SSL gpolygons.xml Sprache: XML

This is the chapter of the documentation describing generalised polygons.
-->

<Chapter Label="gpolygons">
<Heading>Generalised Polygons</Heading>

A <E>generalised n-gon</E> is a point/line geometry whose incidence graph is bipartite
of diameter <A>n</A> and girth <A>2n</A>. Although these rank 2 structures are very
much a subdomain of <Package>GRAPE</Package> and <Package>Design</Package>, their significance in
finite geometry warrants their inclusion in <Package>FinInG</Package>. By the famous
theorem of Feit and Higman, a generalised n-gon which has at least three points on every line,
must have <M>n</M> in <M>\{2,3,4,6,8\}</M>. The case <M>n=2</M> concerns the complete multipartite graphs,
which we disregard. The more interesting cases are accordingly projective planes (<M>n=3</M>),
generalised quadrangles (<M>n=4</M>), generalised hexagons (<M>n=6</M>), and generalised octagons (<M>n=8</M>).

<Package>FinInG</Package> provides some basic functionality to deal with generalised polygons as
incidence geometries. A lot of non-trivial interaction with the package <Package>GRAPE</Package> has been
very useful and even necessary. Currently, generic functions to create generalised polygons, to create
elements of generalised polygons, and to explore the elements are implemented. This generic functionality
allows the user to construct generalised polygons through many different objects available in
GAP and <Package>FinInG</Package>. Apart from these generic functions, some particular generalised polygons are
available: the classical generalised hexagons and elation generalised quadrangles from different perspectives
can be constructed.

<Section>
<Heading>Categories</Heading>

<ManSection>
<Filt Name="IsGeneralisedPolygon" Type="Category"/>
<Filt Name="IsGeneralisedPolygonRep" Type="Representation"/>
<Description>
This category is a subcategory of <C>IsIncidenceGeometry</C>, and contains all generalised polygons.
Generalised polygons constructed through functions described in this chapter, all belong
to <C>IsGeneralisedPolygonRep</C>.
</Description>
</ManSection>

<ManSection>
<Heading>Subcategories in <C>IsGeneralisedPolygon</C></Heading>
<Filt Name="IsProjectivePlaneCategory" Type="Category"/>
<Filt Name="IsGeneralisedQuadrangle" Type="Category"/>
<Filt Name="IsGeneralisedHexagon" Type="Category"/>
<Filt Name="IsGeneralisedOctagon" Type="Category"/>
<Description>
All generalised polygons in <Package>FinInG</Package> belong to one of these four categories. It is not
possible to construct generalised polygons of which the gonality is not known (or checked). Note that the
classical generalised quadrangles (which are the classical polar spaces of rank 2) belong
also to <C>IsGeneralisedQuadrangle</C> and that the desarguesian projective planes (which are the projective
spaces of dimension 2) also belong to <C>IsProjectivePlaneCategory</C>, but both do not belong
to <C>IsGeneralisedPolygonRep</C>.
</Description>
</ManSection>

<ManSection>
<Filt Name="IsWeakGeneralisedPolygon" Type="Category"/>
<Description>
<C>IsWeakGeneralisedPolygon</C> is the category for weak generalised polygons.
</Description>
</ManSection>

<ManSection>
<Heading>Subcategories in <C>IsProjectivePlaneCategory</C></Heading>
<Filt Name="IsDesarguesianPlane" Type="Category"/>
<Description>
<C>IsDesarguesianPlane</C> is declared as a subcategory of <C>IsProjectivePlaneCategory</C> and
<C>IsProjecticeSpace</C>. Projective spaces of dimension 2 constructed using <F>ProjectiveSpace</F>
belong to <C>IsDesarguesianPlane</C>.
</Description>
</ManSection>

<ManSection>
<Heading>Subcategories in <C>IsGeneralisedQuadrangle</C></Heading>
<Filt Name="IsClassicalGQ" Type="Category"/>
<Filt Name="IsElationGQ" Type="Category"/>
<Description>
<C>IsClassicalGQ</C> is declared as a subcategory of <C>IsGeneralisedQuadrangle</C> and
<C>IsClassicalPolarSpace</C>. All classical polar spaces of rank 2 belong to <C>IsClassicalGQ</C>.
<C>IsElationGQ</C> is declared as subcategory of <C>IsGeneralisedQuadrangle</C>. Elation GQs will be
discussed in detail in Section <Ref Sect="EGQ"/>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_categories1.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Filt Name="IsClassicalGeneralisedHexagon" Type="Category"/>
<Description>
<C>IsClassicalGeneralisedHexagon</C> is declared as subcategory of <C>IsGeneralisedHexagon</C> and <C>IsLieGeometry</C>.
The so called classical generalised hexagons are the hexagons that come from the triality of the hyperbolic quadric
&qplus7q;.
The Split Cayley hexagon is embedded in the parabolic quadric &q6q;,
or &w5q; in even characteristic. The twisted triality hexagon
is embedded in the hyperbolic quadric &qplus7q;. The construction of these hexagons in a subcategory
of <C>IsLieGeometry</C> means that the usual operations for Lie geometries become applicable.
The classical generalised hexagons are in detail discussed in Section <Ref Sect="CGH"/>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_categories2.include">
</Example>
</Description>
</ManSection>

</Section>

<Section>
<Heading>Generic functions to create generalised polygons</Heading>

<ManSection>
<Oper Name="GeneralisedPolygonByBlocks" Arg="l"/>
<Returns>a generalised polygon</Returns>
<Description>
The argument <A>l</A> is a finite homogeneous list consisting of
ordered sets of a common size <M>n+1</M>. This operation will assume that
each element of <A>l</A> represents a line of the generalised polygon. Its
points are assumed to be the union of all elements of <A>l</A>. The incidence
is assumed to be symmetrised containment. From this information, an incidence
graph is computed using <Package>GRAPE</Package>. If this graph has diameter <M>d</M>
and girth <M>2d</M>, a generalised polygon is returned. The thickness condition is not checked.
If <M>d \in \{3,4,6,8\}</M>, a projective plane, a generalised quadrangle, a generalised hexagon,
a generalised octagon respectively, is returned. Note that for large input, this operation
can be time consuming.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_gpbyblocks.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="GeneralisedPolygonByIncidenceMatrix" Arg="incmat"/>
<Returns>a generalised polygon</Returns>
<Description>
The argument <M>incmat</M> is a matrix representing the incidence matrix of a point line geometry.
The points are represented by the columns, the rows represent the lines. From <A>incmat</A> a homogeneous
list of sets of column entries is derived, which is then passed to <F>GeneralisedPolygonByBlocks</F>. When
<A>incmat</A> indeed represents a generalised polygon, it is returned. The checks are performed
by <F>GeneralisedPolygonByBlocks</F>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_gpbyincmat.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="GeneralisedPolygonByElements" Arg="pts, lns, inc"/>
<Oper Name="GeneralisedPolygonByElements" Arg="pts, lns, inc, grp, act"/>
<Returns>a generalised polygon</Returns>
<Description>
The argument <A>pts</A>, <A>lns</A> and <A>inc</A> are respectively a set of objects, a set of objects
and a function. The function <A>inc</A> must represent an incidence relation between objects of <A>pts</A>
and <A>lns</A>. The first version of <F>GeneralisedPolygonByElements</F> will construct an incidence graph, and
if this graph has diameter <M>d</M>
and girth <M>2d</M>, a generalised polygon is returned. The thickness condition is not checked.
If <M>d \in \{3,4,6,8\}</M>, a projective plane, a generalised quadrangle, a generalised hexagon,
a generalised octagon respectively, is returned.
The argument <A>grp</A> is a group, and <A>act</A> a function, representing
an action of the elements of <A>grp</A> on the objects in the lists <A>pts</A> and <A>lns</A>, preserving
the incidence. The second version of <F>GeneralisedPolygonByElements</F> acts as the first version, but
uses <A>grp</A> and <A>act</A> to construct the incidence graph in a more efficient way, so
if <A>grp</A> is a non trivial group, the construction of the graph will
be faster. This operation can typically be used to construct generalised polygons from objects that are available
in <Package>FinInG</Package>. This difference in time is shown in the first two examples. The third examples
shows the construction of the generalised quadrangle
<Alt Only="HTML MathJax"><M>T_2(O)</M></Alt><Alt Not="HTML"><M>T_2(O)</M></Alt><Alt Only="HTML noMathJax">T2(O)</Alt>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_gpbyelements.include">
</Example>
</Description>
</ManSection>

</Section>

<Section Label="gp:operations">
<Heading>Attributes and operations for generalised polygons</Heading>

All operations described in this section are applicable on objects in the category <C>IsGeneralisedPolygon</C>.

<ManSection>
<Attr Name="Order" Arg="gp"/>
<Returns>a pair of positive integers</Returns>
<Description>
This method returns the parameters <M>(s,t)</M> of the generalised polygon <A>gp</A>. That is,
<M>s+1</M> is the number of points on any line of <A>gp</A>, and <M>t+1</M> is the number
of lines incident with any point of <A>gp</A>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_order.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="IncidenceGraphAttr" Arg="gp"/>
<Description>
This attribute is declared for objects in <C>IsGeneralisedPolygon</C>. It is a mutable
attribute and can be accessed by the operation <F>IncidenceGraph</F>.
</Description>
</ManSection>

<ManSection>
<Oper Name="IncidenceGraph" Arg="gp"/>
<Returns>a graph</Returns>
<Description>
The argument <A>gp</A> is a generalised polygon. This operation returns the incidence graph of
<A>gp</A>. If <A>gp</A> is constructed using <F>GeneralisedPolygonByBlocks</F>, <F>GeneralisedPolygonByElements</F>
or <F>GeneralisedPolygonByIncidenceMatrix</F>, an incidence graph is computed to check the input, and is stored
as an attribute. For the particular generalised polygons available in <Package>FinInG</Package>, there is
no precomputed incidence graph. Note that computing an incidence graph may require some time, especially when
the <A>gp</A> has no collineation group computed. Therefore, this operation will return an error when <A>gp</A>
has no collineation group computed. As <F>CollineationGroup</F> is an attribute for objects in
<C>IsGeneralisedPolygon</C>, the user should compute the collineation group and then reissue the command to compute
the incidence graph.
We should also point out that this method returns a <E>mutable</E> attribute
of <A>gp</A>, so that acquired information about the incidence graph can be added. For example,
the automorphism group of the incidence graph may be computed and stored as a record component
after the incidence graph is stored as an attribute of <A>gp</A>. Normally, attributes of GAP objects are immutable.
Note that the factor 2 as difference in the order of the collineation group of <M>Q(4,4)</M> and the order of the
automorphism group of its incidence graph is easily explained by the fact that the <M>Q(4,4)</M> is self dual.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_incgraph.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="IncidenceMatrixOfGeneralisedPolygon" Arg="gp"/>
<Returns>a matrix</Returns>
<Description>
This method returns the incidence matrix of the generalised polygon via
the operation <C>CollapsedAdjacencyMat</C> in the <Package>GRAPE</Package>
package. The rows of the matrix
correspond to the points of <A>gp</A>, and the columns correspond to the lines.
Note that since this operation relies on <F>IncidenceGraph</F>,
for some generalised polygons, it is necessary to compute a collineation group first.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_incmatrix.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="CollineationGroup" Arg="gp"/>
<Returns>a group</Returns>
<Description>
This attribute returns the full collineation group of the generalised polygon <A>gp</A>.
For some particular generalised polygons, a (subgroup) of the full collineation group
can be computed efficiently without computing the incidence graph of <A>gp</A>: the full
collineation group of classical generalised quadrangles and classical generalised hexagons;
and an elation group with relation to a base-point of an elation generalised quadrangle. For
generalised polygons constructed by the operations <F>GeneralisedPolygonByBlocks</F>,
<F>GeneralisedPolygonByElements</F> or <F>GeneralisedPolygonByIncidenceMatrix</F>, the
full collineation group is computed using the full automorphism group of the underlying
incidence graph, the latter being computed by the package <Package>GRAPE</Package>.
The collineation groups computed for classical generalised quadrangles and classical
generalised hexagons are collineation groups in the sense of <Package>FinInG</Package>, and
come equipped with a NiceMonomorphism. The collineation groups computed in all other cases
are permutations groups, acting on the vertices of the underlying incidence graph.
Note that the computation of the automorphism group of the underlying graph can be time
consuming, also if the complete collineation group of the generalised polygon has been
used as an argument in e.g. <F>GeneralisedPolygonByElements</F>.
The first example illustrates that <F>CollineationGroup</F> is naturally applicable to
all classical generalised Polygons.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineationgroup1.include">
</Example>
The second example illustrates the computation of collineation groups of generalised polygons constructed
using different objects.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineationgroup2.include">
</Example>
In the third example, the use of an precomputed automorphism group is illustrated. It speeds up
the construction of the underlying graph and the computation of the automorphism group of the
underlying graph. However, as is also illustrated in the example, despite that the precomputed
automorphism group of the generalised polygon is actually the full collineation group, still some
time is needed to compute the automorphism group of the underlying graph. The timings after both
<F>CollineationGroup</F> commands are wrong. This is because <Package>GRAPE</Package> relies on
an external binary to computed the automorphism group of a graph. The generalised quadrangle in this example
is known as <Alt Only="HTML MathJax"><M>T_2^*(O)</M></Alt><Alt Not="HTML"><M>T_2^*(O)</M></Alt><Alt Only="HTML noMathJax">T2*(O)</Alt>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineationgroup3.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="CollineationAction" Arg="group"/>
<Returns>a function</Returns>
<Description>
<A>group</A> is a collineation group of a generalised polygon, computed using
<F>CollineationGroup</F>. The collineation group of classical generalised polygons
will be a collineation group in the sense of <Package>FinInG</Package>. The natural action
is <F>OnProjectiveSubspaces</F>. The collineation group of any other generalised polygons
will be a permutation group. The result of <F>CollineationAction</F> for such a group
is a function with input a pair <A>(x,g)</A> where <A>x</A> is an element of
the generalised polygon, and <A>g</A> is a collineation of the generalised polygon,
so an element of <A>group</A>. The example illustrates the use in the generalised quadrangle.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineations.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="BlockDesignOfGeneralisedPolygon" Arg="gp"/>
<Returns>a block design</Returns>
<Description>
This method allows one to use the GAP package <Package>DESIGN</Package>
to analyse a generalised polygon, so the user must first load this package.
The argument <A>gp</A> is a generalised polygon, and if it has a collineation group,
the block design is computed with this extra information and
thus the resulting design is easier to work with. Likewise, if <A>gp</A> is an elation
generalised quadrangle and it has an elation group, then we use the elation group's action
to efficiently compute the block design.

We should also point out that this method returns a <E>mutable</E> attribute
of <A>gp</A>, so that acquired information about the block design can be added. For example,
the automorphism group of the block design may be computed after the design is stored
as an attribute of <A>gp</A>. Normally, attributes of GAP objects are immutable.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_blockdesign.include">
</Example>
</Description>
</ManSection>

</Section>

<Section Label="gp:elements">
<Heading>Elements of generalised polygons</Heading>

<ManSection>
<Heading>Collections of elements of generalised polygons</Heading>
<Attr Name="ElementsOfIncidencStructure" Arg="gp, i"/>
<Attr Name="Points" Arg="gp"/>
<Attr Name="Lines" Arg="gp"/>
<Returns>a collection of elements of a generalised polygon</Returns>
<Description>
<A>gp</A> is any generalised polygon, <A>i</A> is a natural number, necessarily <M>1</M> or <M>2</M>.
<F>ElementsOfIncidenceStructure</F> returns the elements of type <M>i</M> of <A>gp</A>, <F>Points</F>
and <F>Lines</F> are the usual shortcuts.
</Description>
</ManSection>

<ManSection>
<Oper Name="Size" Arg="els"/>
<Returns>a number</Returns>
<Description>
<A>els</A> is a collection of elements of a generalised polygon. This operation returns
the number of element in <A>els</A>.
</Description>
</ManSection>

<ManSection>
<Heading>Creating elements from objects and retrieving objects from elements</Heading>
<Oper Name="ObjectToElement" Arg="gp, obj"/>
<Oper Name="ObjectToElement" Arg="gp, type, obj"/>
<Oper Name="UnderlyingObject" Arg="el"/>
<Returns>a collection of elements of a generalised polygon</Returns>
<Description>
To create elements in <A>gp</A> (of type <A>type</A>), one of the versions of
<F>ObjectToElement</F> can be used. It is checked whether <A>obj</A> represents an element
(of type <A>type</A>). To retrieve an underlying object of an element <A>el</A>, <F>UnderlyingObject</F>
can be used.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_objectselements.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Incidence</Heading>
<Oper Name="IsIncident" Arg="v,w"/>
<Oper Name="\*" Arg="v,w"/>
<Returns>true or false</Returns>
<Description>
Let <A>v</A> and <A>w</A> be two elements of a generalised polygon. It is checked if
the ambient geometry of the two elements are identical, and true is returned if and only if
the two elements are incident in their ambient geometry.
</Description>
</ManSection>

<ManSection>
<Oper Name="Span" Arg="v,w"/>
<Returns>a line of a generalised polygon or fail</Returns>
<Description>
Let <A>v</A> and <A>w</A> be two elements of a generalised polygon. It is checked if
the ambient geometries of the two elements are identical, and if the two elements are points.
If <A>v</A> and <A>w</A> are incidence with a common line, this line is returned. Otherwise
<K>fail</K> is returned. For generalised polygons constructed with <C>GeneralisedPolygonByBlocks</C>,
<C>GeneralisedPolygonByElements</C> an <C>GeneralisedPolygonByInidenceMatrix</C>, the underlying graph
is used. Note that the behaviour of <F>Span</F> is different for elements of
generalised polygons that belong to <C>IsLieGeometry</C>, see <Ref Label="liespan"/>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_span.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="Meet" Arg="v,w"/>
<Returns>a point of a generalised polygon or fail</Returns>
<Description>
Let <A>v</A> and <A>w</A> be two elements of a generalised polygon. It is checked if
the ambient geometries of the two elements are identical, and if the two elements are lines.
If <A>v</A> and <A>w</A> are incidence with a common point, this point is returned. Otherwise
<K>fail</K> is returned. For generalised polygons constructed with <C>GeneralisedPolygonByBlocks</C>,
<C>GeneralisedPolygonByElements</C> an <C>GeneralisedPolygonByInidenceMatrix</C>, the underlying graph
is used. Note that the behavior of <F>Meet</F> is different for elements of
generalised polygons that belong to <C>IsLieGeometry</C>, see <Ref Label="liemeet"/>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_meet.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Shadow elements</Heading>
<Oper Name="ShadowOfElement" Arg="geo, v, j"/>
<Oper Name="Points" Arg="el"/>
<Oper Name="Lines" Arg="el"/>
<Oper Name="ElementsIncidentWithElementOfIncidenceStructure" Arg="el, i"/>
<Returns> A collection of elements</Returns>
<Description>
<A>geo</A> is a generalised polygon, <A>v</A> must be an element of <A>geo</A>,
<A>j</A> is an integer equal to 1 or 2, since <A>geo</A> is a rank two geometry.
The operation <F>ShadowOfElement</F> returns the collection of elements of <A>geo</A> of type
<A>j</A>, incident with the element <A>v</A>. The operations <F>Points</F> and <F>Lines</F> with argument
are the usual shortcuts to <F>ShadowOfElement</F> with <A>j</A> respectively equal to 1, 2. The operation
<F>ElementsIncidentWithElementOfIncidenceStructure</F> is the usual shortcut to <F>ShadowOfElement</F>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_shadow.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="DistanceBetweenElements" Arg="v,w"/>
<Returns>a number</Returns>
<Description>
Let <A>v</A> and <A>w</A> be two elements of a generalised polygon. It is checked if
the ambient geometry of the two elements are identical, and the distance between the two
elements in the incidence graph of their ambient geometry is returned.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_distance.include">
</Example>
</Description>
</ManSection>

</Section>

<Section Label="CGH">
<Heading>The classical generalised hexagons</Heading>

<Subsection>
<Heading>Trialities of the hyperbolic quadric and generalised hexagons</Heading>

Consider the hyperbolic quadric &qplus7q;. This is a polar space of rank 4.
It is well known that its generators fall into two systems. Each system
contains exactly <Alt Only="HTML MathJax"><M>(q^3+q^2+q+1)(q^2+q+1)</M></Alt>
<Alt Not="HTML"><M>(q^3+q^2+q+1)(q^2+q+1)</M></Alt>
<Alt Only="HTML noMathJax">(q3+q2+q+1)(q2+q+1)</Alt>
generators, which is equal to the number of points of &qplus7q;. Generators from the
same system meet each other in an empty subspace or in a line, Generators from a different system
meet each other in a point or a plane. One defines the rank <M>4</M> geometry
<Alt Not="HTML"><M>\Omega(7,q)</M></Alt><Alt Only="HTML MathJax"><M>\Omega(7,q)</M></Alt>
<Alt Only="HTML noMathJax">Ω(7,q)</Alt> as follows. The 0-points are the points of &qplus7q;,
the 1-points are the generators of the first system, the 2-points are the generators of the second system,
and the lines are the lines of &qplus7q;. The incidence is the
natural incidence of the underlying projective space. Denote the set of <M>i</M>-points as
<Alt Not="HTML"><M>P^{(i)}</M></Alt>
<Alt Only="HTML MathJax"><M>P^{(i)}</M></Alt>
<Alt Only="HTML noMathJax">P(i)</Alt>, <M>i=0,1,2</M>.
A triality of <Alt Only="HTML MathJax"><M>\Omega(7,q)</M></Alt><Alt Not="HTML"><M>\Omega(7,q)</M></Alt>
<Alt Only="HTML noMathJax">Ω(7,q)</Alt> is a map <Alt Not="HTML"><M>\tau: P^{(i)} \to P^{(i+1)}</M></Alt>
<Alt Only="HTML MathJax"><M>\tau: P^{(i)} \to P^{(i+1)}</M></Alt>
<Alt Only="HTML noMathJax"><M>\tau</M>:P(i)→P(i+1)</Alt>
(where <M>i+1</M> is computed modulo 3) preserving the incidence and for which <Alt Not="HTML"><M>\tau^3 = 1</M></Alt>
<Alt Only="HTML MathJax"><M>\tau^3 = 1</M></Alt><Alt Only="HTML noMathJax"><M>\tau</M>3=1</Alt>.
Note that the image of a line under <M>\tau</M> is
determined by the image of the points incident with the line. An <M>i</M>-point is absolute with
respect to a fixed triality if it is incident with its image under the triality. Consequently, a line is absolute
with if it is fixed by the triality.
A generalised hexagon can be constructed as geometry of absolute points of one kind and absolute lines with relation to
a fixed triality. Note that not all trialities yield (thick) generalised quadrangles. There are different
types of trialities, for some of them the absolute geometry is degenerate.
The triality used in <Package>FinInG</Package> to construct the classical generalised hexagons is fixed. It is
described explicitly in <Cite Key="HVM"/>. To describe the triality, a trilinear form expressing the incidence
between <M>i</M>-points of <M>\Omega(7,q)</M> is used. Given the fact that, because of the existence of a triality,
the role of the 0, 1 and 2 points are the same, each of the 1 and 2 points can be labelled the same way as the 0-points,
which are effectively labelled by 8-tuples <M>(x_0,\ldots,x_7) \in V(8,q)=V</M> where each 8-tuples represents a projective point
of &qplus7q;.
Consider the hyperbolic quadric determined by the quadratic form
<Alt Not="HTML"><M>X_0X_4+X_1X_5+X_2X_6+X_3X_7</M></Alt>
<Alt Only="HTML MathJax"><M>X_0X_4+X_1X_5+X_2X_6+X_3X_7</M></Alt>
<Alt Only="HTML noMathJax">X0X4+X1X5+X2X6+X3X7</Alt>.
Consider the trilinear map <M>T</M>,
<Alt Only="HTML MathJax"><M>T(x,y,z) = \left|\begin{array}{ccc}
x_0 & x_1 & x_2 \\
y_0 & y_1 & y_2 \\
z_0 & z_1 & z_2 \\
\end{array}\right| +
\left|\begin{array}{ccc}
x_4 & x_5 & x_6 \\
y_4 & y_5 & y_6 \\
z_4 & z_5 & z_6 \\
\end{array}\right| +
x_3(z_0y_4+z_1y_5+z_2y_6) + x_7(y_0z_4+y_1z_5+y_2z_6) +
y_3(x_0z_4+x_1z_5+x_2z_6) + \\ y_7(z_0x_4+z_1x_5+z_2x_6) +
z_3(y_0x_4+y_1x_5+y_2x_6) + z_7(x_0y_4+x_1y_5+x_2y_6) -
x_3y_3z_3 - x_7y_7z_7.</M></Alt>
<Alt Only="LaTeX"><Display>T(x,y,z) = \left|\begin{array}{ccc}
 x_0 & x_1 & x_2 \\
 y_0 & y_1 & y_2 \\
 z_0 & z_1 & z_2 \\
 \end{array}\right| +
 \left|\begin{array}{ccc}
 x_4 & x_5 & x_6 \\
 y_4 & y_5 & y_6 \\
 z_4 & z_5 & z_6 \\
 \end{array}\right| +
 x_3(z_0y_4+z_1y_5+z_2y_6) + x_7(y_0z_4+y_1z_5+y_2z_6) +</Display></Alt>
<Alt Only="LaTeX"><Display>
 y_3(x_0z_4+x_1z_5+x_2z_6) + \\ y_7(z_0x_4+z_1x_5+z_2x_6) +
 z_3(y_0x_4+y_1x_5+y_2x_6) + z_7(x_0y_4+x_1y_5+x_2y_6) -
 x_3y_3z_3 - x_7y_7z_7.</Display></Alt><Alt Only="HTML noMathJax">
<#Include SYSTEM "./include/beautiful_formula_in_ugly_html.xml">
</Alt><Alt Only="Text">Display of this formula is not possible in text mode</Alt>

Now a pair <M>(x,y) \in V \times V</M> represents an incident 0-1 pair of points if and only if <M>T(x,y,z)</M> vanishes in the variable
<M>z</M>, and similarly for any cyclic permutation of the letters <M>x,y,z</M>. So given a <M>1</M>-point <M>y</M>, <M>T(x,y,z) = 0</M>,
where <M>z</M> is a variable, and <M>x</M> is an unknown, gives a set of equations representing a generator of &qplus7q;
this is the generator of &qplus7q; represented by <M>y</M> as label of a <M>1</M>-point.
Let <M>\sigma</M> be an automorphism of &gfq; of order <M>3</M>, or the identity. Consider the map
<Alt Only="HTML noMathJax"> τσ: P(i)→P(i+1)</Alt>
<Alt Only="HTML MathJax"><M>\tau_{\sigma}: P^{(i)} \to P^{(i+1)}</M></Alt>
<Alt Not="HTML"><M>\tau_{\sigma}: P^{(i)} \to P^{(i+1)}</M></Alt>
<Alt Not="HTML"><M>(x_j) \mapsto (x_j^{\sigma})</M>, <M>j=0\ldots 7</M></Alt>
<Alt Only="HTML MathJax"><M>(x_j) \mapsto (x_j^{\sigma})</M>, <M>j=0\ldots 7</M></Alt>
<Alt Only="HTML noMathJax"> (xj) → (xjσ), j=0...7</Alt>.
This map clearly preserves <M>T(x,y,z)</M>, so preserves the incidence, and has order three, so it is a triality of
<Alt Not="HTML"><M>\Omega(7,q)</M></Alt><Alt Only="HTML MathJax"><M>\Omega(7,q)</M></Alt><Alt Only="HTML noMathJax">Ω(7,q)</Alt>. We call an element <M>p</M> absolute with
respect to a triality <M>\tau</M> if and only if <Alt Only="HTML MathJax"><M>p \mathrm{I} p^\tau</M></Alt><Alt Not="HTML"><M>p \mathrm{I} p^\tau</M></Alt>
<Alt Only="HTML noMathJax">p I pτ</Alt>. Consequently, a line is absolute if and only if it
is fixed by the triality. Denote the set of <M>i</M>-points that are absolute with respect to the triality as
<Alt Only="HTML MathJax"><M>P_{\mathrm{abs}}^{(i)}</M></Alt><Alt Not="HTML"><M>P_{\mathrm{abs}}^{(i)}</M></Alt>
<Alt Only="HTML noMathJax">P(i)abs</Alt>, and the
set of absolute lines with respect to the triality as <Alt Not="HTML"><M>L_{\mathrm{abs}}</M></Alt>
<Alt Only="HTML MathJax"><M>L_{\mathrm{abs}}</M></Alt><Alt Only="HTML noMathJax">Labs</Alt>.
Then a famous theorem of Tits (<Cite Key="Tits1959"/>) says that for the triality
<Alt Only="HTML noMathJax">τσ</Alt><Alt Only="HTML MathJax"><M>\tau_{\sigma}</M></Alt>
<Alt Not="HTML"><M>\tau_{\sigma}</M></Alt>, the point-line geometry
<Alt Not="HTML"><M>\Gamma^{(i)} = (P^{(i)}_{\mathrm{abs}},L_{\mathrm{abs}},\mathrm{I})</M></Alt>
<Alt Only="HTML MathJax"><M>\Gamma^{(i)} = (P^{(i)}_{\mathrm{abs}},L_{\mathrm{abs}},\mathrm{I})</M></Alt>
<Alt Only="HTML noMathJax">Γ(i) = (P(i)abs, Labs, I)</Alt>
is a generalised hexagon of order <M>(|K|,|L|)</M>, <M>K</M> the field &gfq; and <M>L</M> the subfield of invariant elements of <M>K</M>
under the field automorphism <M>\sigma</M>. Note that a finite field has a field automorphism of order three if and only if its
order equals <Alt Only="HTML MathJax"><M>q^3</M></Alt><Alt Not="HTML"><M>q^3</M></Alt><Alt Only="HTML noMathJax">q3</Alt>.
So, for <M>K</M> equal to &gfq; and <M>\sigma=1</M>, <Alt Only="HTML noMathJax">Γ(i)</Alt>
<Alt Only="HTML MathJax"><M>\Gamma^{(i)}</M></Alt><Alt Not="HTML"><M>\Gamma^{(i)}</M></Alt> is a generalised hexagon
of order <M>q</M>, which is called the <E>split Cayley hexagon of order <M>q</M></E>, denoted <M>H(q)</M>.
For <M>K</M> equal to &gfq3;, and
<M>\sigma</M> a non-trivial field automorphism of order <M>3</M>, <Alt Only="HTML noMathJax">Γ(i)</Alt>
<Alt Only="HTML MathJax"><M>\Gamma^{(i)}</M></Alt><Alt Not="HTML"><M>\Gamma^{(i)}</M></Alt> is a generalised hexagon
of order <Alt Only="HTML MathJax"><M>(q^3,q)</M></Alt><Alt Not="HTML"><M>(q^3,q)</M></Alt>
<Alt Only="HTML noMathJax">(q3,q)</Alt>, which is called the <E>twisted triality hexagon
of order <Alt Only="HTML MathJax"><M>(q^3,q)</M></Alt><Alt Not="HTML"><M>(q^3,q)</M></Alt>
<Alt Only="HTML noMathJax">(q3,q)</Alt></E>, denoted <Alt Only="HTML MathJax"><M>T(q^3,q)</M></Alt><Alt Not="HTML"><M>T(q^3,q)</M></Alt>
<Alt Only="HTML noMathJax">T(q3,q)</Alt>. Note that for a given triality, the hexagons
<Alt Only="HTML noMathJax">Γ(i)</Alt><Alt Not="HTML"><M>\Gamma^{(i)}</M></Alt>
<Alt Only="HTML MathJax"><M>\Gamma^{(i)}</M></Alt>, <M>i=0,1,2</M> are isomorphic. Consequently, <Alt Only="HTML noMathJax">Γ(0)</Alt>
<Alt Only="HTML MathJax"><M>\Gamma^{(0)}</M></Alt><Alt Not="HTML"><M>\Gamma^{(0)}</M></Alt>
is a point-line geometry of which the point set, line set respectively, is a subset
of the points, lines respectively of &qplus7q;. Finally, we mention the following important theorem, which was shown by Tits (<Cite Key="Tits1959"/>):
the split Cayley hexagon, obtained by the triality with <M>\sigma=1</M>, is contained in the hyperplane with equation
<Alt Not="HTML"><M>X_3+X_7=0</M></Alt><Alt Only="HTML MathJax"><M>X_3+X_7=0</M></Alt>
<Alt Only="HTML noMathJax">X3 + X7</Alt>, which intersects
the hyperbolic quadric in the parabolic quadric &q6q;.
The points of the split Cayley hexagon are the points of &q6q;.
This above description of the triality and the associated generalised hexagons, contains sufficient analytical information
to implement the split Cayley hexagon and the twisted triality hexagon in an efficient way. The user is allowed to choose
a representation for the ambient polar space. For <Alt Not="HTML"><M>q=2^h</M></Alt><Alt Only="HTML MathJax"><M>q=2^h</M></Alt>
<Alt Only="HTML noMathJax">q=2h</Alt>
the polar spaces &q6q; and &w5q; are isomorphic.
Consequently, the user may choose &w5q; as ambient polar space for the split Cayley hexagon of even order.
This embedding in &w5q; is called the perfect symplectic embedding of
the split Cayley hexagon. Finally, <Cite Key="HVM"/> contains an explicit description of the generators of the collineation groups
of both generalised hexagons.

</Subsection>

<ManSection>
<Filt Name="IsLieGeometry" Type="Category"/>
<Description>
Recall that the classical generalised hexagons are constructed as an object in <F>IsLieGeometry</F>. This makes most operations
described in the appropriate chapters on Lie geometries, projective spaces and polar spaces applicable.
</Description>
</ManSection>

<ManSection>
<Oper Name="SplitCayleyHexagon" Arg="q"/>
<Oper Name="SplitCayleyHexagon" Arg="f"/>
<Oper Name="SplitCayleyHexagon" Arg="ps"/>
<Returns>a generalised hexagon</Returns>
<Description>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_SplitCayleyHexagon.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="TwistedTrialityHexagon" Arg="q"/>
<Oper Name="TwistedTrialityHexagon" Arg="f"/>
<Oper Name="TwistedTrialityHexagon" Arg="ps"/>
<Returns>a generalised hexagon</Returns>
<Description>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_TwistedTrialityHexagon.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="vector spaceToElement" Arg="gh, vec"/>
<Returns>an element of a classical generalized hexagon</Returns>
<Description>
The argument <A>vec</A> is one vector or a list of vectors from the underlying vector space of <A>gh</A>.
This operation checks whether <A>vec</A> represents a point or a line of <A>gh</A>. Note that
vectors and matrices in different representations are allowed as argument.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_cghvectorspacetoelement.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="ObjectToElement" Arg="gh, obj"/>
<Returns>an element of a classical generalized hexagon</Returns>
<Description>
The argument <A>obj</A> is one vector or a list of vectors from the underlying vector space of <A>gh</A>.
This operation checks whether <A>obj</A> represents a point or a line of <A>gh</A>. Note that
vectors and matrices in different representations are allowed as argument.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_cghobjecttoelement.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="UnderlyingObject" Arg="gh, obj"/>
<Returns>a vector or a matrix</Returns>
<Description>
</Description>
</ManSection>

<ManSection>
<Oper Name="\in" Arg="x, gh"/>
<Returns>true or false</Returns>
<Description>
<Example>
<#Include SYSTEM "../examples/include/gpolygons_cghin.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Span and meet of elements</Heading>
<Oper Name="Span" Arg="x, y"/>
<Oper Name="Meet" Arg="x, y"/>
<Returns>a subspace of a projective space</Returns>
<Description>
<A>x</A> and <A>y</A> are two elements of a classical generalised hexagon. The operation <F>Span</F> returns the projective
line spanned by <A>x</A> and <A>y</A>. The operation <F>Meet</F> returns the intersection of the elements <A>x</A> and <A>y</A>.
Note that the classical generalised hexagons are Lie geometries, so their elements belong to a subcategory of <C>IsSubspaceOfProjectiveSpace</C>. Therefore,
the operations <F>Span</F> and <F>Meet</F> behave as described in <Ref Sect="polar:span"/> and <Ref Sect="polar:meet"/>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_spanmeet.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="CollineationGroup" Arg="gh"/>
<Returns>a group of collineations</Returns>
<Description>
<A>gh</A> is a classical generalised hexagon. This attribute returns the full collineation group, equipped with a nice monomorphism.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineationgroup4.include">
</Example>
</Description>
</ManSection>

</Section>

<Section Label="EGQ">
<Heading>Elation generalised quadrangles</Heading>

<Subsection>
<Heading>Elation generalised quadrangles and Kantor families</Heading>

Suppose <M>S=(P,B,I)</M> is a generalised quadrangle of order <M>(s,t)</M> for which
there exists a point <M>p</M> and a group of collineations <M>G</M> fixing
<M>p</M> and each line through <M>p</M>, with the extra property that
<M>G</M> acts regularly on the points not collinear with <M>p</M>. Then
<M>S</M> is called an <E>elation generalised quadrangle</E> with base-point <M>p</M>
and elation group <M>G</M>, and <M>G</M> has order <Alt Only="HTML MathJax"><M>s^2t</M></Alt><Alt Not="HTML"><M>s^2t</M></Alt><Alt Only="HTML noMathJax">s2t</Alt>.
Let <M>y</M> be a fixed point of <M>S</M>, not collinear with <M>p</M>. Denote the <M>t+1</M>
lines incident with <M>p</M> as <Alt Only="HTML MathJax"><M>L_i, i=0 \ldots t</M></Alt><Alt Not="HTML"><M>L_i, i=0 \ldots t</M></Alt>
<Alt Only="HTML noMathJax">Li, i=0 ... t</Alt>. Define for each line
<Alt Only="HTML MathJax"><M>L_i</M></Alt><Alt Only="HTML noMathJax">Li</Alt><Alt Not="HTML"><M>L_i</M></Alt> the unique
point-line pair <Alt Only="HTML MathJax"><M>(z_i,M_i)</M></Alt><Alt Only="HTML noMathJax">(zi,Mi)</Alt>
<Alt Not="HTML"><M>(z_i,M_i)</M></Alt> such that <Alt Only="HTML MathJax"><M>L_i I z_i I M_i I y</M></Alt>
<Alt Not="HTML"><M>L_i I z_i I M_i I y</M></Alt><Alt Only="HTML noMathJax">Li I zi I Mi I y</Alt>.
Define the groups <Alt Only="HTML MathJax"><M>S_i</M></Alt><Alt Not="HTML"><M>S_i</M></Alt>
<Alt Only="HTML noMathJax">Si</Alt> as the subgroups of <M>G</M> fixing the lines
<Alt Only="HTML MathJax"><M>M_i</M></Alt><Alt Not="HTML"><M>M_i</M></Alt><Alt Only="HTML noMathJax">Mi</Alt>, and define the groups
<Alt Only="HTML MathJax"><M>S_i^*</M></Alt><Alt Not="HTML"><M>S_i^*</M></Alt>
<Alt Only="HTML noMathJax">Si*</Alt> as the subgroups of
<M>G</M> fixing the point <Alt Only="HTML MathJax"><M>z_i</M></Alt><Alt Not="HTML"><M>z_i</M></Alt>
<Alt Only="HTML noMathJax">zi</Alt>.
Define the set <Alt Only="HTML MathJax"><M>J = \{S_i: i=0 \ldots t\}</M></Alt><Alt Not="HTML"><M>J = \{S_i: i=0 \ldots t\}</M></Alt>
<Alt Only="HTML noMathJax">J = {Si: i=0 ... t}</Alt>,
and the set <Alt Only="HTML MathJax"><M>J^* = \{S_i: i=0 \ldots t\}</M></Alt><Alt Not="HTML"><M>J^* = \{S_i: i=0 \ldots t\}</M></Alt>
<Alt Only="HTML noMathJax">J* = {Si*: i=0 ... t}</Alt>. Since <M>S</M> is an elation generalised quadrangle,
<M>J</M> is a collection of <M>t+1</M> subgroups of <M>G</M> of order <M>s</M>, and each <Alt Only="HTML MathJax"><M>S_i^*</M></Alt>
<Alt Not="HTML"><M>S_i^*</M></Alt><Alt Only="HTML noMathJax">Si*</Alt>
contains has order <M>st</M> and contains <Alt Only="HTML MathJax"><M>S_i</M></Alt><Alt Not="HTML"><M>S_i</M></Alt>
<Alt Only="HTML noMathJax">Si</Alt> as a subgroup. Furthermore, the following two conditions are satisfied.
(K1) <Alt Only="HTML MathJax"><M>S_iS_j \cap S_k = \{1\}</M></Alt><Alt Not="HTML"><M>S_iS_j \cap S_k = \{1\}</M></Alt>
<Alt Only="HTML noMathJax">Si Sj <M>\cap</M>Sk = {1}</Alt>, for distinct <M>i,j,k</M>.
(K2) <Alt Only="HTML MathJax"><M>S_i \cap S_j^* = \{1\}</M></Alt><Alt Not="HTML"><M>S_i \cap S_j^* = \{1\}</M></Alt>
<Alt Only="HTML noMathJax">Si* <M>\cap</M>Sj = {1}</Alt>, for distinct <M>i,j</M>.
The pair <Alt Only="HTML MathJax"><M>(J,J^*)</M></Alt><Alt Not="HTML"><M>(J,J^*)</M></Alt>
<Alt Only="HTML noMathJax">(J,J*)</Alt> is called a <E>4-gonal family</E> or
<E>Kantor family</E> in <M>G</M>.
Remarkably, each Kantor family in a group of order <Alt Only="HTML MathJax"><M>s^2t</M></Alt><Alt Not="HTML"><M>s^2t</M></Alt>
<Alt Only="HTML noMathJax">s2t</Alt>
gives rise to an elation generalised quadrangle. Kantor families and elation generalised quadrangles are equivalent objects,
and one of the motivations to study Kantor families in groups was to find examples of non-classical elation generalised quadrangles.
Given a group <M>G</M>, together with a Kantor family <Alt Only="HTML MathJax"><M>(J,J^*)</M></Alt><Alt Not="HTML"><M>(J,J^*)</M></Alt>
<Alt Only="HTML noMathJax">(J,J*)</Alt>,
a generalised quadrangle is defined as follows.
The points are of three types:
(i) points of type 1 are the elements of <M>G</M>;
(ii) points of of type 2 are the right cosets <Alt Only="HTML MathJax"><M>S^*g, s^* \in J^*</M></Alt>
<Alt Not="HTML"><M>S^*g, s^* \in J^*</M></Alt><Alt Only="HTML noMathJax">S*g, S*<M>\in</M> J*</Alt>
(iii) the unique point of type (iii) is the symbol <Alt Only="HTML MathJax"><M>(\infty)</M></Alt><Alt Not="HTML"><M>(\infty)</M></Alt>
<Alt Only="HTML noMathJax">(∞)</Alt>.
The lines are of two types:
(a) lines of type (a) are the right cosets <M>Sg</M>, <M>S \in J</M>;
(b) Lines of type (b) are the symbols <M>[S], S \in J</M>.
Incidence is defined as follows. A point <M>g</M> of type (i) is incident with each
line <M>Sg, S \in J</M> of type (a). A point of type (ii) <Alt Only="HTML MathJax"><M>S^*g</M></Alt>
<Alt Not="HTML"><M>S^*g</M></Alt>
<Alt Only="HTML noMathJax">S*g</Alt> is incident the line <M>[S]</M> of type (b)
and the <M>t</M> lines of type (a) for which <Alt Only="HTML MathJax"><M>Sh \subset S^*g</M></Alt>
<Alt Not="HTML"><M>Sh \subset S^*g</M></Alt>
<Alt Only="HTML noMathJax">Sh <M>\subset</M> S*g</Alt>. Finally, the unique point of type (iii)
is incident with the lines of type (b), and there are no further incidences.
It is shown, see e.g. the standard work in this field of Payne and Thas <Cite Key="FGQ"/>, that this
point-line geometry is a generalised quadrangle of order <M>(s,t)</M>.
<Package>FinInG</Package> provides functions to construct elation generalised quadrangles
directly from a Kantor family. The constructed generalised quadrangles are generalised polygons in the
sense of <Package>FinInG</Package>, i.e. all generic operations described in Sections <Ref Sect="gp:operations"/> and
<Ref Sect="gp:elements"/>.

</Subsection>

<ManSection>
<Heading>Categories</Heading>
<Filt Name="IsEGQByKantorFamily" Type="Category"/>
<Filt Name="IsElementOfKantorFamily" Type="Category"/>
<Description>
<C>IsEGQByKantorFamily</C> is a subcategory of <C>IsElationGQ</C>. It contains all elations generalised quadrangles
that are constructed from a Kantor family. <C>IsElementOfKantorFamily</C> is a subcategory of <C>IsElementOfGeneralisedPolygon</C>.
It contains the elements from generalised quadrangles in the category <C>IsEGQByKantorFamily</C>.
</Description>
</ManSection>

<ManSection>
<Heading>Kantor families</Heading>
<Oper Name="IsKantorFamily" Arg="g,f,fstar"/>
<Returns>true or false</Returns>
<Description>
There is no specific way to construct a Kantor family in <Package>FinInG</Package>. However, given a group <M>G</M> and
two collections of subgroups, <F>IsKantorFamily</F> will check whether the input satisfies the conditions of a Kantor family.
If so, the input can be used directly for the operation <F>EGQByKantorFamily</F>.
</Description>
</ManSection>

<ManSection>
<Oper Name="EGQByKantorFamily" Arg="g,f,fstar"/>
<Returns>a generalised quadrangle</Returns>
<Description>
Let <A>g</A> be a group and <A>f</A> and <A>fstar</A> two collections of subgroups, satisfying the conditions of a Kantor
family. This operation returns the corresponding elation generalised quadrangle. Note that this operation <E>does not</E> check
if the input satisfies the conditions to be a Kantor family, it only checks whether the group <A>f[i]</A> is a subgroup of the group
<A>fstar[i]</A>. In the example below, the use of <F>IsKantorFamily</F> is also
demonstrated, and some categories are displayed.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_EGQByKantorFamily.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Representation of elements and underlying objects</Heading>
<Oper Name="ObjectToElement" Arg="egq, t, obj"/>
<Oper Name="ObjectToElement" Arg="egq, obj"/>
<Oper Name="BasePointOfEGQ" Arg="egq"/>
<Oper Name="UnderlyingObject" Arg="el"/>
<Description>
For technical reasons, the underlying objects of the elements of an elation generalised quadrangle constructed from a Kantor family,
are not exactly the mathematical objects from the definition. However, these technicalities are almost completely hidden for the user, except
for the representation of lines of type (b), which are represented in <Package>FinInG</Package> by the elements of the collection
<Alt Only="HTML MathJax"><M>J^*</M></Alt><Alt Not="HTML"><M>J^*</M></Alt><Alt Only="HTML noMathJax">J*</Alt> (instead of the elements of the collection <M>J</M>). This change
from the original definition has no mathematical implications, since there is a bijective correspondence between the elements of
<Alt Only="HTML MathJax"><M>J^*</M></Alt><Alt Not="HTML"><M>J^*</M></Alt><Alt Only="HTML noMathJax">J*</Alt> and <M>J</M>. Notice also that it is only possible to obtain
the base-point of an elation generalised quadrangle constructed from a Kantor family through calling the attribute <F>BasePointOfEGQ</F>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_objectselementsofegq.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Elation group and natural action on elements</Heading>
<Attr Name="ElationGroup" Arg="egq"/>
<Func Name="OnKantorFamily" Arg="g,el"/>
<Attr Name="CollineationAction" Arg="g"/>
<Description>
The attribute <F>ElationGroup</F> is naturally set upon creation of an elation generalised quadrangle from a Kantor
family. The elements of the elation group act "naturally" on the elements of the elation generalised quadrangle. This
natural action is implemented in the action function <F>OnKantorFamily</F>. When <A>g</A> is the elation group of an elation
generalised quadrangle constructed from a Kantor family, the attribute <F>CollineationAction</F> will return the action function
<F>OnKantorFamily</F>. The action function makes use of generic GAP operations possible, as is demonstrated in the example.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_elationgroup.include">
</Example>
</Description>
</ManSection>

<Subsection>
<Heading>Kantor families, q-clans, and elation generalised quadrangles</Heading>

Let <Alt Only="HTML MathJax"><M>C=\{A_i: i=1 \ldots q\}</M></Alt><Alt Not="HTML"><M>C=\{A_i: i=1 \ldots q\}</M></Alt><Alt Only="HTML noMathJax">C = {Ai: i=1 ... q}</Alt>
be a set of <M>q</M> distinct upper triangle <M>2 \times 2</M>
matrices over the finite field &gfq;. Then <M>C</M> is called a <E>q-clan</E> if
<Alt Only="HTML MathJax"><M>A_r - A_t</M></Alt><Alt Not="HTML"><M>A_r - A_t</M></Alt><Alt Only="HTML noMathJax">Ar - At</Alt>
is anisotropic for <M>r \neq t</M>. Define
<Alt Only="HTML MathJax"><M>G = \{(\alpha,c,\beta): \alpha,\beta \in GF(q)^2, c \in GF(q)\}</M></Alt>
<Alt Not="HTML"><M>G = \{(\alpha,c,\beta): \alpha,\beta \in GF(q)^2, c \in GF(q)\}</M></Alt>
<Alt Only="HTML noMathJax">G = {(α,c,β): α,β ∈ GF(q)2, c ∈ GF(q) }</Alt>,
and define the binary operator <M>\times</M> as
<Alt Only="HTML MathJax"><M>(\alpha,c,\beta) \times (\alpha',c',\beta') =
(\alpha + \alpha',c+c'+\beta \cdot \alpha'^{T},\beta + \beta')</M></Alt>
<Alt Not="HTML"><M>(\alpha,c,\beta) \times (\alpha',c',\beta') =
 (\alpha + \alpha',c+c'+\beta \cdot \alpha'^{T},\beta + \beta')</M></Alt>
<Alt Only="HTML noMathJax">(α,c,β) <M>\times</M>
(α',c',β') = (α+α',c+c'+βα'T,β+β'). Then G,\times
is a group with center <M>Z(G)=\{(0,c,0): c \in GF(q)\}</M>. Consider a q-clan
<Alt Only="HTML MathJax"><M>C=\{A_i: i=1 \ldots q\}</M></Alt><Alt Not="HTML"><M>C=\{A_i: i=1 \ldots q\}</M></Alt><Alt Only="HTML noMathJax">C = {Ai: i=1 ... q}</Alt>, define
<Alt Only="HTML MathJax"><M>K_i = A_i+A_i^T</M></Alt><Alt Not="HTML"><M>K_i = A_i+A_i^T</M></Alt><Alt Only="HTML noMathJax">Ki = Ai +
AiT</Alt>. Now define the following subgroups of <M>G</M>
<Alt Not="HTML"><M>A(i) = \{(\alpha,\alpha A_t,\alpha K_t): \alpha \in GF(q)^2, i=1\ldots q\}</M></Alt>
<Alt Only="HTML MathJax"><M>A(i) = \{(\alpha,\alpha A_t,\alpha K_t): \alpha \in GF(q)^2, i=1\ldots q\}</M></Alt>
<Alt Only="HTML noMathJax">A(i)= {(α,αAi,αKi): α ∈ GF(q)2, i=1..q }</Alt> and
<Alt Only="HTML MathJax"><M>A(\infty) = \{(0,0,\gamma): \gamma \in GF(q)^2\}</M></Alt><Alt Not="HTML"><M>A(\infty) = (0,0,\gamma): \gamma \in GF(q)^2\}</M></Alt>
<Alt Only="HTML noMathJax">A(∞)= {(0,0,γ) γ ∈ GF(q)2 }</Alt>, and
<Alt Only="HTML MathJax"><M>A^*(i) = \{(\alpha,b,\alpha K_t): \alpha \in GF(q)^2, b \in GF(q), i=1\ldots q\}</M></Alt>
<Alt Not="HTML"><M>A^*(i) = \{(\alpha,b,\alpha K_t): \alpha \in GF(q)^2, b \in GF(q), i=1\ldots q\}</M></Alt>
<Alt Only="HTML noMathJax">A*(i)= {(α,b,αKi): α ∈ GF(q)2, b ∈ GF(q), i=1..q }</Alt> and
<Alt Only="HTML MathJax"><M>A^*(\infty) = \{(0,b,\gamma): \gamma \in GF(q)^2, b \in GF(q)\}</M></Alt>
<Alt Not="HTML"><M>A^*(\infty) = \{(0,b,\gamma): \gamma \in GF(q)^2, b \in GF(q)\}</M></Alt>
<Alt Only="HTML noMathJax">A*(∞) = {(0,b,γ): γ ∈ GF(q)2, b ∈ GF(q)}</Alt>
Define <Alt Only="HTML MathJax"><M>J=\{A(i):i=1\ldots q\} \cup \{A(\infty)\}</M></Alt>
<Alt Not="HTML"><M>J=\{A(i):i=1\ldots q\} \cup \{A(\infty)\}</M></Alt>
<Alt Only="HTML noMathJax">J = {A(i): i= 1 ... q} ∪ {A(∞)}</Alt> and
<Alt Only="HTML MathJax"><M>J^*=\{A^*(i):i=1\ldots q\} \cup \{A^*(\infty)\}</M></Alt>
<Alt Not="HTML"><M>J^*=\{A^*(i):i=1\ldots q\} \cup \{A^*(\infty)\}</M></Alt>
<Alt Only="HTML noMathJax">J* = {A*(i): i= 1 ... q} ∪ {A*(∞)}</Alt>
A combination of results of Payne and Kantor yield the famous theorem that <Alt Only="HTML MathJax">(J,J^*)</Alt><Alt Not="HTML">(J,J^*)</Alt>
<Alt Only="HTML noMathJax">(J,J*)</Alt> is a Kantor family in <M>G</M> if and only if <M>C</M> is a q-clan.
<Package>FinInG</Package> provides functionality to construct q-clans and to construct the corresponding Kantor family.
As such, elation generalised quadrangles can directly constructed from q-clans.
</Subsection>

<ManSection>
<Oper Name="qClan" Arg="list, f"/>
<Returns>a q-clan.</Returns>
<Description>
Given a list <A>list</A> of <M>2 \times</M> matrices over the field <A>f</A>,
it is checked if the matrices in the list satisfy the condition to constitute a q-clan
over <M>f</M>. If so, the q-clan is returned.
</Description>
</ManSection>

<ManSection>
<Heading>Particular q-clans</Heading>
<Oper Name="LinearqClan" Arg="q"/>
<Oper Name="FisherThasWalkerKantorBettenqClan" Arg="q"/>
<Oper Name="KantorMonomialqClan" Arg="q"/>
<Oper Name="KantorKnuthqClan" Arg="q"/>
<Oper Name="FisherqClan" Arg="q"/>
<Returns>
a q-clan
</Returns>
<Description>
Some famous q-clans are built in. We refer to ... for more information on these.
</Description>
</ManSection>

<ManSection>
<Oper Name="KantorFamilyByqClan" Arg="clan"/>
<Returns>
A Kantor family.
</Returns>
<Description>
The Kantor family constructed from a q-clan will be a matrix group together with the
corresponding collections <M>J</M> and <Alt Only="HTML MathJax">J^*</Alt><Alt Not="HTML">J^*</Alt>
<Alt Only="HTML noMathJax">J*</Alt>.
</Description>
</ManSection>

<ManSection>
<Oper Name="EGQByqClan" Arg="clan"/>
<Returns>
An elation generalised quadrangle constructed from a q-clan.
</Returns>
<Description>
Given a q-clan <A>clan</A>, the operation <F>KantorFamilyByqClan</F> will be used to construct the Kantor family
from <A>clan</A>, followed by the construction of the elation generalised quadrangle using the operation
<F>EGQByKantorFamily</F>. The first example shows also the use of <F>qClan</F>, and shows that a linear q-clan
yields a classical generalised quadrangle.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_EGQByqClan.include">
</Example>
</Description>
</ManSection>

<Subsection>
<Heading>BLT-sets, flocks, q-clans, and elation generalised quadrangles</Heading>

A <E>flock</E> is a partition of the points of a quadratic cone in &pg3q;
minus its vertex into conics. Each conic is determined by a plane, and each plane is determined
uniquely by a triple of elements of the field &gfq;. So a flock is determined by <M>q</M> such
triples. Remarkably, as was shown by J.A. Thas, the conditions for these triples to constitute a flock,
are exactly the same conditions for these triples to constitute <M>q</M> upper triangle matrices making
a q-clan over &gfq;. Hence, q-clans and flocks, and thus flocks and elation generalised quadrangles,
are equivalent objects.
The quadratic cone can be embedded as a hyperplane section into the parabolic quadric &q4q;. L. Bader,
G. Lunardon and J.A. Thas observed that a set of <M>q</M> points of &q4q; can be constructed from
the <M>q</M> planes determining the flock, and this set of points satisfies certain geometric conditions. Such
a set is called a <E>BLT-set</E>. Dually, a BLT-set corresponds to a set of lines of &w3q;. Furthermore,
from this BLT-set of lines, it is possible to construct directly an elation generalised quadrangle from carefully
selecting points and lines from the symplectic space &w5q;. This construction is called the Knarr construction.
Consider the symplectic polar space &w5q; and choose a point <M>P \in W(5,q)</M>. Its tangent space is
a <M>4</M>-dimensional space <M>F</M>. Embed &w3q; in a solid of <M>F</M> not on the point <M>p</M>, and
let <M>L</M> be the set of BLT lines. Each line spans with <M>p</M> a plane of &w5q;, call these <M>q</M> planes
the <E>BLT-planes</E>. Now we can define the points an lines of the elation generalised quadrangle as follows.
Points of type (i) are the points of <M>W(5,q) \setminus F</M>, points of type (ii) are the lines of each BLT-plane
not on <M>p</M> and the unique point of type (iii) is the point <M>p</M>. Lines of type (a) are the
planes of &w5q; meeting a BLT-plane in a line. Note that no plane of &w5q; meeting two BLT planes in a
line can exist. Lines of type (b) are the BLT-planes. Incidence is the natural incidence (so the incidence inherited)
from the polar space &w5q;, and this geometry is a elation generalised quadrangle with base-point <M>p</M> and
of order <Alt Only="HTML noMathJax">(q2,q)</Alt><Alt Only="HTML MathJax"><M>(q^2,q)</M></Alt>
<Alt Not="HTML"><M>(q^2,q)</M></Alt>.
<Package>FinInG</Package> provides functions to construct elation generalised quadrangles using this model from
BLT-sets, and provides a function to compute a BLT set from a q-clan directly. The advantage of constructing
a elation generalised from elements of Lie geometries is the availability of the underlying projective groups and
their action on elements of Lie geometries.
</Subsection>

<ManSection>
<Filt Name="IsEGQByBLTSet" Type="Category"/>
<Description>
<C>IsEGQByBLTSet</C> is a subcategory of <C>IsElationGQ</C>. It contains all elations generalised quadrangles
that are constructed from a BLT set.
</Description>
</ManSection>

<ManSection>
<Oper Name="BLTSetByqClan" Arg="clan"/>
<Returns>
A BLT-set.
</Returns>
<Description>
The BLT-set is a set of points of the parabolic quadric in <M>PG(4,q)</M> with particular
equation <Alt Only="HTML MathJax"><M>2X_1x_5+2X_2X_4+w^{(q+1)/2}=0</M></Alt><Alt Not="HTML"><M>2X_1x_5+2X_2X_4+w^{(q+1)/2}=0</M></Alt>
<Alt Only="HTML noMathJax">X1X5+
X2X4+w(q+1)/2X32=0</Alt>, where <M>w</M>
is a primitive element of the underlying field of <A>clan</A>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_bltsetbyqclan.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Oper Name="EGQByBLTSet" Arg="blt"/>
<Returns>
An elation generalised quadrangle.
</Returns>
<Description>
<A>blt</A> is a BLT-set, this operation returns an elation generalised quadrangle constructed
as described above consisting of elements of &w5q;. Notice in the example that computing the
full collineation group of the GQ constructed directly from the q-clan (hence a group coset geometry) is
substantially slower than computing the full collineation group of the GQ constructed from the BLT-set.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_EGQByBLTSet.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="DefiningPlanesOfEGQByBLTSet" Arg="egq"/>
<Description>
For an elation generalised quadrangle in the category <C>IsEGQByBLTSet</C> (constructed from a BLT-set),
the planes of the polar space &w5q;, as described in the introduction, determine the generalised quadrangle
completely. This attribute returns these <M>q</M> planes of &w5q;.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_definingplanes.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Heading>Representation of elements and underlying objects</Heading>
<Oper Name="ObjectToElement" Arg="egq, t, obj"/>
<Oper Name="ObjectToElement" Arg="egq, obj"/>
<Oper Name="UnderlyingObject" Arg="el"/>
<Description>
The underlying objects of the elements of an elation generalised quadrangle in the category <C>IsEGQByBLTSet</C>
are elements of the polar space &w5q; in its standard representation. These elements can be used as underlying
object to construct elements of <A>egq</A>. The example also demonstrates the use of <F>DistanceBetweenElements</F>.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_objectselementsblt.include">
</Example>
</Description>
</ManSection>

<ManSection>
<Attr Name="CollineationSubgroup" Arg="egq"/>
<Description>
For an elation generalised quadrangle in the category <C>IsEGQByBLTSet</C> (constructed from a BLT set),
the planes of the polar space &w5q;, as described in the introduction, determine the generalised quadrangle
completely. The setwise stabiliser of these planes in the collineation group of &w5q; is a subgroup of the
completely collineation group of the elation generalised quadrangle, and can be computed much faster than the
complete collineation group. This attribute returns this setwise stabiliser. The returned group is equipped
with the <F>CollineationACtion</F> attribute. If <F>CollineationSubgroup</F> is computed, this group will be used
instead of the elation group to compute the incidence graph.
<Example>
<#Include SYSTEM "../examples/include/gpolygons_collineationsubgroup.include">
</Example>
</Description>
</ManSection>

</Section>

</Chapter>

quality100%

¤ Dauer der Verarbeitung: 0.65 Sekunden (vorverarbeitet) ¤

Wurzel

Suchen

Beweissystem der NASA

Beweissystem Isabelle

NIST Cobol Testsuite

Cephes Mathematical Library

Wiener Entwicklungsmethode

Haftungshinweis

Die Informationen auf dieser Webseite wurden nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit, noch Qualität der bereit gestellten Informationen zugesichert.

Bemerkung:

Die farbliche Syntaxdarstellung ist noch experimentell.