[Up] [Previous] [Next] [Index]

# 8 Classifying semi-Latin squares

### Sections

This chapter describes the function `SemiLatinSquareDuals` which can classify semi-Latin squares with certain given properties, and return a list of their duals as block designs.

## 8.1 Semi-Latin squares and SOMAs

Let n and k be positive integers. An (ntimesn)/k semi-Latin squareindexsemi-Latin square is an n by n array A, whose entries are k-subsets of a kn-set X (the symbol-set), such that each element of X occurs exactly once in each row and exactly once in each column of A. (Thus an (ntimesn)/1 semi-Latin square is the same thing as a Latin square of order n.) For extensive useful information on semi-Latin squares, see http://www.maths.qmul.ac.uk/~rab/sls.html.

A SOMA(k,n)indexSOMA is an (ntimesn)/k semi-Latin square A, with nge2, in which no 2-subset of the symbol-set is contained in more than one entry of A. For extensive useful information on SOMAs, see http://www.maths.qmul.ac.uk/~leonard/soma/.

Let A and B be (ntimesn)/k semi-Latin squares. We say that B is (weakly) isomorphic to A if B can be obtained from A by applying one or more of: a row permutation; a column permutation; transposing; renaming the symbols. If transposing is not allowed then we get the concept of strong isomorphism. More formally, B is strongly isomorphic to A if B can be obtained from A by applying one or more of: a row permutation; a column permutation; renaming the symbols.

Let A be an (ntimesn)/k semi-Latin square. Then the dual of A can be represented as a binary block design as follows. The point-set of D is taken to be the Cartesian square of {1,...,n}, with [x,y] representing the [x,y]-entry of A. The blocks of D are in one-to-one correspondance with the symbols of A, with the i-th block of D consisting of the ordered pairs [x,y] such that the i-th symbol of A is contained in the [x,y]-entry of A. Given D, the semi-Latin square A can be recovered, up to the naming of its symbols.

## 8.2 The function SemiLatinSquareDuals

• `SemiLatinSquareDuals( `n`, `k` )`
• `SemiLatinSquareDuals( `n`, `k`, `maxmult` )`
• `SemiLatinSquareDuals( `n`, `k`, `maxmult`, `blockintsizes` )`
• `SemiLatinSquareDuals( `n`, `k`, `maxmult`, `blockintsizes`, `isolevel` )`

Let n and k be positive integers. Then this function (which makes heavy use of the function `BlockDesigns`) returns a list DL of block designs which are the duals of the (ntimesn)/k semi-Latin squares whose properties are specified by the given parameters, described below. In practice, depending on the specified properties, this function can be useful for n up to about 6 or 7.

The parameter maxmult, if given, must be a positive integer or the string `"default"`. If it is a positive integer, then maxmult specifies an upper bound on the multiplicity of each block in each semi-Latin square dual in DL. The default value for maxmult (if omitted or if given as `"default"`) is k, which poses no constraint on the block multiplicities.

The parameter blockintsizes, if given, must be a set of non-negative integers or the string `"default"`. If it is given as a set, then blockintsizes specifies, for each semi-Latin square dual in DL, the set of possible sizes for the intersection of a block B with a different block (but possibly a repeat of B). The default value for blockintsizes (if omitted or if given as `"default"`) is `[0..`n`]`, which poses no constraint on the block intersection sizes. Note that block intersection sizes in the dual of a semi-Latin square correspond to concurrencies of points in the semi-Latin square itself. Also note that if nge2 and blockintsizes is specified to be `[0,1]` then the (ntimesn)/k semi-Latin squares being considered are SOMA(k,n)s.

The parameter isolevel, if given, must be 0, 1, 2, 3, 4 or the string `"default"` (the default value is 2). The value 0 specifies that DL will contain at most one (semi-Latin square dual given as a) block design, and will contain one such block design if and only if a semi-Latin square with the required properties exists. The value 1 specifies that DL will contain a list of duals representing all weak isomorphism classes of semi-Latin squares with the required properties (possibly with some classes represented more than once) and the value 2 specifies that DL will contain precisely one dual semi-Latin square representative for each weak isomorphism class of semi-Latin squares with the required properties. The values 3 and 4 for isolevel play the roles of 1 and 2, respectively, but with weak isomorphism replaced by strong isomorphism. Thus, isolevel=3 specifies that DL will contain a list of duals representing all strong isomorphism classes of semi-Latin squares with the required properties (possibly with some classes represented more than once) and isolevel=4 specifies that DL will contain precisely one dual semi-Latin square representative for each strong isomorphism class of semi-Latin squares with the required properties.

For example, we determine the numbers of weak and strong isomorphism classes of (4times4)/k semi-Latin squares for k=1,...,6. (These numbers disagree with P. E. Chigbu's classification for the cases k=3,4 BaCh.)

```gap> List([1..6],k->Length(SemiLatinSquareDuals(4,k))); # weak
[ 2, 10, 40, 164, 621, 2298 ]
gap> List([1..6],k->Length(SemiLatinSquareDuals(4,k,"default","default",4))); # strong
[ 2, 11, 46, 201, 829, 3343 ]
```

Next, we determine one SOMA(3,6).

```gap> SemiLatinSquareDuals(6,3,"default",[0,1],0);
[ rec( isBlockDesign := true, v := 36,
blocks := [ [ 1, 8, 15, 22, 29, 36 ], [ 1, 9, 16, 23, 30, 32 ],
[ 1, 12, 14, 21, 28, 35 ], [ 2, 9, 17, 24, 25, 34 ],
[ 2, 11, 18, 22, 27, 31 ], [ 2, 12, 16, 19, 29, 33 ],
[ 3, 10, 14, 24, 29, 31 ], [ 3, 11, 16, 20, 25, 36 ],
[ 3, 12, 13, 23, 26, 34 ], [ 4, 7, 14, 23, 27, 36 ],
[ 4, 8, 17, 21, 30, 31 ], [ 4, 9, 18, 19, 26, 35 ],
[ 5, 7, 15, 20, 30, 34 ], [ 5, 8, 13, 24, 28, 33 ],
[ 5, 10, 18, 21, 25, 32 ], [ 6, 7, 17, 22, 26, 33 ],
[ 6, 10, 13, 20, 27, 35 ], [ 6, 11, 15, 19, 28, 32 ] ],
tSubsetStructure := rec( t := 1, lambdas := [ 3 ] ), isBinary := true,
isSimple := true, blockSizes := [ 6 ], blockNumbers := [ 18 ], r := 3,
autSubgroup := <permutation group of size 72 with 3 generators>,
pointNames := [ [ 1, 1 ], [ 1, 2 ], [ 1, 3 ], [ 1, 4 ], [ 1, 5 ],
[ 1, 6 ], [ 2, 1 ], [ 2, 2 ], [ 2, 3 ], [ 2, 4 ], [ 2, 5 ],
[ 2, 6 ], [ 3, 1 ], [ 3, 2 ], [ 3, 3 ], [ 3, 4 ], [ 3, 5 ],
[ 3, 6 ], [ 4, 1 ], [ 4, 2 ], [ 4, 3 ], [ 4, 4 ], [ 4, 5 ],
[ 4, 6 ], [ 5, 1 ], [ 5, 2 ], [ 5, 3 ], [ 5, 4 ], [ 5, 5 ],
[ 5, 6 ], [ 6, 1 ], [ 6, 2 ], [ 6, 3 ], [ 6, 4 ], [ 6, 5 ],
[ 6, 6 ] ] ) ]
```

[Up] [Previous] [Next] [Index]

design manual
November 2011