We provide lower bounds on the number of subgroups of a group G as a function of the primes and exponents appearing in the prime factorization of |G|. Using these bounds, we classify all abelian groups with 22 or fewer subgroups, and all non-abelian groups with 19 or fewer subgroups. This allows us to extend the integer sequence A274847 [13] introduced by Slattery in [12].It is a classic problem in a first course in group theory to show that a group G has exactly two subgroups if and only if G ∼ = Z p for a prime p. The main idea here is to observe that if | Sub G| = 2 or if G ∼ = Z p , then every non-identity element x ∈ G must by necessity generate all of G, i.e. x = G for all x ∈ G. Slightly less frequently, a course may follow up by considering groups G with exactly three or four subgroups. In those cases, it turns out that we can again argue that G must be cyclic.Indeed, if | Sub G| = 3 and H ≤ G is the unique, non-trivial, proper subgroup of G, then observe that for any x ∈ G \ H, we must have x = G as these elements are non-trivial, must generate a subgroup of G, and cannot generate the trivial subgroup or H. Since G must be cyclic, the fact that cyclic groups have exactly one subgroup for each positive divisor of |G| implies that |G| = p 2 for some prime p and thus G ∼ = Z p 2 as cyclic groups of order |G| are unique up to isomorphism.Similarly, if | Sub G| = 4 and H = K are the two non-trivial subgroups, then recall thatOnce again, x = G and G is cyclic. As before, a cyclic group G must have exactly one subgroup for each divisor of |G|, hence it follows that G ∼ = Z pq or Z p 3 for primes p and q. We summarize these classic results below.
Classic Results.1. If | Sub G| = 2, then G ∼ = Z p for some prime p.
Iffor some primes p and q.These classic results beg the question: Which (necessarily finite) groups G have exactly k subgroups when k ≥ 5? Fortunately this becomes much more interesting moving forward as G need not be cyclic when | Sub G| ≥ 5. Thus, from here on we will need a completely different approach.Miller explored this topic previously in a series of obscure and terse papers [6,7,8,9,10] in which he claims to classify the groups with 16 or fewer subgroups, but it is unclear to the authors exactly how he arrives at his conclusions. Despite that, his results agree with ours, except in the case when | Sub G| = 14 where he seems to have skipped a case, causing him to miss S 3 × Z 3 and Z 3 ⋊ Z 32 . Given the assertions within, it is certainly clear that Miller is not applying the techniques we use here. Recently, Slattery [12] explored this idea once more; reducing any group G by factoring out any cyclic central Sylow p-subgroups of G first. Using this method, he worked to classify groups with 12 or fewer subgroups up to similarity defined in the following sense: Definition (From [12]). Let G and H be finite groups.