Numerical semigroups are additive submonoids $A$ of the natural numbers such that the greatest common divisor of all elements of $A$ is 1. The complement of a numerical semigroup in $\mathbb{N}$ is finite and is called the genus of the numerical semigroup.

The sequence A007323 in the OEIS gives the number of numerical semigroups with a given genus $g$. Zagier writes in the comments that this also gives the number of sets of $g$ power sum symmetric functions which form a $\mathbb{Q}$-algebra basis for the field of symmetric functions in $g$ variables. He attributes this result to Kakutani but says he forgot the reference.

Does anyone know the source of this result?


I think Zagier must have been thinking of the following papers of Kakeya, instead of Kakutani

Kakeya, S.: On fundamental systems of symmetric functions. I, II. Jap. J. Math.2, 69–80 (1925) ; 4, 77–85 (1927)

where the following theorem is established:

Theorem: Suppose that the complement of the sequence $k_1,k_2,\dots,k_n$ is a numerical semigroup, then the power sums $p_{k_1},p_{k_2},\dots,p_{k_n}$ generate the field $\mathbb Q(p_1,p_2,\dots,p_n)$ of symmetric functions in $n$ variables.

He made the conjecture that the converse also holds: if $Q(p_{k_1},p_{k_2},\cdots,p_{k_n})\cong Q(p_1,p_2,\cdots,p_n)$ then $k_1,k_2,\cdots,k_n$ must be the complement of a numerical semigroup. However this is still open in general (for example see section 6 of this recent paper of Dvornicich and Zannier) therefore the two descriptions of sequence A007323 are only conjecturally the same.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.