On the 430-cap of \(\mathrm{PG}(6,4)\) having two intersection sizes with respect to hyperplanes (Q6123061): Difference between revisions

A \(t\)-cap in \(\mathrm{PG}(d,q)\), \(d\geq 2\), that is, a set of \(t\) points no three on a line, which is at the same time a two-character set, that is, the size of a hyperplane intersection takes exactly two values, has the characteristic property that the dual of the associated code has minimal distance 4. Hence there is some interest in classifying such objects. It is known since more than 40 years that such a cap is either a hyperoval in \(\mathrm{PG}(2,q)\), an ovoid in \(\mathrm{PG}(3,q)\), the complement of a hyperplane in \(\mathrm{PG}(d,2)\), the \(11\)-cap in \(\mathrm{PG}(4,3)\) associated to the dual ternary Golay code, the Hill \(56\)-cap in \(\mathrm{PG}(5,3)\), a \(78\)-cap in \(\mathrm{PG}(5,4)\), or a \(430\)-cap in \(\mathrm{PG}(6,4)\). Whilst it is known that the Hill \(56\)-cap in \(\mathrm{PG}(5,3)\) is unique, the other two sporadic cases remain mysterious in that an example of a \(78\)-cap in \(\mathrm{PG}(5,4)\) is known, namely the Hill \(78\)-cap, but it is not known whether it is unique; furthermore, no example of a \(430\)-cap in \(\mathrm{PG}(6,4)\) is known to exist. The paper under review connects both mysteries with each other by showing that, if the Hill \(78\)-cap is unique, then no \(430\)-cap exists. The following quite clever method is employed. The intersection numbers with respect to hyperplanes of a putative \(430\)-cap \(K\) in \(\mathrm{PG}(6,4)\) are \(78\) and \(110\). If a hyperplane \(H\) intersects \(K\) in \(78\) points, then this intersection is again a cap and a two-character set (although the author claims this is well-known, he does not have a reference and prove it). But then the author considers the partial quadrangle \(\Gamma\) whose point set is \(\mathrm{PG}(6,4)\setminus H\) and whose lines are the lines of \(\mathrm{PG}(6,4)\) intersecting \(H\) in precisely one point, which on top belongs to \(K\) (this is the so-called linear representation of \(\Gamma\)). The points of \(K\) off \(H\) then form a coclique in the point graph of \(\Gamma\) of size \(352\) attaining the Delsarte/Hoffman bound. The author shows that the point graph of \(\Gamma\) does not contain such cocliques. And this is done using the Schurian scheme for the automorphism group of \(\Gamma\). Not everything in the proof can be done by hand, but the author includes his GAP code in an appendix.