Automated Reasoning for Knot Semigroups and π-orbifold Groups of Knots 9
Mace4 disproves the goal by finding a model in which both the cancellative
semigroup axioms and the defining relations are satisfied, but at the same time,
the goal statement is false. The model found by Mace4 is the dihedral group D
3
,
which is the knot’s π-orbifold group, as discussed in Subsect. 1.5.
Table 2 shows the results for all standard knot diagrams with up to 9 cross-
ings. For each diagram we list the size of the model found and the time taken to
find this model. The results presented in non-bold font are obtained by running
Mace4 with the default iterative search strategy; that is, the search for a model
starts with the size 2; if no model is found by an exhaustive search of models of a
certain size, the size is increased by 1 and the search continues. Thus, assuming
correctness of Mace4, entries in non-bold font represent smallest possible mod-
els. In all these cases the size of the model is two times the size of a smallest kei
model computed in [6]; this observation has led us to formulating the following
conjecture.
Conjecture 2. Consider a knot diagram d. Suppose the kei of d has a factor kei
of size n. Then the semigroup Kd has a factor semigroup of size 2n.
To add some more details regarding the conjecture, the smallest semigroup model
is frequently the knot’s π-orbifold group, which is frequently (see Fact 2) a dihe-
dral group, and the size of a dihedral group is two times the size of the corre-
sponding dihedral kei (that is, the kei consisting of reflections), which is then
the smallest kei model of the same knot diagram (Proposition 2 in [6], Theorem
3in[7]). In some other cases (for example, 8
19
in Table 2, which is not a 4-plat),
the knot’s π-orbifold group is not a dihedral group, but the smallest semigroup
model, which is is a factor group of the knot’s π-orbifold group, happens to be
isomorphic to the dihedral group D
3
. We don’t know what happens to smallest
model sizes when the smallest semigroup model is not a dihedral group and the
smallest kei model is not a dihedral kei.
Table 2 contains remarks related to Conjecture 2. The entries in bold font
represent the diagrams for which the default iterative strategy of Mace4 has
failed to find a model in 50000 s. In this case we used Conjecture 2 to guess a
possible model size. These entries further split into three categories:
(1) the size is given with a mark
a
, meaning the search has been completed
successfully for this particular size, predicted by Conjecture 2; the conjecture
is confirmed, but the model found is not necessarily minimal;
(2) the size is given with a mark
b
, meaning the search has been done for increas-
ing model sizes and ended successfully, but this was not an exhaustive search,
as a time limit was imposed on search for each size; Conjecture 2 is con-
firmed, but the model found is not necessarily minimal;
(3) the size is given with a question mark and the time is given as N/F for ‘not
found’, meaning neither search strategy has succeeded to find a model in
50000 s; an estimated model size is given as predicted by Conjecture 2.
It is interesting to note that we could not find a model of the predicted size
30 in any entry in the table, as Mace4 search has timed out, although for larger