u^2=(-1)^u, \qquad L_{2u+1}=0. \tag{10}

Exact moment determinants for both parameters

Let μj=[xj]M(x)\mu_j=[x^j]M(x). Equation (1) gives

μ2j={1,j=0,0,j≥1,μ2j+1=wj,(11)\mu_{2j}=\begin{cases} 1,&j=0,\\ 0,&j\geq1, \end{cases} \qquad \mu_{2j+1}=w_j, \tag{11}

where

wj=brj+(c−b)1{j=0}.(12)w_j=br_j+(c-b)\mathbf1_{\{j=0\}}. \tag{12}

Define

Wt=det⁡(wi+j)0≤i,j<t,W0=1.(13)W_t=\det(w_{i+j})_{0\leq i,j<t}, \qquad W_0=1. \tag{13}

The matrix in (13) differs from b(ri+j)b(r_{i+j}) only at its upper-left entry. Consequently,

Wt=bt−1(bRt+(c−b)Lt−1)(t≥1).(14)W_t=b^{t-1}\left(bR_t+(c-b)L_{t-1}\right) \qquad(t\geq1). \tag{14}

When tt is even, (10) gives Lt−1=0L_{t-1}=0. When t=2u+1t=2u+1 is odd, (6) and (10) give

Lt−1=(−1)u=Rt.L_{t-1}=(-1)^u=R_t.

Therefore

Wt=Rt b t−(t mod 2)c t mod 2.(15)\boxed{ W_t=R_t\,b^{\,t-(t\bmod2)}c^{\,t\bmod2}. } \tag{15}

Now define the ordinary and once-shifted moment determinants

ΔN=det⁡(μi+j)0≤i,j<N,ΓN=det⁡(μi+j+1)0≤i,j<N,Δ0=Γ0=1.(16)\Delta_N=\det(\mu_{i+j})_{0\leq i,j<N}, \qquad \Gamma_N=\det(\mu_{i+j+1})_{0\leq i,j<N}, \qquad \Delta_0=\Gamma_0=1. \tag{16}

For Δ2m\Delta_{2m}, the two parity classes have equal size, and their off-diagonal block is (wi+j)(w_{i+j}). For Δ2m+1\Delta_{2m+1}, expansion along the unique nonzero same-parity entry leaves two blocks (wi+j+1)(w_{i+j+1}), whose determinants are bmJmb^mJ_m. Thus

Δ2m=(−1)mWm2,Δ2m+1=(−1)mb2m.(17)\boxed{ \Delta_{2m}=(-1)^mW_m^2, \qquad \Delta_{2m+1}=(-1)^mb^{2m}. } \tag{17}

For ΓN\Gamma_N, the mixed-parity blocks vanish. Its two diagonal blocks are (wi+j)(w_{i+j}) and (wi+j+1)(w_{i+j+1}). Hence

ΓN=W⌈N/2⌉ b⌊N/2⌋J⌊N/2⌋.(18)\boxed{ \Gamma_N =W_{\lceil N/2\rceil}\, b^{\lfloor N/2\rfloor}J_{\lfloor N/2\rfloor}. } \tag{18}

Since b,c≠0b,c\neq0, (15), (17), and (18) show that every ΔN\Delta_N and ΓN\Gamma_N is nonzero.

Recovery of all Stieltjes coefficients

The standard Stieltjes determinant formulas are

s2n=Γn+1ΔnΓnΔn+1,s2n+1=Δn+2ΓnΔn+1Γn+1.(19)\begin{aligned} s_{2n} &=\frac{\Gamma_{n+1}\Delta_n} {\Gamma_n\Delta_{n+1}},\\ s_{2n+1} &=\frac{\Delta_{n+2}\Gamma_n} {\Delta_{n+1}\Gamma_{n+1}}. \end{aligned} \tag{19}

From (15)--(18), the ordinary determinants simplify to

Δ4k=b4k,Δ4k+1=b4k,Δ4k+2=−b4kc2,Δ4k+3=−b4k+2.(20)\begin{aligned} \Delta_{4k}&=b^{4k},& \Delta_{4k+1}&=b^{4k},\\ \Delta_{4k+2}&=-b^{4k}c^2,& \Delta_{4k+3}&=-b^{4k+2}. \end{aligned} \tag{20}

Write

Ek=(−1)kRkJk.(21)E_k=(-1)^kR_kJ_k. \tag{21}

Using (6) and (8), the shifted determinants in (18) become

Γ4k=Ekb4k,Γ4k+1=Ekcb4k,Γ4k+2=(−1)kEkcb4k+1,Γ4k+3=(−1)k+1Ekb4k+3.(22)\begin{aligned} \Gamma_{4k}&=E_kb^{4k},\\ \Gamma_{4k+1}&=E_kcb^{4k},\\ \Gamma_{4k+2}&=(-1)^kE_kcb^{4k+1},\\ \Gamma_{4k+3}&=(-1)^{k+1}E_kb^{4k+3}. \end{aligned} \tag{22}

Substitution of (20) and (22) into (19) gives

ns2ns2n+14kc−c4k+1(−1)k+1b/c2(−1)kb/c24k+2−cc4k+3−JkJk+1/bJkJk+1/b.(23)\begin{array}{c|c|c} n&s_{2n}&s_{2n+1}\\ \hline 4k&c&-c\\ 4k+1&(-1)^{k+1}b/c^2&(-1)^kb/c^2\\ 4k+2&-c&c\\ 4k+3&-J_kJ_{k+1}/b&J_kJ_{k+1}/b. \end{array} \tag{23}

It remains to identify the sign in the final row. Set

εk=JkJk+1.(24)\varepsilon_k=J_kJ_{k+1}. \tag{24}

Equations (6) and (8) give

ε2t=RtRt+1=(−1)t,ε2t+1=JtJt+1=εt.(25)\begin{aligned} \varepsilon_{2t} &=R_tR_{t+1}=(-1)^t,\\ \varepsilon_{2t+1} &=J_tJ_{t+1}=\varepsilon_t. \end{aligned} \tag{25}

Comparing with the defining paperfolding recurrences (3), and using ε0=1=2P(0)−1\varepsilon_0=1=2P(0)-1, we obtain

JkJk+1=εk=2P(k)−1(k≥0).(26)J_kJ_{k+1}=\varepsilon_k=2P(k)-1 \qquad(k\geq0). \tag{26}

Substitution into (23) gives exactly the four asserted families in (4), proving the complete two-parameter Stieltjes-continued-fraction formula.