Uniqueness of the -primitive decomposition for fixed square-free part

From papers

Let sNs\in\mathbb{N} be square-free, and let P(s)P(s) be the set of Δ\Delta-primitives with square-free part ss. A natural number nn has a Δ\Delta-primitive decomposition if it can be written as n=α2mn=\alpha^2m with αN\alpha\in\mathbb{N} and mm Δ\Delta-primitive. Fixed-square-free-part uniqueness conjecture. If P(s)=1|P(s)|=1, then every non-square nN(Δ)n\in\mathbb{N}(\Delta) with square-free part ss admits a unique Δ\Delta-primitive decomposition. The conjecture refines a uniqueness question that is false in general: the paper gives 5616=32624=2214045616=3^2\cdot624=2^2\cdot1404 as the smallest counterexample, while the classification of integers with multiple decompositions remains open.

Progress summary

Open

No public discussion or published progress on this conjecture was found.

No public discussion or published progress was found for this problem.

Current status (as of August 2026): the conjecture appears open, with no recorded public activity.

Sources & referencesView supporting material

Primary source

Victor N. Schvöllner, “The Δ property: a bridge between split graphs and Number Theory”, arXiv:2605.25264 (2026).

Solutions 1

Proof

Complete proof. For a positive integer u=ppepu=\prod_p p^{e_p}, let

sf(u)=ep oddp\operatorname{sf}(u)=\prod_{e_p\text{ odd}}p

denote its square-free part. For every positive integer cc,

sf(c2u)=sf(u).\operatorname{sf}(c^2u)=\operatorname{sf}(u).

Proposition 3.2 of the source establishes that every nN(Δ)n\in\mathbb N(\Delta) admits a decomposition n=c2mn=c^2m, where c1c\ge1 and mm is Δ\Delta-primitive. Equivalently, repeatedly remove a nontrivial square factor whenever its quotient remains in N(Δ)\mathbb N(\Delta). The positive quotients strictly decrease, so the process terminates at a primitive quotient.

Fix a square-free ss with P(s)=1|P(s)|=1, and suppose nN(Δ)n\in\mathbb N(\Delta) has square-free part ss. If

n=c12m1=c22m2n=c_1^2m_1=c_2^2m_2

are primitive decompositions, square-free-part invariance gives

sf(m1)=sf(n)=s=sf(m2).\operatorname{sf}(m_1)=\operatorname{sf}(n)=s=\operatorname{sf}(m_2).

Thus m1,m2P(s)m_1,m_2\in P(s). Since P(s)P(s) is a singleton, m1=m2m_1=m_2; consequently c12=c22c_1^2=c_2^2, and positivity gives c1=c2c_1=c_2. The decomposition therefore exists and is unique. The nonsquare assumption is unnecessary.

More precisely, writing n=su2n=su^2, its primitive decompositions are in bijection with

{du:sd2P(s)}\{d\mid u:sd^2\in P(s)\}

via m=sd2m=sd^2 and c=u/dc=u/d. Conversely, if P(s)P(s) contains distinct mi=sdi2m_i=sd_i^2, then

N=slcm(d1,d2)2N=s\,\operatorname{lcm}(d_1,d_2)^2

has two distinct primitive decompositions, and every square multiple of NN does as well. For s>1s>1, all these examples are nonsquares.

0 endorsements
Shivam Patel ·