Namespaces
Variants
Actions

Difference between revisions of "Howell design"

From Encyclopedia of Mathematics
Jump to: navigation, search
m (fix formatting)
m (link)
 
(One intermediate revision by one other user not shown)
Line 1: Line 1:
A Howell design of side <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103301.png" /> and order <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103302.png" />, or more briefly an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103303.png" />, is an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103304.png" />-array in which each cell is either empty or contains an unordered pair of distinct elements from some <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103305.png" />-set <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103306.png" /> such that:
+
<!--
 +
h1103301.png
 +
$#A+1 = 78 n = 2
 +
$#C+1 = 78 : ~/encyclopedia/old_files/data/H110/H.1100330 Howell design
 +
Automatically converted into TeX, above some diagnostics.
 +
Please remove this comment and the {{TEX|auto}} line below,
 +
if TeX found to be correct.
 +
-->
  
1) every element of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103307.png" /> occurs in precisely one cell of each row and each column;
+
{{TEX|auto}}
 +
{{TEX|done}}
  
2) every unordered pair of elements from <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103308.png" /> is in at most one cell of the array. It follows immediately from the definition of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h1103309.png" /> that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033010.png" />. An example of a Howell design is the following <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033011.png" />:
+
A Howell design of side  $  s $
 +
and order  $  2n $,
 +
or more briefly an  $  H ( s,2n ) $,
 +
is an  $  ( s \times s ) $-
 +
array in which each cell is either empty or contains an unordered pair of distinct elements from some  $  2n $-
 +
set  $  V $
 +
such that:
  
<table border="0" cellpadding="0" cellspacing="0" style="background-color:black;"> <tr><td> <table border="0" cellspacing="1" cellpadding="4" style="background-color:black;"> <tbody> <tr> <td colname="1" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033012.png" /> 0</td> <td colname="2" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1">1 3</td> <td colname="4" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033013.png" /> 2</td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1">2 3</td> <td colname="2" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033014.png" /> 1</td> <td colname="3" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033015.png" /> 0</td> <td colname="4" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033016.png" /> 3</td> <td colname="3" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033017.png" /> 2</td> <td colname="4" style="background-color:white;" colspan="1">0 1</td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033018.png" /> 1</td> <td colname="2" style="background-color:white;" colspan="1">0 2</td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033019.png" /> 3</td> </tr> </tbody> </table>
+
1) every element of  $  V $
 +
occurs in precisely one cell of each row and each column;
 +
 
 +
2) every unordered pair of elements from  $  V $
 +
is in at most one cell of the array. It follows immediately from the definition of an  $  H ( s,2n ) $
 +
that  $  n \leq  s \leq  2n - 1 $.
 +
An example of a Howell design is the following  $  H ( 4,6 ) $:
 +
 
 +
<table border="0" cellpadding="0" cellspacing="0" style="background-color:black;"> <tr><td> <table border="0" cellspacing="1" cellpadding="4" style="background-color:black;"> <tbody> <tr> <td colname="1" style="background-color:white;" colspan="1"> $  \alpha $
 +
0</td> <td colname="2" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1">1 3</td> <td colname="4" style="background-color:white;" colspan="1"> $  \infty $
 +
2</td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1">2 3</td> <td colname="2" style="background-color:white;" colspan="1"> $  \alpha $
 +
1</td> <td colname="3" style="background-color:white;" colspan="1"> $  \infty $
 +
0</td> <td colname="4" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"> $  \infty $
 +
3</td> <td colname="3" style="background-color:white;" colspan="1"> $  \alpha $
 +
2</td> <td colname="4" style="background-color:white;" colspan="1">0 1</td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"></td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="2" style="background-color:white;" colspan="1"></td> </tr> <tr> <td colname="1" style="background-color:white;" colspan="1"> $  \infty $
 +
1</td> <td colname="2" style="background-color:white;" colspan="1">0 2</td> <td colname="3" style="background-color:white;" colspan="1"></td> <td colname="4" style="background-color:white;" colspan="1"> $  \alpha $
 +
3</td> </tr> </tbody> </table>
  
 
</td></tr> </table>
 
</td></tr> </table>
  
An <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033020.png" /> is also called a [[Room square|Room square]] of side <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033021.png" />. At the other extreme, the existence of a pair of mutually [[Orthogonal Latin squares|orthogonal Latin squares]] implies the existence of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033022.png" />. The existence of Howell designs has been completely determined [[#References|[a1]]], [[#References|[a5]]]: Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033023.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033024.png" /> be positive integers such that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033025.png" />. There exists an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033026.png" /> if and only if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033027.png" /> or <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033028.png" />. The proof uses a variety of direct and recursive constructions.
+
An $  H ( 2n - 1,2n ) $
 +
is also called a [[Room square|Room square]] of side $  2n - 1 $.  
 +
At the other extreme, the existence of a pair of mutually [[Orthogonal Latin squares|orthogonal Latin squares]] implies the existence of an $  H ( n,2n ) $.  
 +
The existence of Howell designs has been completely determined [[#References|[a1]]], [[#References|[a5]]]: Let $  s $
 +
and $  n $
 +
be positive integers such that 0 \leq  n \leq  s \leq  2n - 1 $.  
 +
There exists an $  H ( s,2n ) $
 +
if and only if $  ( s,2n ) \neq ( 2,4 ) , ( 3,4 ) , ( 5,6 ) $
 +
or $  ( 5,8 ) $.  
 +
The proof uses a variety of direct and recursive constructions.
  
An <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033029.png" /> is an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033030.png" /> in which there is a subset <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033031.png" /> of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033032.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033033.png" />, such that no pair of elements from <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033034.png" /> appears in the design. <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033036.png" />-designs are quite useful in recursive constructions. There exist <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033037.png" /> for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033038.png" /> even, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033039.png" />, with two exceptions: there is no <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033040.png" /> and there is no <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033041.png" /> [[#References|[a1]]]. The existence of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033042.png" /> for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033043.png" /> odd remains open, see [[#References|[a5]]]. The only known case where an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033044.png" /> exists but an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033045.png" /> does not is for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033046.png" />.
+
An $  H  ^ {*} ( s,2n ) $
 +
is an $  H ( s,2n ) $
 +
in which there is a subset $  W $
 +
of $  V $,  
 +
$  | W | = 2n - s $,  
 +
such that no pair of elements from $  W $
 +
appears in the design. $  * $-
 +
designs are quite useful in recursive constructions. There exist $  H  ^ {*} ( s,2n ) $
 +
for $  s $
 +
even, $  n \leq  s \leq  2n - 2 $,  
 +
with two exceptions: there is no $  H  ^ {*} ( 2,4 ) $
 +
and there is no $  H  ^ {*} ( 6,12 ) $[[#References|[a1]]]. The existence of $  H  ^ {*} ( s,2n ) $
 +
for $  s $
 +
odd remains open, see [[#References|[a5]]]. The only known case where an $  H ( s,2n ) $
 +
exists but an $  H  ^ {*} ( s,2n ) $
 +
does not is for $  s = n = 6 $.
  
The pairs of elements in the cells of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033047.png" /> can be thought of as the edges of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033048.png" />-regular [[Graph|graph]] on the <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033049.png" />-set <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033050.png" />, the underlying graph of the Howell design. The existence of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033051.png" /> is equivalent to the existence of a pair of orthogonal one-factorizations of the underlying graph of the <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033052.png" /> (cf. [[One-factorization|One-factorization]]). The underlying graph of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033053.png" /> is the complete graph <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033054.png" />, and the underlying graph of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033055.png" /> is the [[cocktail party graph]] <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033056.png" />, where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033057.png" /> is a [[One-factor|one-factor]]. An <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033058.png" /> with underlying graph <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033059.png" /> is equivalent to a pair of mutually [[Orthogonal Latin squares|orthogonal Latin squares]] of order <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033060.png" />. The general problem of determining which graphs are the underlying graphs of a Howell design remains open (1996), see [[#References|[a3]]].
+
The pairs of elements in the cells of an $  H ( s,2n ) $
 +
can be thought of as the edges of an $  s $-
 +
[[regular graph]] on the $  2n $-
 +
set $  V $,  
 +
the underlying graph of the Howell design. The existence of an $  H ( s,2n ) $
 +
is equivalent to the existence of a pair of orthogonal one-factorizations of the underlying graph of the $  H ( s,2n ) $(
 +
cf. [[One-factorization|One-factorization]]). The underlying graph of an $  H ( 2n - 1,2n ) $
 +
is the complete graph $  K _ {2n }  $,  
 +
and the underlying graph of an $  H ( 2n - 2,2n ) $
 +
is the [[cocktail party graph]] $  K _ {2n }  - f $,  
 +
where $  f $
 +
is a [[One-factor|one-factor]]. An $  H ( n,2n ) $
 +
with underlying graph $  K _ {n,n }  $
 +
is equivalent to a pair of mutually [[Orthogonal Latin squares|orthogonal Latin squares]] of order $  n $.  
 +
The general problem of determining which graphs are the underlying graphs of a Howell design remains open (1996), see [[#References|[a3]]].
  
Several special types of Howell designs have been studied, including <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033061.png" />-designs, skew designs, complementary designs, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033062.png" />-complementary designs, cyclic Howell designs (used for Howell movements in duplicate bridge), and Howell designs with Howell sub-designs (see [[#References|[a3]]] [[#References|[a4]]]).
+
Several special types of Howell designs have been studied, including $  ** $-
 +
designs, skew designs, complementary designs, $  * $-
 +
complementary designs, cyclic Howell designs (used for Howell movements in duplicate bridge), and Howell designs with Howell sub-designs (see [[#References|[a3]]] [[#References|[a4]]]).
  
A <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033064.png" />-dimensional Howell design, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033065.png" />, is a <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033066.png" />-dimensional array in which every cell either is empty or contains an unordered pair of elements from a <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033067.png" />-set <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033068.png" /> and such that each two-dimensional projection is an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033069.png" />. An <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033070.png" /> is called a Howell cube. An <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033071.png" /> is equivalent to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033072.png" /> mutually orthogonal one-factorizations of the underlying graph. Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033073.png" /> denote the maximum value of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033074.png" /> such that an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033075.png" /> exists. Very little is known about upper bounds for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033076.png" />. It is easy to see that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033077.png" />, and it has been conjectured that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033078.png" />. See [[#References|[a3]]], [[#References|[a2]]] for results on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033079.png" /> and existence results on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h110/h110330/h11033080.png" />.
+
A $  d $-
 +
dimensional Howell design, $  H _ {d} ( s,2n ) $,  
 +
is a $  d $-
 +
dimensional array in which every cell either is empty or contains an unordered pair of elements from a $  2n $-
 +
set $  V $
 +
and such that each two-dimensional projection is an $  H ( s,2n ) $.  
 +
An $  H _ {3} ( s,2n ) $
 +
is called a Howell cube. An $  H _ {d} ( s,2n ) $
 +
is equivalent to $  d $
 +
mutually orthogonal one-factorizations of the underlying graph. Let $  \nu ( s,2n ) $
 +
denote the maximum value of $  d $
 +
such that an $  H _ {d} ( s,2n ) $
 +
exists. Very little is known about upper bounds for $  \nu ( s,2n ) $.  
 +
It is easy to see that $  \nu ( s,2n ) \leq  s - 1 $,  
 +
and it has been conjectured that $  \nu ( s,2n ) \leq  n - 1 $.  
 +
See [[#References|[a3]]], [[#References|[a2]]] for results on $  \nu ( s,2n ) $
 +
and existence results on $  H _ {d} ( s,2n ) $.
  
 
The survey article [[#References|[a3]]] includes results and references on Howell designs.
 
The survey article [[#References|[a3]]] includes results and references on Howell designs.

Latest revision as of 09:31, 19 January 2021


A Howell design of side $ s $ and order $ 2n $, or more briefly an $ H ( s,2n ) $, is an $ ( s \times s ) $- array in which each cell is either empty or contains an unordered pair of distinct elements from some $ 2n $- set $ V $ such that:

1) every element of $ V $ occurs in precisely one cell of each row and each column;

2) every unordered pair of elements from $ V $ is in at most one cell of the array. It follows immediately from the definition of an $ H ( s,2n ) $ that $ n \leq s \leq 2n - 1 $. An example of a Howell design is the following $ H ( 4,6 ) $:

<tbody> </tbody>
$ \alpha $ 0 1 3 $ \infty $ 2
2 3 $ \alpha $ 1 $ \infty $ 0
$ \infty $ 3 $ \alpha $ 2 0 1
$ \infty $ 1 0 2 $ \alpha $ 3

An $ H ( 2n - 1,2n ) $ is also called a Room square of side $ 2n - 1 $. At the other extreme, the existence of a pair of mutually orthogonal Latin squares implies the existence of an $ H ( n,2n ) $. The existence of Howell designs has been completely determined [a1], [a5]: Let $ s $ and $ n $ be positive integers such that $ 0 \leq n \leq s \leq 2n - 1 $. There exists an $ H ( s,2n ) $ if and only if $ ( s,2n ) \neq ( 2,4 ) , ( 3,4 ) , ( 5,6 ) $ or $ ( 5,8 ) $. The proof uses a variety of direct and recursive constructions.

An $ H ^ {*} ( s,2n ) $ is an $ H ( s,2n ) $ in which there is a subset $ W $ of $ V $, $ | W | = 2n - s $, such that no pair of elements from $ W $ appears in the design. $ * $- designs are quite useful in recursive constructions. There exist $ H ^ {*} ( s,2n ) $ for $ s $ even, $ n \leq s \leq 2n - 2 $, with two exceptions: there is no $ H ^ {*} ( 2,4 ) $ and there is no $ H ^ {*} ( 6,12 ) $[a1]. The existence of $ H ^ {*} ( s,2n ) $ for $ s $ odd remains open, see [a5]. The only known case where an $ H ( s,2n ) $ exists but an $ H ^ {*} ( s,2n ) $ does not is for $ s = n = 6 $.

The pairs of elements in the cells of an $ H ( s,2n ) $ can be thought of as the edges of an $ s $- regular graph on the $ 2n $- set $ V $, the underlying graph of the Howell design. The existence of an $ H ( s,2n ) $ is equivalent to the existence of a pair of orthogonal one-factorizations of the underlying graph of the $ H ( s,2n ) $( cf. One-factorization). The underlying graph of an $ H ( 2n - 1,2n ) $ is the complete graph $ K _ {2n } $, and the underlying graph of an $ H ( 2n - 2,2n ) $ is the cocktail party graph $ K _ {2n } - f $, where $ f $ is a one-factor. An $ H ( n,2n ) $ with underlying graph $ K _ {n,n } $ is equivalent to a pair of mutually orthogonal Latin squares of order $ n $. The general problem of determining which graphs are the underlying graphs of a Howell design remains open (1996), see [a3].

Several special types of Howell designs have been studied, including $ ** $- designs, skew designs, complementary designs, $ * $- complementary designs, cyclic Howell designs (used for Howell movements in duplicate bridge), and Howell designs with Howell sub-designs (see [a3] [a4]).

A $ d $- dimensional Howell design, $ H _ {d} ( s,2n ) $, is a $ d $- dimensional array in which every cell either is empty or contains an unordered pair of elements from a $ 2n $- set $ V $ and such that each two-dimensional projection is an $ H ( s,2n ) $. An $ H _ {3} ( s,2n ) $ is called a Howell cube. An $ H _ {d} ( s,2n ) $ is equivalent to $ d $ mutually orthogonal one-factorizations of the underlying graph. Let $ \nu ( s,2n ) $ denote the maximum value of $ d $ such that an $ H _ {d} ( s,2n ) $ exists. Very little is known about upper bounds for $ \nu ( s,2n ) $. It is easy to see that $ \nu ( s,2n ) \leq s - 1 $, and it has been conjectured that $ \nu ( s,2n ) \leq n - 1 $. See [a3], [a2] for results on $ \nu ( s,2n ) $ and existence results on $ H _ {d} ( s,2n ) $.

The survey article [a3] includes results and references on Howell designs.

References

[a1] B.A. Anderson, P.J. Schellenberg, D.R. Stinson, "The existence of Howell designs of even side" J. Combin. Th. A , 36 (1984) pp. 23–55
[a2] J.H. Dinitz, "Howell designs" C.J. Colbourn (ed.) J.H. Dinitz (ed.) , CRC Handbook of Combinatorial Designs , CRC (1996) pp. 381–385
[a3] J.H. Dinitz, D.R. Stinson, "Room squares and related designs" J.H. Dinitz (ed.) D.R. Stinson (ed.) , Contemporary Design Theory: A Collection of Surveys , Wiley (1992) pp. 137–204
[a4] E.R. Lamken, S.A. Vanstone, "The existence of skew Howell designs of side and order " J. Combin. Th. A , 54 (1990) pp. 20–40
[a5] D.R. Stinson, "The existence of Howell designs of odd side" J. Combin. Th. A , 32 (1982) pp. 53–65
How to Cite This Entry:
Howell design. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Howell_design&oldid=37181
This article was adapted from an original article by E.R. Lamken (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article