Hughes planes and their collineation groups
Angelo Sonnino
Università degli Studi della Basilicata Potenza, Italy
Finite Geometry Conference & Workshop
Szeged 10–14 June 2013
Projective geometries
Let V be an (n+1) dimensional vector space over a field F.
(x0, . . . ,xn)∼(y0, . . . ,yn) if and only if there exists k∈F\ {0} such that xi =kyi for all 0≤i ≤n. Then, the quotient V\{0}∼ =PG(n,F) is called the n-dimensional projective space over F.
Projective geometries
Let V be an (n+1) dimensional vector space over a field F.
(x0, . . . ,xn)∼(y0, . . . ,yn) if and only if there exists k∈F\ {0} such that xi =kyi for all 0≤i ≤n.
Then, the quotient V\{0}∼ =PG(n,F) is called the n-dimensional projective space over F.
Projective geometries
Let V be an (n+1) dimensional vector space over a field F.
(x0, . . . ,xn)∼(y0, . . . ,yn) if and only if there exists k∈F\ {0} such that xi =kyi for all 0≤i ≤n.
Then, the quotient V\{0}∼ =PG(n,F) is called the n-dimensional projective space over F.
Projective geometries
1-dimensional subspaces of V ⇒ point (projective subspace of dimension 0).
2-dimensional subspaces of V ⇒ line (projective subspace of dimension 1).
. . .
n-dimensional subspaces of V ⇒ hyperplane (projective subspace of dimension n−1).
Projective geometries
1-dimensional subspaces of V ⇒ point (projective subspace of dimension 0).
2-dimensional subspaces of V ⇒ line (projective subspace of dimension 1).
. . .
n-dimensional subspaces of V ⇒ hyperplane (projective subspace of dimension n−1).
Projective geometries
1-dimensional subspaces of V ⇒ point (projective subspace of dimension 0).
2-dimensional subspaces of V ⇒ line (projective subspace of dimension 1).
. . .
n-dimensional subspaces of V ⇒ hyperplane (projective subspace of dimension n−1).
Projective geometries
1-dimensional subspaces of V ⇒ point (projective subspace of dimension 0).
2-dimensional subspaces of V ⇒ line (projective subspace of dimension 1).
. . .
n-dimensional subspaces of V ⇒ hyperplane (projective subspace of dimension n−1).
The finite case
If F=GF(q) then we have the finite projective space of orderq containing exactly
qn+1−1
q−1 =qn+qn−1+· · ·+q+1
points.
LIttle Wedderburn’s theorem
Theorem
Every finite division ring is a field
The construction we have seen yields all possible finite projective geometries. . .
whenn >2.
LIttle Wedderburn’s theorem
Theorem
Every finite division ring is a field
The construction we have seen yields all possible finite projective geometries. . .
when n >2.
LIttle Wedderburn’s theorem
Theorem
Every finite division ring is a field
The construction we have seen yields all possible finite projective geometries. . . when n >2.
Projective planes
If n=2 then we have a projective plane, that is, an incidence structure of points and lines satisfying:
- Every two distinct points are on exactly one line;
- Every two lines meet at exactly one point;
- There are four points no three of which are collinear.
When n=2 and q ≥9 this construction does not produce all possible projective planes, but only the Desarguesian ones.
Projective planes
If n=2 then we have a projective plane, that is, an incidence structure of points and lines satisfying:
- Every two distinct points are on exactly one line;
- Every two lines meet at exactly one point;
- There are four points no three of which are collinear.
When n=2 and q ≥9 this construction does not produce all possible projective planes, but only the Desarguesian ones.
The Desargues’ theorem
XXXX
XXXX
XXXX XX
"
"
"
"
""
D D D D D D D DD
J J J J JJ
Q Q Q Q Q Q Q
s s s
s
s s s
s s s
The Desargues’ theorem
XXXX
XXXX
XXXX XX
"
"
"
"
""
D D D D D D D DD
J J J J JJ
Q Q Q Q Q Q Q
s s s
s
s s s
s s s
A first example
Veblen & Wedderburn (1907) exhibited a
non-Desarguesian planeπ of order 9 which turned out to. . .
- admit a collineation group of order 78, and
- be self-dual, that is isomorphic to its dual plane π∗ which is obtained from π by interchanging the role of points and lines.
A first example
Veblen & Wedderburn (1907) exhibited a
non-Desarguesian planeπ of order 9 which turned out to. . .
- admit a collineation group of order 78, and
- be self-dual, that is isomorphic to its dual plane π∗ which is obtained from π by interchanging the role of points and lines.
A first example
Veblen & Wedderburn (1907) exhibited a
non-Desarguesian planeπ of order 9 which turned out to. . .
- admit a collineation group of order 78, and
- be self-dual, that is isomorphic to its dual plane π∗ which is obtained from π by interchanging the role of points and lines.
V-W systems and nearfields
A left (right) Veblen-Wedderburn system R is a finite set containing ar least two elements: 0 and 1, with two binary operations “+” and “·” satisfying:
1. R is an additive group with identity 0;
2. R\ {0} is a multiplicative loop with identity 1
I the equationax+b=0 has a unique solutionx ∈R;
I the equationxa+b=0 has a unique solution x ∈R;
Distributive laws:
a(b+c) =ab+ac ⇒ left V-W system; (a+b)c =ac+bc ⇒ right V-W system.
V-W systems and nearfields
A left (right) Veblen-Wedderburn system R is a finite set containing ar least two elements: 0 and 1, with two binary operations “+” and “·” satisfying:
1. R is an additive group with identity 0;
2. R\ {0} is a multiplicative loop with identity 1
I the equationax+b=0 has a unique solution x ∈R;
I the equationxa+b=0 has a unique solution x∈R;
Distributive laws:
a(b+c) =ab+ac ⇒ left V-W system; (a+b)c =ac+bc ⇒ right V-W system.
V-W systems and nearfields
A left (right) Veblen-Wedderburn system R is a finite set containing ar least two elements: 0 and 1, with two binary operations “+” and “·” satisfying:
1. R is an additive group with identity 0;
2. R\ {0} is a multiplicative loop with identity 1
I the equationax+b=0 has a unique solution x ∈R;
I the equationxa+b=0 has a unique solution x∈R;
Distributive laws:
a(b+c) =ab+ac ⇒ left V-W system;
(a+b)c =ac+bc ⇒ right V-W system.
V-W systems and nearfields
A left (right) V-W system with the multiplicative associative law is a left (right) nearfield.
The centre of R is the set
Z(R) ={z ∈R|xz =zx for all x ∈R}.
V-W systems and nearfields
A left (right) V-W system with the multiplicative associative law is a left (right) nearfield.
The centre of R is the set
Z(R) ={z ∈R|xz =zx for all x ∈R}.
Exixtence of nearfields
Theorem (Zassenhous 1936)
There exists a nearfieldR (which is not a field) of orderq2=p2h for any odd primep and positive integerh whose centre is F 'GF(q).
Exixtence of nearfields
Take GF(q2) with q an odd prime power and define a new algebraic structureR with
elements: same as GF(q2);
sum: same as GF(q2);
multiplication:
x◦y =
(xy if x is a square inGF(q2) xyq if x is not a square in GF(q2).
R is a nearfield (not a field) of order q2 whose centre isGF(q).
Exixtence of nearfields
Take GF(q2) with q an odd prime power and define a new algebraic structureR with
elements: same as GF(q2);
sum: same as GF(q2);
multiplication:
x◦y =
(xy if x is a square inGF(q2) xyq if x is not a square in GF(q2).
R is a nearfield (not a field) of order q2 whose centre isGF(q).
Sporadic examples
Seven exceptional examples of orders 52, 112, 72, 232, again 112, 292 and 592 whose generators of the
multiplicative group are represented by2×2 matrices.
For q2=112, take the matrix group hA,B,Ci generated by
A=
0 −1 1 0
, B =
1 5
−5 −2
, C =
4 0 0 4
;
elements of R are of type x=x1+x2t with t2=τ, τ a non-square element of GF(q).
Sporadic examples
Seven exceptional examples of orders 52, 112, 72, 232, again 112, 292 and 592 whose generators of the
multiplicative group are represented by2×2 matrices.
For q2=112, take the matrix group hA,B,Ci generated by
A=
0 −1 1 0
, B=
1 5
−5 −2
, C =
4 0 0 4
;
elements of R are of type x=x1+x2t with t2=τ, τ a non-square element ofGF(q).
(Left) vector spaces
R a left nearfield of order q2 =p2h, with p and odd prime and F 'GF(q) the centre of R.
V set of all triples (x,y,z) with x,y,z ∈R, and V0 subset of all triples (x,y,z) with x,y,z ∈F.
Let A be a linear transformation of V0 over F. (x,y,z)Am=
(a11x+a12y+a13z,a21x+a22y+a23z,a31x+a32y+a33z),
where aij ∈F depend on m.
(Left) vector spaces
R a left nearfield of order q2 =p2h, with p and odd prime and F 'GF(q) the centre of R.
V set of all triples (x,y,z) with x,y,z ∈R, and V0 subset of all triples (x,y,z) with x,y,z ∈F.
Let A be a linear transformation of V0 over F. (x,y,z)Am=
(a11x+a12y+a13z,a21x+a22y+a23z,a31x+a32y+a33z),
where aij ∈F depend on m.
(Left) vector spaces
R a left nearfield of order q2 =p2h, with p and odd prime and F 'GF(q) the centre of R.
V set of all triples (x,y,z) with x,y,z ∈R, and V0 subset of all triples (x,y,z) with x,y,z ∈F.
Let A be a linear transformation of V0 over F. (x,y,z)Am=
(a11x+a12y+a13z,a21x+a22y+a23z,a31x+a32y+a33z),
where aij ∈F depend on m.
(Left) vector spaces
Suppose that A has the property that for any v0 ∈V0
v0Am=kv0 for some k ∈R\ {0} if and only if m≡0 (modq2+q+1).
Then, A induces a linear transformation of V over R. Further, V is partitioned into point-orbits of length q2+q+1.
(Left) vector spaces
Suppose that A has the property that for any v0 ∈V0
v0Am=kv0 for some k ∈R\ {0} if and only if m≡0 (modq2+q+1).
Then, A induces a linear transformation of V over R.
Further, V is partitioned into point-orbits of length q2+q+1.
(Left) vector spaces
Suppose that A has the property that for any v0 ∈V0
v0Am=kv0 for some k ∈R\ {0} if and only if m≡0 (modq2+q+1).
Then, A induces a linear transformation of V over R.
Further, V is partitioned into point-orbits of length q2+q+1.
A new incidence structure π
Points
Triples (x,y,z)∈V with the identification (x,y,z) = (kx,ky,kz) for all k ∈R\ {0}.
Lines
Formal symbols LtAm where either t =1 or t ∈R\F, 0≤m≤q2+q and we write just Lt when m=0.
Incidence
The point (x,y,z) is on Lt if and only if x+yt +z =0 whileLtAm contains all the points (x,y,z)Am such that (x,y,z) is in Lt.
A new incidence structure π
Points
Triples (x,y,z)∈V with the identification (x,y,z) = (kx,ky,kz) for all k ∈R\ {0}.
Lines
Formal symbols LtAm where either t =1 or t ∈R\F, 0≤m≤q2+q and we write just Lt when m=0.
Incidence
The point (x,y,z) is on Lt if and only if x+yt +z =0 whileLtAm contains all the points (x,y,z)Am such that (x,y,z) is in Lt.
A new incidence structure π
Points
Triples (x,y,z)∈V with the identification (x,y,z) = (kx,ky,kz) for all k ∈R\ {0}.
Lines
Formal symbols LtAm where either t =1 or t ∈R\F, 0≤m≤q2+q and we write just Lt when m=0.
Incidence
The point (x,y,z) is on Lt if and only if x+yt +z =0 whileLtAm contains all the points (x,y,z)Am such that (x,y,z) is in Lt.
A new projective plane π
- There are exactly q4+q2+1 points in π.
- There are exactly q4+q2+1 lines in π.
- Two lines meet at exactly one point because. . . Theorem (Ryser 1950)
LetX ={x1, . . . ,xv} be a finite set and letT1, . . . , Tv be sets consistinf of elements fromX. If eachTj contains exactlys distinct elements ofX end every pair of distinct sets Ti,Tj shares exactlyλ elements then
λ= s(s−1) v−1 .
A new projective plane π
- There are exactly q4+q2+1 points in π.
- There are exactly q4+q2+1 lines in π.
- Two lines meet at exactly one point because. . . Theorem (Ryser 1950)
LetX ={x1, . . . ,xv} be a finite set and letT1, . . . , Tv be sets consistinf of elements fromX. If eachTj contains exactlys distinct elements ofX end every pair of distinct sets Ti,Tj shares exactlyλ elements then
λ= s(s−1) v−1 .
A new projective plane π
- There are exactly q4+q2+1 points in π.
- There are exactly q4+q2+1 lines in π.
- Two lines meet at exactly one point because. . .
Theorem (Ryser 1950)
LetX ={x1, . . . ,xv} be a finite set and letT1, . . . , Tv be sets consistinf of elements fromX. If eachTj contains exactlys distinct elements ofX end every pair of distinct sets Ti,Tj shares exactlyλ elements then
λ= s(s−1) v−1 .
A new projective plane π
- There are exactly q4+q2+1 points in π.
- There are exactly q4+q2+1 lines in π.
- Two lines meet at exactly one point because. . . Theorem (Ryser 1950)
LetX ={x1, . . . ,xv} be a finite set and letT1, . . . , Tv be sets consistinf of elements fromX. If eachTj contains exactlys distinct elements ofX end every pair of distinct sets Ti,Tj shares exactlyλ elements then
λ= s(s −1) v−1 .
A new projective plane π
The existence a linear transformation A0 of V0 over F with the required property is guaranteed by the work of Singer (1938).
The generator of the cyclic collineation group of PG(2,q) provided by Singer yields the required A0
mapping V0 onto itself, and hence preserving a
“subplane” π0 of π which turns out to be PG(2,q). The lines of π0 are those of the form L1Am.
There are exactly q2+q+1 such lines.
A new projective plane π
The existence a linear transformation A0 of V0 over F with the required property is guaranteed by the work of Singer (1938).
The generator of the cyclic collineation group of PG(2,q) provided by Singer yields the required A0
mapping V0 onto itself, and hence preserving a
“subplane” π0 of π which turns out to be PG(2,q).
The lines of π0 are those of the form L1Am.
There are exactly q2+q+1 such lines.
A new projective plane π
The existence a linear transformation A0 of V0 over F with the required property is guaranteed by the work of Singer (1938).
The generator of the cyclic collineation group of PG(2,q) provided by Singer yields the required A0
mapping V0 onto itself, and hence preserving a
“subplane” π0 of π which turns out to be PG(2,q).
The lines of π0 are those of the form L1Am.
There are exactly q2+q+1 such lines.
A new projective plane π
The existence a linear transformation A0 of V0 over F with the required property is guaranteed by the work of Singer (1938).
The generator of the cyclic collineation group of PG(2,q) provided by Singer yields the required A0
mapping V0 onto itself, and hence preserving a
“subplane” π0 of π which turns out to be PG(2,q).
The lines of π0 are those of the form L1Am. There are exactly q2+q+1 such lines.
A simpler construction
By Rosati (1960)
Back to R with multiplication defined by x◦y =
(xy if x is a square inGF(q2) xyq if x is not a square in GF(q2).
Let P and L be two copies of the set of triples (x,y,z) with x,y,z ∈R and identification as before.
A simpler construction
By Rosati (1960)
Back to R with multiplication defined by x◦y =
(xy if x is a square inGF(q2) xyq if x is not a square in GF(q2).
Let P and L be two copies of the set of triples (x,y,z) with x,y,z ∈R and identification as before.
A simpler construction
Fix t∈R\F.
Let P = (x,y,z)∈P and L= (u,v,w)∈L with u =a+a1t, v=b+b1t, w =c+c1t, and a,a1,b,b1,c,c1 ∈F.
Define a relation “∼” between elements of P and elements of L where P ∼L if and only if
xa+yb+zc+ (xa1+yb1+zc1)◦t =0.
A simpler construction
Fix t∈R\F.
Let P = (x,y,z)∈P and L= (u,v,w)∈L with u =a+a1t, v=b+b1t, w =c+c1t, and a,a1,b,b1,c,c1∈F.
Define a relation “∼” between elements of P and elements of L where P ∼L if and only if
xa+yb+zc+ (xa1+yb1+zc1)◦t =0.
A simpler construction
Fix t∈R\F.
Let P = (x,y,z)∈P and L= (u,v,w)∈L with u =a+a1t, v=b+b1t, w =c+c1t, and a,a1,b,b1,c,c1∈F.
Define a relation “∼” between elements of P and elements of L where P ∼L if and only if
xa+yb+zc+ (xa1+yb1+zc1)◦t =0.
A simpler construction
- The relation ∼ is independent of the choice of t ∈R\F.
- P ∼L if and only if L∼P.
- If P ∼L and k1,k2 ∈R\ {0} then (k1◦P)∼(k2◦L).
The realtion ∼ induces an incidence relation between P and L, and the incidence structure (P,L,∼) so defined is a Hughes plane of order q2=p2h.
A simpler construction
- The relation ∼ is independent of the choice of t ∈R\F.
- P ∼L if and only if L∼P.
- If P ∼L and k1,k2 ∈R\ {0} then (k1◦P)∼(k2◦L).
The realtion ∼ induces an incidence relation between P and L, and the incidence structure (P,L,∼) so defined is a Hughes plane of order q2 =p2h.
A simpler construction
In this setting, for some 0≤m≤q2+q the equation xa+yb+zc + (xa1+yb1+zc1)◦t =0
represents a lineLtAm in a natural manner.
Collineation groups
The Hughes plane π of order q2 admits a cyclic
collineation group S of order q2+q+1, that is, some collineations of π can be obtained by extending to the whole plane the action of the collineations of the cyclic collineation group of its (Desarguesian) subplane π0.
For q=3 (Veblen-Wedderburn) π admits also a collineation group T of order 6 fixing π0 pointwise. This is the automorphism group of the nearfield R which fixes F =GF(3) and is isomorphic to Sym(3).
Collineation groups
The Hughes plane π of order q2 admits a cyclic
collineation group S of order q2+q+1, that is, some collineations of π can be obtained by extending to the whole plane the action of the collineations of the cyclic collineation group of its (Desarguesian) subplane π0. For q=3 (Veblen-Wedderburn) π admits also a collineation group T of order 6 fixing π0 pointwise.
This is the automorphism group of the nearfield R which fixes F =GF(3) and is isomorphic to Sym(3).
Collineation groups
If θ∈Aut(R) =T, then under the action of θ we have that
L1Am 7→L1Am for every 0≤m≤q2+q while
LtAm 7→LtθAm with t6=tθ when t 6=1.
Collineation groups
S and T have no common element (apart from the identity) and for each pairσ ∈S and τ ∈T we have στ =τ σ.
Hence π admits the collineation group G =S ×T, whose order is 78.
Is that it?
Collineation groups
S and T have no common element (apart from the identity) and for each pairσ ∈S and τ ∈T we have στ =τ σ.
Henceπ admits the collineation group G =S ×T, whose order is 78.
Is that it?
Collineation groups
S and T have no common element (apart from the identity) and for each pairσ ∈S and τ ∈T we have στ =τ σ.
Henceπ admits the collineation group G =S ×T, whose order is 78.
Is that it?
Collineation groups
Zappa (1957) & Rosati (1958)
Every collineation of π0 can be extended to a (unique) collineation ofπ so that π admits a collineation group which is isomorphic to the total collineation group of π0 'PG(2,q).
If q2=9 then the total collineation group Σ of π is Σ =S×T 'PGL(3,q)×Sym(3),
hence |Σ|=6·5,616=33,696.
Collineation groups
Zappa (1957) & Rosati (1958)
Every collineation of π0 can be extended to a (unique) collineation ofπ so that π admits a collineation group which is isomorphic to the total collineation group of π0'PG(2,q).
If q2=9 then the total collineation group Σ of π is Σ =S×T 'PGL(3,q)×Sym(3),
hence |Σ|=6·5,616=33,696.
Collineation groups
Zappa (1957) & Rosati (1958)
Every collineation of π0 can be extended to a (unique) collineation ofπ so that π admits a collineation group which is isomorphic to the total collineation group of π0'PG(2,q).
If q2=9 then the total collineation group Σ of π is Σ =S×T 'PGL(3,q)×Sym(3),
hence |Σ|=6·5,616=33,696.
Collineation groups
Zappa (1957) & Rosati (1958)
Every collineation of π0 can be extended to a (unique) collineation ofπ so that π admits a collineation group which is isomorphic to the total collineation group of π0'PG(2,q).
If q2=9 then the total collineation group Σ of π is Σ =S×T 'PGL(3,q)×Sym(3),
hence |Σ|=6·5,616=33,696.
Collineation groups
Note that one has Σ =S ×T if and only if q2=p2 with p an odd prime, and in this case each collineation of S commutes with each collineation of T.
If q2=p2h then T is a cyclic group of order 2h. If q2=p2h6=9 then
|Σ|=2mq3(q2+q+1)(q−1)2(q+1),
that is,
|Σ|=2|PΓL(3,q)|.
Collineation groups
Note that one has Σ =S ×T if and only if q2=p2 with p an odd prime, and in this case each collineation of S commutes with each collineation of T.
If q2=p2h then T is a cyclic group of order 2h.
If q2=p2h6=9 then
|Σ|=2mq3(q2+q+1)(q−1)2(q+1),
that is,
|Σ|=2|PΓL(3,q)|.
Collineation groups
Note that one has Σ =S ×T if and only if q2=p2 with p an odd prime, and in this case each collineation of S commutes with each collineation of T.
If q2=p2h then T is a cyclic group of order 2h.
If q2=p2h6=9 then
|Σ|=2mq3(q2+q+1)(q−1)2(q+1),
that is,
|Σ|=2|PΓL(3,q)|.
Collineation groups
In particular, when q2 =25 we have
|Σ|=2·31·30·25·16=2·372,000=744,000,
and Σ has two orbits on π:
- one orbit is π0 and - the other one is π\π0.
Another construction (Bose 1973)
In PG(2,q2), with q an odd prime power, fix a real line as the line at the infinity `∞, let α=AG(2,q2) be the affine plane and α0 =AG(2,q) its real part.
There is a natural classification for the lines of α with respect to the collineation induced by the Frobenius automorphismx 7→xq of GF(q2):
- Lines of type R: real lines, with real direction, q real points and q2−q complex points; `q=`.
- Lines of type 0P: complex lines with real direction;
`q is parallel to `.
- Lines of type 1: complex lines with one real point and complex direction; `q∩`={V} with V ∈α0 (the point V is the vertex of `).
Another construction (Bose 1973)
In PG(2,q2), with q an odd prime power, fix a real line as the line at the infinity `∞, let α=AG(2,q2) be the affine plane and α0 =AG(2,q) its real part.
There is a natural classification for the lines of α with respect to the collineation induced by the Frobenius automorphismx 7→xq of GF(q2):
- Lines of type R: real lines, with real direction, q real points and q2−q complex points; `q=`.
- Lines of type 0P: complex lines with real direction;
`q is parallel to `.
- Lines of type 1: complex lines with one real point and complex direction; `q∩`={V} with V ∈α0 (the point V is the vertex of `).
Another construction (Bose 1973)
In PG(2,q2), with q an odd prime power, fix a real line as the line at the infinity `∞, let α=AG(2,q2) be the affine plane and α0 =AG(2,q) its real part.
There is a natural classification for the lines of α with respect to the collineation induced by the Frobenius automorphismx 7→xq of GF(q2):
- Lines of type R: real lines, with real direction, q real points and q2−q complex points; `q=`.
- Lines of type 0P: complex lines with real direction;
`q is parallel to `.
- Lines of type 1: complex lines with one real point and complex direction; `q∩`={V} with V ∈α0 (the point V is the vertex of `).
Another construction (Bose 1973)
In PG(2,q2), with q an odd prime power, fix a real line as the line at the infinity `∞, let α=AG(2,q2) be the affine plane and α0 =AG(2,q) its real part.
There is a natural classification for the lines of α with respect to the collineation induced by the Frobenius automorphismx 7→xq of GF(q2):
- Lines of type R: real lines, with real direction, q real points and q2−q complex points; `q=`.
- Lines of type 0P: complex lines with real direction;
`q is parallel to `.
- Lines of type 1: complex lines with one real point and complex direction; `q∩`={V} with V ∈α0 (the point V is the vertex of `).
Another construction (Bose 1973)
Let ` be a line of type 1. Then, if d= (l,m) is the
direction vector of `, any point P ∈` can be written as P =V +λd
with λ∈GF(q2).
P is called a “green” or a “red” point of ` according as λ is a square element or a non-square element of GF(q2).
G(`) ={P ∈`\ {V} |P is green}, R(`) ={P ∈`\ {V} |P is red}.
Another construction (Bose 1973)
Let ` be a line of type 1. Then, if d= (l,m) is the
direction vector of `, any point P ∈` can be written as P =V +λd
with λ∈GF(q2).
P is called a “green” or a “red” point of ` according as λ is a square element or a non-square element of GF(q2).
G(`) ={P ∈`\ {V} |P is green}, R(`) ={P ∈`\ {V} |P is red}.
Another construction (Bose 1973)
Let ` be a line of type 1. Then, if d= (l,m) is the
direction vector of `, any point P ∈` can be written as P =V +λd
with λ∈GF(q2).
P is called a “green” or a “red” point of ` according as λ is a square element or a non-square element of GF(q2).
G(`) ={P ∈`\ {V} |P is green}, R(`) ={P ∈`\ {V} |P is red}.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2); - lines are:
I the lines of type R ofPG(2,q2)with the same
“direction”;
I the lines of type 0p of PG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2);
- lines are:
I the lines of type R ofPG(2,q2)with the same
“direction”;
I the lines of type 0p of PG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2);
- lines are:
I the lines of type R ofPG(2,q2)with the same
“direction”;
I the lines of type 0p ofPG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2);
- lines are:
I the lines of type R of PG(2,q2)with the same
“direction”;
I the lines of type 0p ofPG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2);
- lines are:
I the lines of type R of PG(2,q2)with the same
“direction”;
I the lines of type 0p ofPG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Define a new incidence structure α of points and lines (with parallelism) where
- points are the same as in PG(2,q2);
- lines are:
I the lines of type R of PG(2,q2)with the same
“direction”;
I the lines of type 0p ofPG(2,q2) with the same
“direction”;
I the lines defined as
{V} ∪ R(`)∪ G(`q) with the “direction” of `.
Another construction (Bose 1973)
Incidende is the set-theoretical inclusion and
two lines are “parallel” if and only if they have the same
“direction”.
The incidence structure so defined turns out to be a non-Desarguesian affine plane whose projective closure (by means of the line at the infinity `∞) is a Hughes plane of order q2.
Another construction (Bose 1973)
Incidende is the set-theoretical inclusion and
two lines are “parallel” if and only if they have the same
“direction”.
The incidence structure so defined turns out to be a non-Desarguesian affine plane whose projective closure (by means of the line at the infinity `∞) is a Hughes plane of orderq2.
Concluding remarks
The Hughes plane π of order q2, with q and odd prime power can be obtained from the Desarguesian plane PG(2,q2) and one of its partitions into Baer subplanes.
- Take π0 =PG(2,q) as one of the subplanes of such a partition P.
- The other subplanes of P come in conjugate pairs under x 7→xq.
- Keep π0 and give an alternative collinearity condition for the points on the complex lines (tangents to π0).
Concluding remarks
The Hughes plane π of order q2, with q and odd prime power can be obtained from the Desarguesian plane PG(2,q2) and one of its partitions into Baer subplanes.
- Take π0 =PG(2,q) as one of the subplanes of such a partition P.
- The other subplanes of P come in conjugate pairs under x 7→xq.
- Keep π0 and give an alternative collinearity condition for the points on the complex lines (tangents to π0).
Concluding remarks
The Hughes plane π of order q2, with q and odd prime power can be obtained from the Desarguesian plane PG(2,q2) and one of its partitions into Baer subplanes.
- Take π0 =PG(2,q) as one of the subplanes of such a partition P.
- The other subplanes of P come in conjugate pairs under x 7→xq.
- Keep π0 and give an alternative collinearity condition for the points on the complex lines (tangents to π0).
Concluding remarks
The Hughes plane π of order q2, with q and odd prime power can be obtained from the Desarguesian plane PG(2,q2) and one of its partitions into Baer subplanes.
- Take π0 =PG(2,q) as one of the subplanes of such a partition P.
- The other subplanes of P come in conjugate pairs under x 7→xq.
- Keep π0 and give an alternative collinearity condition for the points on the complex lines (tangents to π0).
Concluding remarks
- The complex points in π\π0 are partitioned into q2−q subsets Ωi, with |Ωi|=q2+q+1 for all 1≤i ≤q2−q, coming from the subplanes other than π0 in P.
- Each Ωi comes from an orbit of a point under σq2−q+1, with σ the Singer cycle of PG(2,q2), and in π is the orbit of a point under σ0 =σq2−q+1, with σ0 the Singer Cycle of π0.
- The characters of the sets Ωi other than π0 are: 0, 1, 2, q+1
2 , q+3 2 . - See de Resmini (1985) and (1987) for the
combinatorial properties of these sets.
Concluding remarks
- The complex points in π\π0 are partitioned into q2−q subsets Ωi, with |Ωi|=q2+q+1 for all 1≤i ≤q2−q, coming from the subplanes other than π0 in P.
- Each Ωi comes from an orbit of a point under σq2−q+1, with σ the Singer cycle of PG(2,q2), and in π is the orbit of a point under σ0=σq2−q+1, with σ0 the Singer Cycle of π0.
- The characters of the sets Ωi other than π0 are: 0, 1, 2, q+1
2 , q+3 2 . - See de Resmini (1985) and (1987) for the
combinatorial properties of these sets.
Concluding remarks
- The complex points in π\π0 are partitioned into q2−q subsets Ωi, with |Ωi|=q2+q+1 for all 1≤i ≤q2−q, coming from the subplanes other than π0 in P.
- Each Ωi comes from an orbit of a point under σq2−q+1, with σ the Singer cycle of PG(2,q2), and in π is the orbit of a point under σ0=σq2−q+1, with σ0 the Singer Cycle of π0.
- The characters of the sets Ωi other than π0 are:
0, 1, 2, q+1
2 , q+3 2 .
- See de Resmini (1985) and (1987) for the combinatorial properties of these sets.
Concluding remarks
- The complex points in π\π0 are partitioned into q2−q subsets Ωi, with |Ωi|=q2+q+1 for all 1≤i ≤q2−q, coming from the subplanes other than π0 in P.
- Each Ωi comes from an orbit of a point under σq2−q+1, with σ the Singer cycle of PG(2,q2), and in π is the orbit of a point under σ0=σq2−q+1, with σ0 the Singer Cycle of π0.
- The characters of the sets Ωi other than π0 are:
0, 1, 2, q+1
2 , q+3 2 . - See de Resmini (1985) and (1987) for the
combinatorial properties of these sets.