This chapter consists essentially of four parts, according to the four different types of data structures that are described, after the usual brief discussion of the objects (see More about Row Spaces, Row Space Bases, Row Space Cosets, Quotient Spaces, Subspaces and Parent Spaces).

The first part introduces row spaces, and their operations and functions (see RowSpace, Operations for Row Spaces, Functions for Row Spaces, IsRowSpace, Subspace, AsSubspace, AsSpace, NormedVectors).

The second part introduces bases for row spaces, and their operations and functions (see Coefficients for Row Space Bases, SiftedVector, Basis, CanonicalBasis, SemiEchelonBasis, IsSemiEchelonBasis, NumberVector, ElementRowSpace).

The third part introduces row space cosets, and their operations and functions (see Operations for Row Space Cosets, Functions for Row Space Cosets, IsSpaceCoset).

The fourth part introduces quotient spaces of row spaces, and their operations and functions (see Operations for Quotient Spaces, Functions for Quotient Spaces).

The obligatory last sections describe the details of the implementation of the data structures (see Row Space Records, Row Space Basis Records, Row Space Coset Records, Quotient Space Records).

**Note**:
The current implementation of row spaces provides no homomorphisms of
row spaces (linear maps), and also quotient spaces of quotient spaces are
not supported.

- More about Row Spaces
- Row Space Bases
- Row Space Cosets
- Quotient Spaces
- Subspaces and Parent Spaces
- RowSpace
- Operations for Row Spaces
- Functions for Row Spaces
- IsRowSpace
- Subspace
- AsSubspace
- AsSpace
- NormedVectors
- Coefficients for Row Space Bases
- SiftedVector
- Basis
- CanonicalBasis
- SemiEchelonBasis
- IsSemiEchelonBasis
- NumberVector
- ElementRowSpace
- Operations for Row Space Cosets
- Functions for Row Space Cosets
- IsSpaceCoset
- Operations for Quotient Spaces
- Functions for Quotient Spaces
- Row Space Records
- Row Space Basis Records
- Row Space Coset Records
- Quotient Space Records

A **row space** is a vector space (see chapter Vector Spaces), whose elements
are row vectors, that is, lists of elements in a common field.

**Note** that for a row space *V* over the field *F* necessarily the
characteristic of *F* is the same as the characteristic of the vectors in
*V*.
Furthermore at the moment the field *F* must contain the field spanned by all
the elements in vectors of *V*, since in many computations vectors are
normed, that is, divided by their first nonzero entry.

The implementation of functions for these spaces and their elements uses the
well-known linear algebra methods, such as Gaussian elimination, and many
functions delegate the work to functions for matrices, e.g., a basis of a
row space can be computed by performing Gaussian elimination to the matrix
formed by the list of generators. Thus in a sense, a row space in **GAP3** is
nothing but a **GAP3** object that knows about the interpretation of a matrix
as a generating set, and that knows about the functions that do the work.

Row spaces are constructed using RowSpace `RowSpace`

, full row spaces can
also be constructed by

, for a field `F` ^ `n``F` and a positive integer
`n`.

The **zero element** of a row space `V` in **GAP3** is not necessarily stored in
the row space record. If necessary, it can be computed using `Zero( `

.
`V` )

The `generators`

component may contain zero vectors, so no function should
expect a generator to be nonzero.

For the usual concept of substructures and parent structures see Subspaces and Parent Spaces.

See Operations for Row Spaces and Functions for Row Spaces for an overview of applicable operators and functions, and Row Space Records for details of the implementation.

Many computations with row spaces require the computation of a **basis**
(which will always mean a vector space basis in **GAP3**), such as the
computation of the dimension, or efficient membership test for the row
space.

Most of these computations do not rely on special properties of the chosen basis. The computation of coefficients lists, however, is basis dependent. A natural way to distinguish these two situations is the following.

For basis independent questions the row space is allowed to compute a
suitable basis, and may store bases. For example the dimension of the space
`V` can be computed this way using `Dimension( `

.
In such situations the component `V` )

is used. The value of this
component depends on how it was constructed, so no function that accesses
this component should assume special properties of this basis.
`V`.basis

On the other hand, the computation of coefficients of a vector `v` with
respect to a basis `B` of `V` depends on this basis, so you have to call
`Coefficients( `

, and `B`, `v` )**not** `Coefficients( `

.
`V`, `v` )

It should be mentioned that there are two types of row space bases.
A basis of the first type is **semi-echelonized** (see SemiEchelonBasis
for the definition and examples), its structure allows to perform
efficient calculations of membership test and coefficients.

A basis of the second type is **arbitrary**, that is, it has no special
properties. There are two ways to construct such a (user-defined) basis
that is **not** necessarily semi-echelonized. The first is to call
`RowSpace`

with the optional argument `"basis"`

; this means that the
generators are known to be linearly independent (see RowSpace).
The second way is to call `Basis`

with two arguments (see Basis).
The computation of coefficients with respect to an arbitrary basis is
performed by computing a semi-echelonized basis, delegating the task to
this basis, and then performing the base change.

The functions that are available for row space bases are `Coefficients`

(see Coefficients for Row Space Bases) and `SiftedVector`

(see
SiftedVector).

The several available row space bases are described in Basis, CanonicalBasis, and SemiEchelonBasis. For details of the implementation see Row Space Basis Records.

Let *V* be a vector space, and *U* a subspace of *V*.
The set *v + U = { v + u; u ∈ U}* is called a **coset** of *U* in *V*.

In **GAP3**, cosets are of course domains that can be formed using the
`'+'`

operator, see Operations for Row Space Cosets and Functions
for Row Space Cosets for an overview of applicable operators and
functions, and Row Space Coset Records for details of the
implementation.

A coset *C = v + U* is described by any representative *v* and the space *U*.
Equal cosets may have different representatives. A canonical representative
of the coset *C* can be computed using `CanonicalRepresentative( `

, it
does only depend on `C` )*C*, especially not on the basis of *U*.

Row spaces cosets can be regarded as elements of quotient spaces (see Quotient Spaces).

Let *V* be a vector space, and *U* a subspace of *V*.
The set *{ v + U; v ∈ V }* is again a vector space, the **quotient space**
(or factor space) of *V* modulo *U*.

By definition of row spaces, a quotient space is not a row space. (One reason to describe quotient spaces here is that for general vector spaces at the moment no factor structures are supported.)

Quotient spaces in **GAP3** are formed from two spaces using the `/`

operator. See the sections Operations for Quotient Spaces and
Functions for Quotient Spaces for an overview of applicable operators
and functions, and Quotient Space Records for details of the
implementation.

**Bases for Quotient Spaces of Row Spaces**

A basis *B* of a quotient *V / U* for row spaces *V* and *U* is best
described by bases of *V* and *U*.
If *B* is a basis without special properties then it will delegate the work
to a semi-echelonized basis.
The concept of **semi-echelonized bases** makes sense also for quotient spaces
of row spaces since for any semi-echelonized basis of *U* the set *S* of
pivot columns is a subset of the set of pivot columns of a semi-echelonized
basis of *V*.
So the cosets *v + U* for basis vectors *v* with pivot column not in *S*
form a semi-echelonized basis of *V / U*.
The **canonical basis** of *V / U* is the semi-echelonized basis derived in
that way from the canonical basis of *V* (see CanonicalBasis).

See Functions for Quotient Spaces for details about the bases.

The concept described in this section is essentially the same as the concept of parent groups and subgroups (see More about Groups and Subgroups).

(The section should be moved to chapter Vector Spaces, but for general vector spaces the concept does not yet apply.)

Every row space *U* is either constructed as **subspace** of an existing space
*V*, for example using Subspace `Subspace`

, or it is not.

In the latter case the space is called a **parent space**, in the former case
*V* is called the **parent** of *U*.

One can only form sums of subspaces of the same parent space, form quotient
spaces only for spaces with same parent, and cosets *v + U* only for
representatives *v* in the parent of *U*.

`Parent(`

:`V`)-

returns the parent space of the row space`V`,

`IsParent(`

:`V`)-

returns`true`

if the row space`V`is a parent space, and`false`

otherwise.

See AsSubspace, AsSpace for conversion functions.

`RowSpace( `

`F`, `generators` )

returns the row space that is generated by the vectors `generators` over
the field `F`. The elements in `generators` must be **GAP3** vectors.

`RowSpace( `

`F`, `generators`, `zero` )

Whenever the list `generators` is empty, this call of `RowSpace`

has to
be used, with `zero` the zero vector of the space.

`RowSpace( `

`F`, `generators`, "basis" )

also returns the `F`-space generated by `generators`. When the space is
constructed in this way, the vectors `generators` are assumed to form a
basis, and this is used for example when `Dimension`

is called for the
space.

It is **not** checked that the vectors are really linearly independent.

`RowSpace( `

`F`, `dimension` )

`F` ^ `n`

return the full row space of dimension `n` over the field `F`.
The elements of this row space are all the vectors of length `n` with entries
in `F`.

gap> v1:= RowSpace( GF(2), [ [ 1, 1 ], [ 0, 1 ] ] * Z(2) ); RowSpace( GF(2), [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) gap> v2:= RowSpace( GF(2), [], [ 0, 0 ] * Z(2) ); RowSpace( GF(2), [ [ 0*Z(2), 0*Z(2) ] ] ) gap> v3:= RowSpace( GF(2), [ [ 1, 1 ], [ 0, 1 ] ] * Z(2), "basis" ); RowSpace( GF(2), [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) gap> v4:= RowSpace( GF(2), 2 ); RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> v5:= GF(2) ^ 2 ; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> v3 = v4; true

Note that the list of generators may contain zero vectors.

**Comparisons of Row Spaces**

:`V`=`W`-

returns`true`

if the two row spaces`V`,`W`are equal as sets, and`false`

otherwise.

:`V`<`W`-

returns`true`

if the row space`V`is smaller than the row space`W`, and`false`

otherwise. The first criteria of this ordering are the comparison of the fields and the dimensions, row spaces over the same field and of same dimension are compared by comparison of the reversed canonical bases (see CanonicalBasis).

**Arithmetic Operations for Row Spaces**

:`V`+`W`-

returns the sum of the row spaces`V`and`W`, that is, the row space generated by`V`and`W`. This is computed using the Zassenhaus algorithm.

:`V`/`U`-

returns the quotient space of`V`modulo its subspace`U`(see Quotient Spaces).

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> s:= Subspace( v, [ [ 1, 1 ] * Z(2) ] ); Subspace( v, [ [ Z(2)^0, Z(2)^0 ] ] ) gap> t:= Subspace( v, [ [ 0, 1 ] * Z(2) ] ); Subspace( v, [ [ 0*Z(2), Z(2)^0 ] ] ) gap> s = t; false gap> s < t; false gap> t < s; true gap> u:= s+t; Subspace( v, [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) gap> u = v; true gap> f:= u / s; Subspace( v, [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) / [ [ Z(2)^0, Z(2)^0 ] ]

The following functions are overlaid in the operations record of row spaces.

The **set theoretic functions**

`Closure`

, `Elements`

, `Intersection`

, `Random`

, `Size`

.

`Intersection(`

:`V`,`W`)-

returns the intersection of the two row spaces`V`and`W`that is computed using the Zassenhaus algorithm.

The **vector space specific functions**

`Base(`

:`V`)-

returns the list of vectors of the canonical basis of the row space`V`(see CanonicalBasis).

`Cosets(`

:`V`,`U`)-

returns the list of cosets of the subspace`U`in`V`, as does`Elements(`

.`V`/`U`)

`Dimension(`

:`V`)-

returns the dimension of the row space. For this, a basis of the space is computed if not yet known.

`Zero(`

:`V`)-

returns the zero element of the row space`V`(see More about Row Spaces).

`IsRowSpace( `

`obj` )

returns `true`

if `obj`, which can be an object of arbitrary
type, is a row space and `false`

otherwise.

gap> v:= GF(2) ^ 2; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> IsRowSpace( v ); true gap> IsRowSpace( v / [ v.generators[1] ] ); false

`Subspace( `

`V`, `gens` )

returns the subspace of the row space `V` that is generated by the vectors
in the list `gens`.

gap> v:= GF(3)^2; v.name:= "v";; RowSpace( GF(3), [ [ Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3)^0 ] ] ) gap> s:= Subspace( v, [ [ 1, -1 ] *Z(3)^0 ] ); Subspace( v, [ [ Z(3)^0, Z(3) ] ] )

`AsSubspace( `

`V` ,`U` )

returns the row space `U`, viewed as a subspace of the rows space `V`.
For that, `V` must be a parent space.

gap> v:= GF(2)^2; v.name:="v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> u:= RowSpace( GF(2), [ [ 1, 1 ] * Z(2) ] ); RowSpace( GF(2), [ [ Z(2)^0, Z(2)^0 ] ] ) gap> w:= AsSubspace( v, u ); Subspace( v, [ [ Z(2)^0, Z(2)^0 ] ] ) gap> w = u; true

`AsSpace( `

`U` )

returns the subspace `U` as a parent space.

gap> v:= GF(2)^2; v.name:="v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> u:= Subspace( v, [ [ 1, 1 ] * Z(2) ] ); Subspace( v, [ [ Z(2)^0, Z(2)^0 ] ] ) gap> w:= AsSpace( u ); RowSpace( GF(2), [ [ Z(2)^0, Z(2)^0 ] ] ) gap> w = u; true

`NormedVectors( `

`V` )

returns the set of those vectors in the row space `V` for that the first
nonzero entry is the identity of the underlying field.

gap> v:= GF(3)^2; RowSpace( GF(3), [ [ Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3)^0 ] ] ) gap> NormedVectors( v ); [ [ 0*Z(3), Z(3)^0 ], [ Z(3)^0, 0*Z(3) ], [ Z(3)^0, Z(3)^0 ], [ Z(3)^0, Z(3) ] ]

`Coefficients( `

`B`, `v` )

returns the coefficients vector of the vector `v` with respect to the
basis `B` (see Row Space Bases) of the vector space *V*,
if `v` is an element of *V*. Otherwise `false`

is returned.

gap> v:= GF(3)^2; v.name:= "v";; RowSpace( GF(3), [ [ Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3)^0 ] ] ) gap> b:= Basis( v ); Basis( v, [ [ Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3)^0 ] ] ) gap> Coefficients( b, [ Z(3), Z(3) ] ); [ Z(3), Z(3) ] gap> Coefficients( b, [ Z(3), Z(3)^2 ] ); [ Z(3), Z(3)^0 ]

`SiftedVector( `

`B`, `v` )

returns the residuum of the vector `v` with respect to the basis `B` of
the vector space *V*. The exact meaning of this depends on the special
properties of `B`.

But in general this residuum is obtained on subtracting appropriate
multiples of basis vectors, and `v` is contained in *V* if and only if
`SiftedVector( `

is the zero vector of `B`, `v` )*V*.

gap> v:= GF(3)^2; v.name:= "v";; RowSpace( GF(3), [ [ Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3)^0 ] ] ) gap> s:= Subspace( v, [ [ 1, -1 ] *Z(3)^0 ] ); s.name:= "s";; Subspace( v, [ [ Z(3)^0, Z(3) ] ] ) gap> b:= Basis(s); SemiEchelonBasis( s, [ [ Z(3)^0, Z(3) ] ] ) gap> SiftedVector( b, [ Z(3), 0*Z(3) ] ); [ 0*Z(3), Z(3) ]

`Basis( `

`V` )

`Basis( `

`V`, `vectors` )

`Basis( `

returns a basis of the row space `V` )`V`. If the component

or `V`.canonicalBasis

was bound before the first
call to `V`.semiEchelonBasis`Basis`

for `V` then one of these bases is returned. Otherwise a
semi-echelonized basis (see Row Space Bases) is computed.
The basis is stored in

.
`V`.basis

`Basis( `

returns the basis of `V`, `vectors` )`V` that consists of the
vectors in the list `vectors`. In the case that

was not bound
before the call the basis is stored in this component.
`V`.basis

**Note** that it is not necessarily checked whether `vectors` is really
linearly independent.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> b:= Basis( v, [ [ 1, 1 ], [ 1, 0 ] ] * Z(2) ); Basis( v, [ [ Z(2)^0, Z(2)^0 ], [ Z(2)^0, 0*Z(2) ] ] ) gap> Coefficients( b, [ 0, 1 ] * Z(2) ); [ Z(2)^0, Z(2)^0 ] gap> IsSemiEchelonBasis( b ); false

`CanonicalBasis( `

`V` )

returns the canonical basis of the row space `V`. This is a special
semi-echelonized basis (see SemiEchelonBasis), with the additional
properties that for *j > i* the position of the pivot of row *j* is bigger
than that of the pivot of row *i*, and that the pivot columns contain
exactly one nonzero entry.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> cb:= CanonicalBasis( v ); CanonicalBasis( v ) gap> cb.vectors; [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ]

The canonical basis is obtained on applying a full Gaussian elimination to
the generators of `V`, using BaseMat `BaseMat`

.
If the component

is bound then this basis is used to
compute the canonical basis, otherwise `V`.semiEchelonBasis`TriangulizeMat`

is called.

`SemiEchelonBasis( `

`V` )

`SemiEchelonBasis( `

`V`, `vectors` )

returns a semi-echelonized basis of the row space `V`.
A basis is called **semi-echelonized** if the first non-zero element in every
row is one, and all entries exactly below these elements are zero.

If a second argument `vectors` is given, these vectors are taken as basis
vectors. Note that if the rows of `vectors` do not form a semi-echelonized
basis then an error is signalled.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> SemiEchelonBasis( v ); SemiEchelonBasis( v, [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> b:= Basis( v, [ [ 1, 1 ], [ 0, 1 ] ] * Z(2) ); Basis( v, [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) gap> IsSemiEchelonBasis( b ); true gap> b; SemiEchelonBasis( v, [ [ Z(2)^0, Z(2)^0 ], [ 0*Z(2), Z(2)^0 ] ] ) gap> Coefficients( b, [ 0, 1 ] * Z(2) ); [ 0*Z(2), Z(2)^0 ] gap> Coefficients( b, [ 1, 0 ] * Z(2) ); [ Z(2)^0, Z(2)^0 ]

`IsSemiEchelonBasis( `

`B` )

returns `true`

if `B` is a semi-echelonized basis (see SemiEchelonBasis),
and `false`

otherwise.
If `B` is semi-echelonized, and this was not yet stored before, after the
call the operations record of `B` will be `SemiEchelonBasisRowSpaceOps`

.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> b1:= Basis( v, [ [ 0, 1 ], [ 1, 0 ] ] * Z(2) ); Basis( v, [ [ 0*Z(2), Z(2)^0 ], [ Z(2)^0, 0*Z(2) ] ] ) gap> IsSemiEchelonBasis( b1 ); true gap> b1; SemiEchelonBasis( v, [ [ 0*Z(2), Z(2)^0 ], [ Z(2)^0, 0*Z(2) ] ] ) gap> b2:= Basis( v, [ [ 0, 1 ], [ 1, 1 ] ] * Z(2) ); Basis( v, [ [ 0*Z(2), Z(2)^0 ], [ Z(2)^0, Z(2)^0 ] ] ) gap> IsSemiEchelonBasis( b2 ); false gap> b2; Basis( v, [ [ 0*Z(2), Z(2)^0 ], [ Z(2)^0, Z(2)^0 ] ] )

`NumberVector( `

`B`, `v` )

Let *<v> = ∑ _{i=1}^{n} λ_{i} b_{i}* where

`FFList`

.
Then `NumberVector( `

returns
`B`, `v` )*∑ _{i=1}^{n} λ_{i} q^{i-1}*.

gap> v:= GF(3)^3;; v.name:= "v";; gap> b:= CanonicalBasis( v );; gap> l:= List( [0 .. 6 ], x -> ElementRowSpace( b, x ) ); [ [ 0*Z(3), 0*Z(3), 0*Z(3) ], [ Z(3)^0, 0*Z(3), 0*Z(3) ], [ Z(3), 0*Z(3), 0*Z(3) ], [ 0*Z(3), Z(3)^0, 0*Z(3) ], [ Z(3)^0, Z(3)^0, 0*Z(3) ], [ Z(3), Z(3)^0, 0*Z(3) ], [ 0*Z(3), Z(3), 0*Z(3) ] ]

`ElementRowSpace( `

`B`, `n` )

returns the `n`-th element of the row space with basis `B`, with respect
to the ordering defined in NumberVector `NumberVector`

.

gap> v:= GF(3)^3;; v.name:= "v";; gap> b:= CanonicalBasis( v );; gap> l:= List( [0 .. 6 ], x -> ElementRowSpace( b, x ) );; gap> List( l, x -> NumberVector( b, x ) ); [ 0, 1, 2, 3, 4, 5, 6 ]

**Comparison of Row Space Cosets**

:`C1`=`C2`-

returns`true`

if the two row space cosets`C1`,`C2`are equal, and`false`

otherwise.**Note**that equal cosets need not have equal representatives (see Row Space Cosets).

:`C1`<`C2`-

returns`true`

if the row space coset`C1`is smaller than the row space coset`C2`, and`false`

otherwise. This ordering is defined by comparison of canonical representatives.

**Arithmetic Operations for Row Space Cosets**

:`C1`+`C2`-

If`C1`and`C2`are row space cosets that belong to the same quotient space, the result is the row space coset that is the sum resp. the difference of these vectors. Otherwise an error is signalled.

:`s`*`C`-

returns the row space coset that is the product of the scalar`s`and the row space coset`C`, where`s`must be an element of the ground field of the vector space that defines`C`.

**Membership Test for Row Space Cosets**

:`v`in`C`-

returns`true`

if the vector`v`

is an element of the row space coset`C`, and false otherwise.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> u:= Subspace( v, [ [ 1, 1 ] * Z(2) ] ); u.name:="u";; Subspace( v, [ [ Z(2)^0, Z(2)^0 ] ] ) gap> f:= v / u; v / [ [ Z(2)^0, Z(2)^0 ] ] gap> elms:= Elements( f ); [ ([ 0*Z(2), 0*Z(2) ]+u), ([ 0*Z(2), Z(2)^0 ]+u) ] gap> 2 * elms[2]; ([ 0*Z(2), 0*Z(2) ]+u) gap> elms[2] + elms[1]; ([ 0*Z(2), Z(2)^0 ]+u) gap> [ 1, 0 ] * Z(2) in elms[2]; true gap> elms[1] = elms[2]; false

Since row space cosets are domains, all set theoretic functions are applicable to them.

`Representative`

returns the value of the `representative`

component. **Note** that equal
cosets may have different representatives. Canonical representatives
can be computed using `CanonicalRepresentative`

.

`CanonicalRepresentative(`

:`C`)-

returns the canonical representative of the row space coset`C`, which is defined as the result of`SiftedVector(`

where`B`,`v`)

, and`C`=`v`+`U``B`is the canonical basis of`U`.

`IsSpaceCoset( `

`obj` )

returns `true`

if `obj`, which may be an arbitrary object, is a row space
coset, and `false`

otherwise.

gap> v:= GF(2)^2; v.name:= "v";; RowSpace( GF(2), [ [ Z(2)^0, 0*Z(2) ], [ 0*Z(2), Z(2)^0 ] ] ) gap> u:= Subspace( v, [ [ 1, 1 ] * Z(2) ] ); Subspace( v, [ [ Z(2)^0, Z(2)^0 ] ] ) gap> f:= v / u; v / [ [ Z(2)^0, Z(2)^0 ] ] gap> IsSpaceCoset( u ); false gap> IsSpaceCoset( Random( f ) ); true

:`W1`=`W2`-

returns`true`

if for the two quotient spaces

and`W1`=`V1`/`U1`

the equalities`W2`=`V2`/`U2`

and`V1`=`V2`

hold, and`U1`=`U2``false`

otherwise.

:`W1`<`W2`-

returns`true`

if for the two quotient spaces

and`W1`=`V1`/`U1`

either`W2`=`V2`/`U2`

or`U1`<`U2`

and`U1`=`U2`

hold, and`V1`<`V2``false`

otherwise.

Computations in quotient spaces usually delegate the work to computations in numerator and denominator.

The following functions are overlaid in the operations record for quotient spaces.

The **set theoretic** functions

`Closure`

, `Elements`

, `IsSubset`

, `Intersection`

,

and the **vector space** functions

`Base(`

:`V`)-

returns the vectors of the canonical basis of`V`,

`Generators(`

:`V`)-

returns a list of cosets that generate`V`,

`CanonicalBasis(`

:`V`)-

returns the canonical basis of*<V> = W / U*, this is derived from the canonical basis of*W*.

`SemiEchelonBasis( `

`V` )

`SemiEchelonBasis(`

:`V`,`vectors`)-

return a semi-echelonized basis of the quotient space`V`.`vectors`can be a list of elements of`V`, or of representatives.

`Basis( `

`V` )

`Basis(`

:`V`,`vectors`)-

return a basis of the quotient space`V`.`vectors`can be a list of elements of`V`, or of representatives.

In addition to the record components described in Vector Space Records the following components must be present in a row space record.

`isRowSpace`

:-

is always`true`

,

`operations`

:-

the record`RowSpaceOps`

.

Depending on the calculations in that the row space was involved, it may
have lots of optional components, such as `basis`

, `dimension`

, `size`

.

A vector space basis is a record with at least the following components.

`isBasis`

:-

always`true`

,

`vectors`

:-

the list of basis vectors,

`structure`

:-

the underlying vector space,

`operations`

:-

a record that contains the functions for the basis, at least`Coefficients`

,`Print`

, and`SiftedVector`

. Of course these functions depend on the special properties of the basis, so different basis types have different operations record.

Depending on the type of the basis, the basis record additionally contains
some components that are assumed and used by the functions in the
`operations`

record.

For arbitrary bases these are `semiEchelonBasis`

and `basechange`

,
for semi-echelonized bases these are the lists `heads`

and `ishead`

.
Furthermore, the booleans `isSemiEchelonBasis`

and `isCanonicalBasis`

may
be present.

The operations records for the supported bases are

`BasisRowSpaceOps`

:-

for arbitrary bases,

`CanonicalBasisRowSpaceOps`

:-

for the canonical basis of a space,

`SemiEchelonBasisRowSpaceOps`

:-

for semi-echelonized bases.

A row space coset *v + U* is a record with at least the following components.

`isDomain`

:-

always`true`

,

`isRowSpaceCoset`

:-

always`true`

,

`isSpaceCoset`

:-

always`true`

,

`factorDen`

:-

the row space*U*if the coset is an element of*V / U*for a space*V*,

`representative`

:-

one element of the coset,**note**that equal cosets need not have equal representatives (see Row Space Cosets),

`operations`

:-

the record`SpaceCosetRowSpaceOps`

.

A quotient space *V / U* is a record with at least the following components.

`isDomain`

:-

always`true`

,

`isRowSpace`

:-

always`true`

,

`isFactorSpace`

:-

always`true`

,

`field`

:-

the coefficients field,

`factorNum`

:-

the row space*V*(the numerator),

`factorDen`

:-

the row space*U*(the denominator),

`operations`

:-

the record`FactorRowSpaceOps`

.

gap3-jm

11 Mar 2019