## Product of Projective Cap with Cap in AG(6, b)

Let C1 be an s1-cap in the projective space PG(u1, b). Let C2 be an s2-cap in PG(u2, b) and C2ʹ ⊆ C2 an s2ʹ-cap in the affine space AG(u2, b). Then it is shown in [1, Theorem 8] that an (s1s2ʹ + s2s2ʹ)-cap in PG(u1 + u2, b) can be constructed.

If C2 is already in AG(u2, b) (and therefore C2 = C2ʹ), this construction reduces to the well-known product construction for caps , yielding an (s1s2)-cap in PG(u1 + u2, b) based on an s1-cap in PG(u1, b) and an s2-cap in AG(u2, b). The more general result in  allows to include the extra points in C2 ∖ C2ʹ to the final cap.

### Construction

Let

C1 = {a1,…, as1} ⊂ Fbu1+1,

and

C2 = {   ,…,   ,   ,…,   }

with bjFbu2 for j = 1,…, s2. Then the new cap consists of the disjoint union of the following two sets:

{   :  1 ≤ is1, 1 ≤ js2ʹ},

and

{   :  s2ʹ < js2},

with cardinality s1s2ʹ and s2s2ʹ, respectively.

The parent code / orthogonal array listed by MinT for this propagation rule is the code equivalent to C1. The following pairs C2ʹ ⊆ C2 are used depending on u2:

### The Trivial Cap for Dimension u2 = 1

For u2 = 1, the optimal solution is to use the trivial 2-cap in PG(1, b), which is also in AG(1, b). In this case, the product construction reduces to the well-known doubling of a cap: Given an s1-cap in PG(u1, b), a (2s1)-cap in PG(u1 + 1, b) can be constructed.

### The Oval and Hyperoval for Dimension u2 = 2

For u2 = 2, the oval and hyperoval are the largest caps in AG(2, b) as well as PG(2, b). Therefore, one can construct an (s1s2)-cap in PG(u1 + 2, b) with s2 = b + 1 if b is odd and s2 = b + 2 if b is even.

### The Ovoid for Dimension u2 = 3

For u2 = 3, the optimal solution is to use the affine part of the ovoid. It is a b2-cap in AG(3, b) which can be completed to a (b2 + 1)-cap in PG(3, b).

Using these caps as C2ʹ and C2, respectively, a (b2s1 + 1)-cap in PG(u1 + 3, b) can be constructed based on an s2-cap C1 in PG(u1, b). Thus, we have just recovered Segre’s construction from .

### Dimension u2 = 4

For u2 = 4 the optimal solution is not known for all b. MinT uses the following caps C2ʹ in AG(4, b). In some cases, this cap is already projectively-complete, so C2 = C2ʹ and we can use the original construction from .

 b s2ʹ s2 Reference 3 20 same Doubling the ovoid yields a cap in AG 4 40 same See  5 65 66 computer completed cap 7 127 132 computer completed cap 8 208 same computer completed cap 9 210 same computer completed cap 11 311 316 computer completed cap 13 387 388 computer completed cap 16 628 629 computer completed cap 32 3136 same computer completed cap

### Dimension u2 = 5

For u2 = 5 one can use the series of caps in PG(5, b). It follows from the proof of [1, Theorem 14] that each of these projective caps of size s2 = (b – 1)(b + 1)2 + 4(b + 1) contains an affine subcap of size s2ʹ = (b – 1)(b + 1)2 + 2(b + 1). For b = 9 one can even show that the 840-cap contains a subcap of size s2ʹ = 830 (private communication with Yves Edel).

For some small base fields better results are known:

 b s2ʹ s2 Reference 3 45 56 Hill cap 4 120 126 Glynn cap 5 176 186 computer completed cap 7 427 434 computer completed cap 8 694 695 computer completed cap

### Dimension u2 = 6

For u2 = 6, MinT uses the class of (b4 +2b2)-caps in PG(6, u) resulting from this construction based on two ovoid caps. It is shown in the proof for [1, Theorem 12] that these caps always contain an affine sub-cap of size s2ʹ = b4 + b2 − 1.