Bounded-box reductions in the Subbarao–Warren problem for unitary perfect numbers

Tom Maciejewski · 2026

Read Bounded-box reductions in the Subbarao–Warren problem for unitary perfect numbers by Tom Maciejewski (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-20475.txt.

Introduction

Tom Maciejewski revisits an old hunt for a sixth unitary perfect number. Such a number's unitary divisors sum to twice the number itself, and only five are known. By holding one seed factor explicit and sifting source parts in a bounded graph search, he rules out five impostor kernels. A three-filter certificate does that work, and a stubborn even-number set is bounded to at most 201 members below 40000.

Catalog record

CREATORTom Maciejewski
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE76307
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.20475

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Bounded-box reductions in the Subbarao–Warren problem for unitary perfect numbers Thanks: ORCID: 0009-0004-5702-5302. Tom Maciejewski Abstract . A unitary perfect number is a positive integer n n satisfying σ ∗ ​ ( n ) = 2 ​ n \sigma^{*}(n)=2n . Only five examples are known. We revisit the Subbarao–Warren problem by keeping the seed factor 2 a + 1 2^{a}+1 explicit in the full balance ( 2 a + 1 ) ​ ∏ i ( p i e i + 1 ) = 2 a + 1 ​ ∏ i p i e i . (2^{a}+1)\prod_{i}(p_{i}^{e_{i}}+1)=2^{a+1}\prod_{i}p_{i}^{e_{i}}. Inside a bounded enumeration box for source strongly connected components of the odd dependency graph, every admissible source kernel is either one of the two kernels occurring in the known nonsquarefree examples, 3 2 3^{2} and 5 4 5^{4} , or one of five additional impostor kernels. For these five impostor kernels we give a reproducible three-filter elimination certificate for all relevant seed classes with 1 ≤ a ≤ 10000 1\leq a\leq 10000 . The filters use Zsigmondy-type exponent obstructions, non-3-Higgs witnesses inherited from seed divisors, and a deterministic 2-adic budget overshoot. The remaining obstruction is isolated in the auxiliary set H even H_{\mathrm{even}} of even m m for which every prime divisor of 2 m + 1 2^{m}+1 is 3-Higgs. A structural lemma shows that if m = 2 ​ k ∈ H even m=2k\in H_{\mathrm{even}} , then k k is Higgs-cubefree and every odd divisor of k k gives another element of H even H_{\mathrm{even}} . Thus finiteness of H even H_{\mathrm{even}} is equivalent to finiteness of the prime branch m = 2 ​ p m=2p , while finite computations may also leave composite candidates inherited from unresolved prime divisors. Using the supplied factor cache and APR-CL primality-verification transcripts, we verify | H even ∩ [ 2 , 40000 ] | ≤ 201 , | H even ∩ [ 2 , 50000 ] | ≤ 272 , |H_{\mathrm{even}}\cap[2,40000]|\leq 201,\qquad|H_{\mathrm{even}}\cap[2,50000]|\leq 272, with explicit undecided candidate lists. Finally, Ford’s theorem for downward-closed prime sets gives an unconditional power-saving thinness bound for H even H_{\mathrm{even}} , but not finiteness. We identify the unresolved step as a divisor-level problem for the fixed cyclotomic values Φ 4 ​ p ​ ( 2 ) \Phi_{4p}(2) and formulate conditional criteria that would close the remaining branch. The paper therefore does not prove the full conjecture; it supplies a bounded-box elimination, a finite verified frontier, and a precise analytic target for the remaining obstruction. 2020 Mathematics Subject Classification Primary 11A25; Secondary 11Y05, 11N69, 11Y50, 11A41 1. Introduction For a positive integer n , write \sigma^{*}(n)=\sum_{d\,\|\,n}d where d\,\|\,n means d\mid n and \gcd(d,n/d)=1 . A unitary perfect number (UPN) is an n satisfying \sigma^{*}(n)=2n . The classical list is: 6,\quad 60,\quad 90,\quad 87360,\quad 146361946186458562560000. Subbarao and Warren [ 1 ] introduced UPNs and gave the first four; Wall [ 2 ] added the fifth. No further example is known. Conjecture (strict form) . The five integers above are the only unitary perfect numbers. Wall [ 3 ] proved that any new unitary perfect number has at least nine odd components. Graham [ 4 ] showed that a UPN with squarefree odd part must be one of 6,60,87360 , and that any hypothetical further UPN must have an odd prime divisor exceeding 2^{15}=32768 . The two known examples with nonsquarefree odd part have specific small “kernels”: 3^{2} in 90 and 5^{4} in the fifth UPN. We shall call these the known kernels . Kernels vs. the full UPN. The bound p\leq 2000 in the enumeration box \mathcal{B} (see ( 1 ) below) is consistent with Graham’s lower bound: \mathcal{B} constrains only the source-SCC kernel primes that initiate the odd dependency cascade, not the full prime support of the UPN. Graham’s bound says that any hypothetical sixth UPN contains a prime >2^{15} somewhere in its support; that prime appears as a descendant in the cascade produced by the kernel, not as a kernel prime itself. The kernels we enumerate are the small “roots” of the cascade; the cascade itself, once unfolded through the seed 2^{a}+1 and the multiplicative balance, can and must include arbitrarily large primes. The role of the bounded box is therefore to enumerate root configurations, not to bound the ultimate prime support. Status of results. Because this manuscript mixes fully rigorous results with heuristic analytic conjectures, we adopt the convention that every numbered theorem, proposition, lemma, or conjecture carries one of the following two labels: R (rigorous, unconditional).: Theorems 3 , 8 , 9 – 18 (rigorous bounds |H_{\mathrm{even}}\cap[2,N]|\leq\cdot with 191 undecided through N=40000 ), Theorem 22 (power-saving thinness of H_{\mathrm{even}} ), Lemma 21 (seven deep Pratt-tree closures, with APR-CL primality-verification transcripts for the six large witness primes generated by PARI/GP 2.17.3 and bundled as supplementary material), Proposition 5 (structural lemma), Proposition 29 (loglog obstruction to sublog-growth), and the combined rigorous bound |H_{\mathrm{even}}\cap[2,40000]|\leq 201 . C-A (conditional on analytic hypotheses).: Theorem 28 (conditional sharpened reduction; uses Stewart-radical plus effective Chebotarev), and Theorem 31 (conditional finiteness under the hybrid hypothesis). Conjectures 7 , 24 , 25 , 27 , 30 . The headline computational bound is therefore |H_{\mathrm{even}}\cap[2,40000]|\leq 201 , rigorous (the six large witness primes used in the deep Pratt-tree closures are APR-CL verified). The prime-case reduction (Theorem 8 ) and the power-saving thinness theorem (Theorem 22 ) are likewise fully rigorous. 1.1. Setup Decompose n=2^{a}\prod p_{i}^{e_{i}} with p_{i} odd. Multiplicativity of \sigma^{*} with \sigma^{*}(2^{a})=2^{a}+1 and \sigma^{*}(p^{e})=p^{e}+1 makes the defining equation \sigma^{*}(n)=2n the full balance ( \ast ) (2^{a}+1)\prod(p_{i}^{e_{i}}+1)=2^{a+1}\prod p_{i}^{e_{i}}. In particular, 2^{a}+1 is the odd seed factor that initiates the dependency cascade. For each odd prime p in the support, the exponent e_{p} matches the total incoming p -adic valuation from both the seed and the other components: e_{p}=\sum_{q\in\mathrm{components}}v_{p}(q+1). The 2-adic balance is a+1=\sum_{p\text{ odd}}v_{2}(p^{e_{p}}+1) . A prime p is 3-Higgs (OEIS A057447 [ 8 ] ; see also Burris–Yeats [ 16 ] ) by the following well-founded recursion on prime value. The base case: 2 is 3-Higgs (by convention; equivalently, 2-1=1 divides the empty product). The inductive step: an odd prime p is 3-Higgs if and only if every prime factor q of p-1 is itself a 3-Higgs prime strictly less than p and satisfies v_{q}(p-1)\leq 3 , with the additional exponent cap v_{2}(p-1)\leq 3 . Equivalently, p-1 divides the cube of the product of the smaller 3-Higgs primes. The recursion is well-founded because each invocation reduces the prime value strictly. Standard small examples: 3,5,7,11,13,17 fail the recursion only at 17 (since 17-1=2^{4} violates the v_{2}\leq 3 cap), so 17\notin\mathcal{P}_{3} . Proposition 1 (Every prime divisor of a UPN is 3-Higgs) . If n is a unitary perfect number, then every prime divisor of n lies in \mathcal{P}_{3} . Proof. The prime 2 is in \mathcal{P}_{3} by the base case. Let p be an odd prime divisor of n with p^{e}\,\|\,n . By the multiplicative balance (\ast) , every prime factor q of p^{e}+1 must itself appear in the support of n , since the only primes present in the right-hand side of (\ast) are 2 and the p_{i} in the support. Thus, for any prime q\mid p-1\mid p^{e}+1 (with e odd), q is either 2 or another

Read the complete text