Weight preservation for Fibonacci recurrences

About 1 year old · traced to

Let FmF_m be the set of periodic sequences from the Fibonacci recurrence or its parity transform with initial condition (a0,a1)∈(Z/mZ)2(a_0,a_1)\in(\mathbb Z/m\mathbb Z)^2, taken modulo mm. For a divisor dd of a composite modulus mm, define the weight of a period qq of length ℓq\ell_q in FdF_d by

wd(q)=ℓqd2.w_d(q)=\frac{\ell_q}{d^2}.

If qd∈Fdq_d\in F_d extends to periods qm(1),…,qm(k)∈Fmq_m^{(1)},\ldots,q_m^{(k)}\in F_m reducing to qdq_d modulo dd, Weight Preservation of Fibonacci Recurrences.

wd(qd)=∑i=1kwm(qm(i)).w_d(q_d)=\sum_{i=1}^k w_m(q_m^{(i)}).

Thus the total weight of every period is conjectured to be conserved under extension from dd to mm; the supplied text gives examples but no resolution status.

References

Primary source

Marc T. Pudelko, “Modular Periodicity of Random Initialized Recurrences”, arXiv:2510.24882 (2026).

Progress summary

Refreshed
Claimed solved

Computations support the conjecture, while an unverified posted argument claims a general counting proof.

Marc T. Pudelko formulated the conjecture in a preprint first posted in October 2025. It says that the normalized total length of every period is preserved when Fibonacci or parity-transformed sequences are lifted from a divisor modulus to a composite modulus.

October 2025 computational evidence

The preprint checks exact preservation in examples including lifts from d=2d=2 to m=6m=6, from d=3d=3 to m=6m=6, and prime-power cases such as 1919 to 361361. It gives no general proof, counterexample, withdrawal, or independent verification.

Posted attempt

A posted argument claims a complete proof by counting fibers of the reduction map between finite orbit sets: each fiber has size (m/d)2(m/d)^2, and decomposing the inverse image of an orbit yields the asserted weight identity. This attempt has not been independently verified.

Current status (as of August 2026): The conjecture has an unverified complete proof attempt, but no verified proof or counterexample is recorded, so its mathematical resolution remains open.

Sources

Solutions 1

ProofThis solution needs a summarySee full solutionHide full solution

The claim follows from an orbit-fiber counting identity, valid much more generally than the Fibonacci recurrence.

For ε∈{1,−1}\varepsilon\in\{1,-1\}, let

Aε=(011ε).A_\varepsilon= \begin{pmatrix}0&1\\1&\varepsilon\end{pmatrix}.

The Fibonacci recurrence and its parity transform act on initial-condition pairs by

(aj,aj+1)⟼(aj+1,aj+εaj+1)=Aε(aj,aj+1).(a_j,a_{j+1}) \longmapsto (a_{j+1},a_j+\varepsilon a_{j+1}) =A_\varepsilon(a_j,a_{j+1}).

Since det⁡Aε=−1\det A_\varepsilon=-1, this is a permutation of

Xs=(Z/sZ)2X_s=(\mathbb Z/s\mathbb Z)^2

for every s≥1s\ge1. Its permutation orbits are precisely the periods modulo ss, and orbit cardinality equals period length.

For d∣md\mid m, coordinatewise reduction

π:Xm⟶Xd\pi:X_m\longrightarrow X_d

commutes with both recurrence actions:

πAε,m=Aε,dπ.\pi A_{\varepsilon,m}=A_{\varepsilon,d}\pi.

Every fiber has cardinality

∣π−1(x)∣=(m/d)2.|\pi^{-1}(x)|=(m/d)^2.

Fix an orbit O⊂XdO\subset X_d of length ℓ\ell. Its full inverse image is invariant under Aε,mA_{\varepsilon,m}, so it decomposes into disjoint orbits

π−1(O)=O1⊔⋯⊔Or.\pi^{-1}(O)=O_1\sqcup\cdots\sqcup O_r.

Each OiO_i maps onto OO, since its image is a nonempty invariant subset of a single transitive orbit. Thus the OiO_i are exactly all periods modulo mm reducing to the chosen period modulo dd.

Counting the inverse image yields

∑i=1r∣Oi∣=∣π−1(O)∣=ℓ(md)2.\sum_{i=1}^r|O_i| = |\pi^{-1}(O)| = \ell\left(\frac md\right)^2.

Dividing by m2m^2 gives precisely

∑i=1rwm(Oi)=∑i=1r∣Oi∣m2=ℓd2=wd(O).\sum_{i=1}^r w_m(O_i) = \sum_{i=1}^r\frac{|O_i|}{m^2} = \frac{\ell}{d^2} = w_d(O).

More generally, if AA is any integral r×rr\times r matrix whose determinant is coprime to mm, then for every d∣md\mid m and every orbit OO modulo dd,

∑O′ orbit modulo m\O′modd=O∣O′∣mr=∣O∣dr.\sum_{\substack{O'\text{ orbit modulo }m\O'\bmod d=O}} \frac{|O'|}{m^r} = \frac{|O|}{d^r}.

Thus orbit-weight preservation holds for every invertible integral linear recurrence in every dimension.