<html><head><title>[SONATA-tutorial] 8 Designs</title></head>
<body text="#000000" bgcolor="#ffffff">
[<a href = "chapters.htm">Up</a>] [<a href ="CHAP007.htm">Previous</a>] [<a href = "theindex.htm">Index</a>]
<h1>8 Designs</h1><p>
<P>
<H3>Sections</H3>
<oL>
<li> <A HREF="CHAP008.htm#SECT001">Designs from planar nearrings</a>
<li> <A HREF="CHAP008.htm#SECT002">Information on designs</a>
</ol><p>
<p>
Various designs can be obtained from nearrings. The design structure and
basic functions for the manipulation of designs have been implemented within
SONATA, as well as the means to construct the most popular nearring designs.
Please see the reference manual for other ways to generate designs, e.g.
by a set of blocks or by an incidence matrix.
<p>
<p>
<h2><a name="SECT001">8.1 Designs from planar nearrings</a></h2>
<p><p>
We are going to generate a design with a feasible parameter set from a planar
nearring, respectively from a Ferrero pair.
For notations and definitions we refer to citeClay:Nearrings.
<p>
Let <var>N</var> be a (left) planar nearring. Then we can define a design
<var>(N,B<sup>*</sup>,in)</var> with <var>N</var> as the set of points and
<var>{N<sup>*</sup>cdota+b | a,binN, a not= 0 }</var> as set of blocks.
Here <var>N<sup>*</sup> = { xinN | xcdotN = N }</var>.
We note that such a design is always a BIB-design. Suppose that the planar
nearring <var>N</var> is obtained from the Ferrero pair <var>(G,Phi)</var> as described in the
previous chapter with <var>v := |G|</var>, <var>k := |Phi|</var>.
We can identify <var>G</var> and the additive group of the nearring <var>N</var>.
Then <var>N<sup>*</sup>cdota = Phi(a)</var> for all <var>ainN</var>. The number of points of
<var>(N,B<sup>*</sup>,in)</var> is <var>v</var>, each block has size <var>k</var>, and any <var>2</var> distinct points
are together incident with precisely <var>k-1</var> blocks. We say that <var>(N,B<sup>*</sup>,in)</var>
is a <var>2-(v,k,k-1)</var> design. Note that <var>k</var> divides <var>v-1</var>.
<p>
<strong>The problem:</strong>
Find a <var>2-(16,5,4)</var> design <var>(N,B<sup>*</sup>,in)</var> if possible.
<p>
<strong>The solution:</strong>
Since our design should have <var>16</var> points, we need a planar nearring of size
<var>16</var>. Furthermore, this planar nearring has an underlying Ferrero pair with
a fixed-point-free (fpf) automorphism group of order <var>5</var>, since the blocks
should be of that size.
<p>
Let <var>(G,Phi)</var> be such a Ferrero pair.
We note that <var>Phi</var> is fpf on all characteristic subgroups of <var>G</var>, in
particular, on the center of <var>G</var>. For our example the size of the center
modulo <var>5</var> has to be congruent to <var>1</var>. Therefore the center has order <var>16</var>,
and <var>G</var> is abelian.
Since <var>Phi</var> is also fpf on any factor of <var>G</var> by a characteristic subgroup,
<var>Phi</var> is fpf on the factor of <var>G</var> by its Frattini subgroup.
Thus <var>G</var> is elementary abelian.
<p>
Now we compute an fpf automorphism group of order 5 on the elementary abelian
group of order 16:
<p>
<pre>
gap> aux := FpfAutomorphismGroupsCyclic( [2,2,2,2], 5 );
[ [ [ f1, f2, f3, f4 ] -> [ f4, f1*f2, f2*f3, f3*f4 ] ],
<pc group of size 16 with 4 generators> ]
gap> a := aux[1][1];
[ f1, f2, f3, f4 ] -> [ f4, f1*f2, f2*f3, f3*f4 ]
gap> phi := Group( a );;
gap> G := aux[2];;
</pre>
<p>
Up to conjugacy there is only one fpf automorphism group of order 5.
We note that all planar nearrings obtained from a fixed Ferrero pair
<var>(G,Phi)</var> yield the same design. We build the blocks <var>Phi(a)+b</var> for
<var>a,binG, anot=0</var>, of <var>(N,B<sup>*</sup>,in)</var> directly from <var>(G,Phi)</var> without
actually generating a nearring.
<p>
<pre>
gap> D := DesignFromFerreroPair( G, phi, "*" );
<a 2 - ( 16, 5, 4 ) nearring generated design>
</pre>
<p>
The reference manual describes other options for <code>DesignFromFerreroPair</code>
besides <code>"*"</code> and the correponding designs as well.
<p>
<p>
<h2><a name="SECT002">8.2 Information on designs</a></h2>
<p><p>
We investigate the design <code>D</code>, which we have generated from a planar
nearring in the last section.
<p>
<pre>
gap> D;
<a 2 - ( 16, 5, 4 ) nearring generated design>
gap> DesignParameter( D );
[ 2, 16, 48, 15, 5, 4 ]
</pre>
<p>
<code>DesignParameter( D )</code> returns the set of parameters <var>t, v, b, r, k, lambda</var>
of the design <code>D</code>. Here there are 16 points, 48 blocks,
every point is incident with precisely 15 blocks, every block is incident
with precisely 5 points, every 2 distinct points are together incident
with precisely 4 blocks. The design can be visualized by printing the
incidence matrix. The rows are labelled by the points, the columns by
the blocks. The point of number <var>i</var> is incident with the block of number <var>j</var>
if and only if the entry in the <var>i</var>-th row, <var>j</var>-th column is 1.
<p>
<pre>
gap> PrintIncidenceMat( D );
...1...1...11...1...1..1..11..1...1..1..1...11..
1......1...1..1.1...11...1..1..1...11..1..1....1
.1..1......11..1...1..1...11....1..1..11..1...1.
..1..1..1...1...1.1...1..1...11...1.1....11...1.
1....11..1......1..1.1....1.1...11...1...1..1.1.
.1..1.1...1..1......11....1..11....11....1.1.1..
..1..1.1.1...1...1...1...1.1....1.1...11...1.1..
..11...1..1.1..1..1.....1...1...1..1.1...1.1...1
..1.1...1.1...1..1.1.1.....1..1..1...1..1.1....1
1...1.1....1.1...11...1.1.......1.1.1...1...1..1
1...1...11....11....1..11....1....1..1.1...1..1.
.1...1..1.1.1....1.1...11...1..1....1..1....11..
.1.1....11...1.1....1.1..1.1...1.1.......1..1..1
.1.1..1....1..1..11....1.1..1.1..1....1....1..1.
..11..1..1....1.1..1..1.1....1.1...1..1.1....1..
1....1.1..1..1.1..1....1..1..1.1.1....1.1.1.....
</pre>
<p>
By checking the entries in the incidence matrix we realize this design is
circular, that is, any <var>2</var> distinct blocks have at most <var>2</var> points in common.
Actually, this is not so easy to see. We prefer to do it like this:
<p>
<pre>
gap> IsCircularDesign( D );
true
</pre>
<p>
There are convenient functions to check which points are incident with
a given set of blocks and vice versa.
<p>
<pre>
gap> PointsIncidentBlocks( D, [2,7] );
[ 6, 14 ]
gap> BlocksIncidentPoints( D, [6,14] );
[ 2, 7, 31, 44 ]
</pre>
<p>
Here the 6-th and the 14-th point are incident with the given blocks with
numbers 2 and 7. The numbering corresponds to that of the columns and rows of
the incidence matrix. Blocks 2, 7, 31, and 44 are incident with points 6 and
14.
<p>
The cardinalities of the intersections of one block with all the others,
e.g. of the 4-th block with all 48 blocks are given like this.
<p>
<pre>
gap> BlockIntersectionNumbers( D, 4 );
[ 0, 2, 2, 5, 0, 0, 2, 2, 1, 2, 1, 2, 2, 1, 2, 2, 2, 1, 2, 1, 2, 0,
2, 2, 2, 2, 1, 2, 2, 1, 2, 2, 1, 2, 1, 2, 0, 2, 2, 0, 2, 2, 0, 2,
2, 2, 1, 2 ]
</pre>
<p>
Of course, all these functions can be applied to all kinds of designs no
matter how they have been generated.
<p>
<p>
[<a href = "chapters.htm">Up</a>] [<a href ="CHAP007.htm">Previous</a>] [<a href = "theindex.htm">Index</a>]
<P>
<address>SONATA-tutorial manual<br>November 2008
</address></body></html>
distrib
>
Mandriva
>
2010.0
>
i586
>
media
>
contrib-release
>
by-pkgid
>
91213ddcfbe7f54821d42c2d9e091326
>
files
>
2817
