Research log for issue #47. The full conjecture remains open.
Let be the sum of the proper divisors of . If is a set of positive integers of natural density zero, the conjecture says that also has natural density zero.
Literature
- Erdős, Granville, Pomerance, and Spiro, Conjecture 4 pose the statement in their 1990 paper on iterations of arithmetic functions.
- Pollack, Pomerance, and Thompson, Theorems 1.2 and 1.4 prove the conjecture for arbitrary sets with counting function . They refute a separate proposed uniform bound on individual fibers. A sparse family of large fibers can coexist with the density-zero conjecture.
- Luca and Pomerance, §3 bound the second moment of the fibers on a chosen positive-density set of even deficient inputs with three large prime factors in prescribed windows. This proves positive lower density of their even outputs. The same moment bound gives the density-zero preimage implication on those inputs by Cauchy–Schwarz.
- Pollack and Troupe prove the Erdős–Kac law for . Lebowitz-Lockard, Pollack, and Singha Roy establish equidistribution modulo individual primes up to fixed powers of . These distribution statements address specific kinds of sparse sets.
- Pollack and Singha Roy prove agreement of -free status almost everywhere for . Troupe treats the set of sums of two squares. Benli et al. prove the EGPS conclusion for missing-digit sets. A 2026 follow-up develops further digit-distribution results.
- Erdős Problem 955 and MathWorld record the same full statement. The fiber-tightness audit gives the equivalent question of uniform control of the mass of large finite fibers.
The November 2026 search covered the original papers, later preprints and their references, and the problem discussion. No proof or counterexample for arbitrary density-zero sets was found.
The proof route uses Luca and Pomerance's second moment after separating inputs by their small-prime parts. The new calculation retains quadratic congruences modulo prime powers, so it avoids the use of squarefreeness in the original collision count.
A preimage theorem for a family with three large prime factors
Theorem. Fix any . Let consist of the integers , where are the three largest distinct prime divisors, none divides , and
For every set of natural density zero,
There is no parity or deficiency restriction. The proof adapts Luca and Pomerance, §3. It replaces their squarefree-modulus step with two bounds for a quadratic congruence modulo an arbitrary odd modulus.
Reduction to a second moment
Work on , and put , . Enlarge the allowed ranges slightly to use powers of in the hypotheses. Split the possible values of into finitely many intervals of length at most . On each interval there are fixed exponents such that
Here . It suffices to show the conclusion on each interval.
Write , set , and write for the largest -smooth divisor of . The following conditions can be imposed after discarding an arbitrarily small upper proportion of inputs. Put . For require
For also require , , and . The implied constants may depend on the permitted exceptional proportion. We also take .
Here is the justification for the discards. The small-prime conditions come from Luca and Pomerance's Lemma 2.1, with in place of . For the interval between these thresholds is contained in a common interval of width . Removing inputs divisible by the possible prime in that interval costs . The three prime factors all exceed , so the remaining smooth parts coincide. The reciprocal-divisor condition follows from their Lemma 2.5. Davenport's distribution theorem lets us restrict to a compact subinterval of the positive reals with arbitrarily small exceptional density. The original paper's Theorem 5.1 gives on a set of density one. This bounds on the same retained range because . To transfer an exceptional set of cofactors to inputs , use . For an exceptional set of the extra sum over costs since over primes. An exceptional set of upper density has reciprocal sum at most in the upper-limit sense. Finally follows by summing prime-power divisibility, so costs .
For these inputs . Write for the number of retained inputs giving output . We prove . Cauchy–Schwarz bounds the preimages of by . The exceptional proportion can then be arbitrarily small. The dyadic estimates sum to the stated bound up to .
Counting collisions
A collision has the same on both inputs, since . The diagonal contributes at most after summing all strata. For distinct cofactors put , , and . All prime divisors of exceed . If there is a collision, . Also : otherwise and exceed both coefficients, forcing , and . The two prime variables in the collision equation run along linear forms in one integer variable. The upper-bound sieve used in Luca and Pomerance, (3.7) bounds their count by
The progression parameter has range of scale , which exceeds , and both affine forms have nonzero determinant. As in the cited estimate, the local factors are bounded by a constant times .
For , the collision gives . Writing and , the congruences then give
Indeed . A prime of in is excluded by . If a prime of divides , then it divides . It cannot equal , because that would make it a common prime of and outside . It cannot divide either, by the corresponding gcd condition on . The claimed congruence follows by solving for and . For , fix and one of the possible divisors . The congruences put in fixed residue classes modulo . In dyadic ranges , , their reciprocal sums are and . If the ranges contain a collision, then by . Consequently the product of the two sums, multiplied by , is at most a constant times
This has a power saving in every dyadic range. Summing over , the dyadic ranges, and with weights leaves that saving, up to . The factor is harmless here.
For we have . The reduced denominators on the two sides are and , since . So , and strict monotonicity with that denominator gives .
Thus and . Choose the order . The reciprocal sum of such is at most , even without primality. Using , the bound , over primes, and , the contribution is per stratum. This is negligible in total since .
Small and intermediate moduli
Suppose . Fix , a divisor , and the tail . For a collision we must have
The residue is a unit modulo . Thus are units, the congruences determine and modulo , and satisfies a monic quadratic modulo . For a fixed , is in a single residue class modulo .
We first average over , retaining the factor . Factors from are bounded by the retained totient ratios. Put and . In the factor arising from , the primes at most contribute by Mertens' theorem. Those above contribute , since . In the middle range, primes dividing contribute by the totient ratio of . Those dividing contribute by the retained bound on the sum of reciprocals of divisors of . Define
If , the remaining product is bounded. If , use the overall bound . Together,
For each summand in , the equation fixes a single residue class of modulo . Its modulus combines with by the Chinese remainder theorem. Brun–Titchmarsh and partial summation bound the harmonic sum of in this class by : the modulus is at most , with . A noncoprime residue cannot contain any of the eligible primes. As , this proves, for each fixed in its admissible classes,
For an odd prime power , a monic quadratic has at most roots modulo . Its roots lie in at most two classes modulo . Both statements follow by completing the square. For a nonzero square with valuation the two unit roots lift to two classes modulo . For the zero square the roots are multiples of .
For , apply Brun–Titchmarsh to the -sum in all root classes modulo . Their number is at most . The resulting contribution for each is at most
The final divisor sum is bounded uniformly. At a prime the sum over all positive exponents is , so the product is bounded by a fixed power of . The reciprocal sums over and are each, by the Euler product on -rough integers. Thus this range contributes .
For , the quadratic-root cover gives at most classes modulo . Over those classes, the harmonic sum of integers is . Multiplying by and summing all still yields a power saving, since . The intermediate contribution is negligible in total.
Finally by Mertens' theorem. Summing the main contributions, the negligible terms and the diagonal gives , as required.
Remaining gap
This treats arbitrary zero-density target sets on the stated large-prime family. The proof uses large for the final two-prime sieve and larger than its remaining cofactor to bound the largest collision moduli. These inequalities do not hold on all but inputs. A proof or disproof of the full conjecture must control inputs outside this family.
The margins can be removed on a set of density
Let be the integers admitting with prime, and . These inequalities determine the primes uniquely as the two largest prime factors. For every set of natural density zero,
To remove the margins, work on . For fixed , the number of choices of is by the prime number theorem, since . Fix . The inputs with have total count
Here the inner sum is over primes. For it is by Mertens' theorem. Summing against gives the displayed bound. The term contributes .
Among the remaining inputs, failure of puts between constant multiples of and . The reciprocal sum over these cofactors is , giving inputs. Failure of puts in . As , the prime reciprocal sum there is uniformly in , again by Mertens. This gives inputs.
If the largest prime of is below , the -smooth part of contains and hence exceeds . The average logarithm of this smooth part is , by summing prime-power divisibility. This excludes a further inputs. Finally, the inputs divisible by a square of a prime at least are . Outside these exceptions we are in . Taking and then arbitrarily small gives the preimage assertion.
For the density, first let count of the form with prime. Such representations are unique. The prime number theorem gives
The subtracted sum is , and the last sum is computed by integrating . For , count the possible prime over these . Another use of the prime number theorem gives
Partial summation using multiplies the unrestricted reciprocal sum by . The latter sum tends to by the same integral. This yields the stated density.
Collisions with a zero cofactor determinant
Write each input as , with two distinct primes coprime to . Put and , and use primes on the coefficients for the other input . The shifted-product identity from Pollack, Pomerance, and Thompson, §3 is
In particular, a collision implies
Proposition. Fix and . Consider inputs with , , and the small-prime conditions above for , including . The number of ordered pairs of distinct such inputs having and is . There is no dominance condition on either of the two primes.
In a collision the two small-prime parts are equal. For , the denominator argument above gives . For this common , set and . These are relatively prime, and is bounded in terms of . The two shifted products are equal. Writing the gcd of and as yields a factorization
All four integers are units modulo . Comparing the shifted factors gives , , and . If or , the two unordered prime pairs coincide. For distinct inputs, by interchanging the two inputs and the prime labels we may count only and at a constant factor in the bound.
Split the four integers into intervals of the form with a power of two. Write for the four lower endpoints. In a nonempty box we have , since the respective differences are positive multiples of . Also , since . There are possibilities for : count their possible products in the residue class and use the divisor bound. Once are fixed, the other two residue classes allow choices of . Finally, the bound gives
The count per box is thus at most . There are boxes, and . Summing over gives the claimed bound.
A second-moment argument on a larger family would still need to handle uniformly in the two tails and the ranges of the prime variables.
The small-cofactor theorem also covers inputs whose largest prime is at most the square root and whose two largest primes have product at least . A refinement by prime and cofactor size brings the combined natural density to approximately .
A third-prime theorem also treats with and , including inputs with . The three proved families have combined natural density approximately .