(H,M) = allSemigroups L(H,M) = allSemigroups mUsing Normaliz we compute the face of the Kunz cone containing L (or of the cone of semigroups of multiplicity m).
Every semigroup of multiplicity m has Apery set of the form m, mu_1*m+1, mu_2*m+2..mu_(m-1)*m+(m-1). The sequence of mu_i for a semigroup L is the value of mu(L).
|
|
|
Here the rows of the first matrix, H, form a Hilbert basis, that is, a set of semigroup generators, of the semigroup of the possible mu(L) as L runs over the semigroups of multiplicity 3. The rows of the the second matrix, M, when prepened by the multiplicity 3, are Apery sets of some semigroups such that every semigroup of multiplicity 3 has Apery set equal to a linear combination of the rows of H added to one of the rows of M.
|
This means that all the semigroups on the same (open) face of the Kunz cone as {4,7,9} can be obtained by taking a linear combination of the rows of H, adding one of the rows of M, and prepending the multiplicity 4.
On the face with the buchweitz example there are two facet rays:
|
The first row of H is 13*(mu buchweitz 0), the mu vector of the Buchweitz example. Adding multiples of the first row to the Weierstrass semigroups of an ordinary point on a curve of genus 12, we eventually reach a semigroup that fails the Buchweitz test to be a Weierstrass semigroup:
|
|
|
By Riemann-Roch the quantity 3*(genus L' -1)-#sums(G,G) is non-negative for Weierstrass semigroups. We conjecture that the same thing is true for any semigroup L0 of multiplicity 13 in place of L. Here is a "random" example:
|
|
|
The object allSemigroups is a method function.
The source of this document is in NumericalSemigroups.m2:2246:0.