OFFSET
1,2
COMMENTS
Note that a(57) = 210 = 2*3*5*7, while 330 = 2*3*5*11 is not in the sequence. This demonstrates that this sequence is not determined by prime signature alone. - Charles R Greathouse IV, Oct 17 2015
Every prime number appears in this sequence. Proof: the greedy algorithm processes natural numbers in increasing order. When prime p is considered, every element already in the sequence is less than p, so none is divisible by p. By Euclid's lemma, no existing triple product a(i)*a(j)*a(k) is divisible by p. But any new triple product involving p is divisible by p. Therefore no collision is possible, and p is admitted. For collisions among new triples: a(i)*a(j)*p = a(k)*a(l)*p implies a(i)*a(j) = a(k)*a(l), which would make 1*a(i)*a(j) = 1*a(k)*a(l) an existing collision (since 1 is in the sequence), contradicting the B_3 property. The same argument shows every prime appears in the greedy multiplicative B_k sequence for any k >= 2. Lean 4 formalization and computational verification: see Links. - Nova Spivack, Mar 21 2026
LINKS
Zak Seidov, Table of n, a(n) for n = 1..200
Nova Spivack, Proof, Lean 4 formalization, and computational verification that every prime appears in this sequence, GitHub, 2026.
MATHEMATICA
f[l_List] := Block[{k = 1, p2 = Times @@@ Subsets[l, {2}], p3 = Times @@@ Subsets[l, {3}]}, While[Intersection[p3, p2*k] != {}, k++ ]; Append[l, k]]; Nest[f, {1, 2, 3}, 62] (* Ray Chandler, Feb 12 2007 *)
CROSSREFS
KEYWORD
nonn
AUTHOR
Amarnath Murthy, Feb 19 2003
EXTENSIONS
Extended by Ray Chandler, Feb 12 2007
Typo in name fixed by Zak Seidov, Jul 06 2013
STATUS
approved
