Tilburg University
Orthogonality and Conjugation for the Idempotents of Constacyclic Codes
van Zanten, A.J.
Publication date:
2016
Document Version
Publisher's PDF, also known as Version of record
Link to publication in Tilburg University Research Portal
Citation for published version (APA):
van Zanten, A. J. (2016). Orthogonality and Conjugation for the Idempotents of Constacyclic Codes. Tilburg University.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
1
Orthogonality and Conjugation
for the Idempotents
2
Abstract
3
Contents
1. Introduction p. 4
2. Orthogonality relations for idempotents p. 4
3. Conjugated constacyclotomic cosets p. 18
4. Conjugated irreducible polynomials p. 22
5. Conjugated constacyclonomials p. 26
6. Constacyclotomic cosets p. 33
7. Constacyclotomic (di)graphs p. 36
4
1. Introduction
This report is a continuation of the study of the generating and irreducible polynomials of
constacyclonomials of constacyclic codes as described in [1, 2, 3]. Codes generated by polynomials which divide
x
n
,
GF q
( )*
, are called (
-) constacyclic codes. This family of codes generalizes the well-known class of cyclic codes when
1. Let n ( )( )s s S
x
P x
be the decomposition ofx
n
into monic irreducible polynomials overGF q
( )
,( , ) 1
n q
. The code f xs( ) withf x
s( ) : (
x
n
) /
P
s( )( )
x
is called a minimal or irreducible constacyclic code, contrary to the code Ps( ) ( )x which is called a maximal constacylic code. The idempotent generating polynomials of the minimal constacyclic codes are denoted by
s( )x and those of the maximal constacyclic codes by
s( )x . The polynomials
s( )x are occasionally called primitive idempotent polynomials. For more introductory remarks and references we refer to the introductions in [1, 2, 3].2. Orthogonality relations for idempotents
In order to keep our notation as simple as possible we shall drop the fixed parameters n and q from all variable names (like we did on earlier occasions), and write
T
:
T
n q, ,,S
:
S
n q, ,,, , : n q t t C C , : n q, t t C C , ( ) : n q, , ( ) t t e x e x , ( ) : n q, , ( ) s s c x c x , etc. From [1,2,3] we have the following factorization of
x
n
into monic irreducible polynomials over5
: q( )
k ord
, satisfying
n
. If
i is a zero of the irreducible polynomial Pt( )( )x , then
(
i q)
j for somej
is a zero of the same polynomial . It follows that we can take forT
a subset of{0.1,....,
n
1}
, such that the index t of each polynomial Pt( )( )x
corresponds to one of its zeros
t. Usually, we take the minimal value possible. More precisely, from [2, Theorem 6] it follows that if
t is a zero of Pt ( )x
, then all zeros of that polynomial can be written as
c, where c runs through the constacyclotomic coset Ct,t
{0,1,....,
n
1}
, (cf. [3, Definition 2]). We shall write the polynomial Pt( ) ( )x explicitly as( )
( )
t ,1 t 1....
, t m m t t t mP
x
x
p x
p
. (2)From the lines above it follows that the degree mt of Pt ( )x
is equal to the size of Ct. As for the set
T
, we proved (cf. [3, Corollary 19]) that its size is equal to the size ofS
: {
s
∣cs( )x is an element of a maximal independent set of constacyclonomials}. For reasons of convenience we therefore introduce the variable
n
0:
∣T
∣= ∣S
∣. (3)Since
k,k
:
ordq( )
, an alternative expression for the n zeros of nx
is
ik1,0
i
n
1
(cf. [1, Theorem 3 (v)]). In the next we shall occasionally need the zeros ofx
n
j,0
j
k
1
. Therefore, we sometimes shall use the alternative representation of the zeros, since it makes dealing with them easier (cf. the remarks right after Theorem 13 and Example 10 in [2]). Actually, all zeros of the polynomialx
kn
1
are covered in this way at the same time. By doing so, the setT
will be a subset of{0,1,....,
kn
1}
.In [2,3] the notion of constacyclonomial was introduced. For our convenience we repeat the most relevant facts. For the details we refer the reader to [2,3]. The polynomial
csn q, , ( )x xs xsq .... xsqms 1 mod
x
n
is called a constacyclonomial if n q, , ( )q n q, , ( ) s sc x c x and if it is not the zeropolynomial. This polynomial can be written as a polynomial in
GF q x
( )[ ]
n q, , ( ) d0 c0 d1 c1 .... dms1 cms1
s
c x
x
x
x ,with ciCsn q, , 0 i ms, and
d
0
0
. A necessary and sufficient condition for csn q, ,( )x to be a constacyclonomial isx
sqmsx
s
, or equivalentlys q
(
ms1)
0
modkn
, wherek
ordq( )
.Since
q
for all elements
GF q
( )
, the polynomial
c
sn q, ,( )
x
also satisfies the definition requirement for constacyclonomials, i.e. (
csn q, ,( ))x q
csn q, ,( )x . Therefore, from now on we shall call all polynomials
csn q, ,( )x satisfyingx
sqmsx
sq
modx
n
constacyclonomials. The special case
1
could be dubbed as a monic constacyclonomial , similarly to monic (irreducible)polynomials. We shall often identify the linearly dependent constacyclonomials
c
sn q, ,( )
x
and , , ( ) n q s c x
6
We remind the reader that we defined
S
such that if cj is not self adjoint and ifj
S
, then alson
j
S
. For our convenience we therefore introduce the notion of adjoint constacyclonomial. Ifc
j( )
x
is not self adjoint, we define the adjoint constacyclonomialc
j*( )
x
as
c
j*( ) :
x
c
n j( )
x
, (4) while in case thatc
j( )
x
is self adjoint we just takec
j*( ) :
x
c x
j( )
. So, according to this definition, in all cases the adjoint constacyclonomial is again a (monic) constacyclonomial. In [2] we weoccasionally wrote
c
j( )
x
instead ofc
n j( )
x
. This notation will not give rise to confusion as long as we stick to the ring Zn. However, in expressions like
s, one should keep in mind that it stands forn s
, and hence
s s
n(
o 1). Similarly, we define the adjoint irreducible polynomialPt( )*( ) :x P(t 1)( )x
. So, the zeros of
( )*
( ) t
P x are zeros of
x
n
1 and Pt( )* ( )x itself is a factor of this binomial. It is well known that the inverses of the zeros of Pt( )( )x
are the zeros of its monic reciprocal. ButPt( )*( )x
is not always a factor of the monic reciprocal of
x
n
1 which isx
n
. Hence, we can say that the adjoint of( )
( ) t
P x is equal to its monic reciprocal if and only if
1, i.e. if
1. From [2, Theorem 13] it now follows that then
0 constacyclonomialsc
j
,
j
S
, constitute an orthogonal basis of polynomials in the algebra A with respect to the inner product1 0 ( , ) ( ) ( ) n i i i p q p
q
, i.e. : { s s( ) s S A c x
∣cs C , s GF q( )}
, (5) ( *, ) aj , j k j j k c c nm
, (6) where aj: 1 ifc
j( )
x
is not self adjoint and : ( [mj/2] 1) / j
a j q n if it is self adjoint. Occasionally, we shall write
c
j( )
x
for the adjoint constacyclonomial
c
j*( )
x
instead of
c
n j( )
x
,
where
j
stands for the integer in{0,1,....,
n
1}
which is equal modulo n to
j
. We emphasize that without that conditionc
j( )
x
is not a constacyclonomial and not even an element ofR
n q, .Example 1
We compute for each value of
an orthogonal basis for the vector spaceC
6,5, as described above.7
4
4 0( ) 1 c x , c x14( )x1x5, c x24( )x2x4, c x34( )x3.All these constacyclonomials are self adjoint and hence, for all
andj
we havec
j*
c
j. As a couple of illustrations of eq. (6) we compute:5 5 2* 2 1 5 1 5 2 6 10 1 1 2 0 0 ( , ) ( )( ) x i ( 2 ) x i i i c c x x x x x x x
=6.4
24
4
, 5 5 2* 2 2 4 2 4 4 6 8 2 2 2 0 0 ( , ) ( 2 )( 2 ) x i ( 4 4 ) x i i i c c x x x x x x x
=6.4.2
48 3
, 5 5 4* 4 3 3 6 3 3 4 0 0 ( , ) x i x i 6.4 24 4 i i c c x x x
.Applying (6) yields the following outcomes for the above inner products:
2* 2 1
1 1 2
(c ,c ) 6.2.2 244, (c22*,c22)2 6.2.22 483 and (c34*,c34)6.1.41244, which agree with the direct calculations.
We also compute the inner products of c52*( )x x5x1 and c x42( ) with themselves and obtain
2* 2
5 5 2
(c ,c ) 4 and (c42*,c42)2 2. Indeed, eq. (6) provides us with (c52*,c52)2 6.2.25 4 and
2* 2 4
4 4 2
(c ,c ) 6.2.2 2. The last result can also be obtained by realizing that 3c22 c42, and so
2* 2 2 2* 2
4 4 2 2 2
(c ,c ) 3 (c ,c )9.32. □
It follows that each element p
A
, i.e. the vector space consisting of all polynomials inR
n q, which satisfy p x( )q p x( ), (cf. [2, Theorem 13 (i)]) can be developed w.r.t. the basis {cs∣
s
S
}
, so ( ) s s( ) s S p x c x
, ( *, ) / as s cs p nms
,s
S
, (7)where we used
m
s
m
n s
m
s in the case that cs( )x is not self adjoint. In particular, we can consider p x( )
t( )x ,t
T
, a so-called primitive idempotent polynomial which generates the constacyclic code ( ) /{ } ( ) u u T t P x
of length n overGF q
( )
, which we call a minimal orirreducible code, thus generalizing that notion as known in the theory of cyclic codes. This idempotent
satisfies
t( i)1 if
i (or
ki1) is a zero of Pt( ) ( )x , while
t( i)0 if
i is not a zero of Pt( )( )x
, for
0 i
n
. Another special case is the idempotent t ( )x 1 t ( )x
which generates the maximal code Pt( ) ( )x .
Theorem 1
Let
C
be the constacyclic code of length n generated by some polynomial inR
nq, and let, ,
( )
n qe x
C
be its idempotent generator. Let furthermore {cs( )x 8
of constacyclonomials in ,
q n
R
for the chosen values of n and q.(i) There exist coefficients
s
GF q
( )
such that ( ) s s ( ) s Se x c x
and they are uniquely determined by the relation ( *, ) / ass cs e nms
.(ii) The coefficients
st in the expression for the primitive idempotent generator t ( ) st s( ) s S x c x
,t
T
, are equal to 's/
's s t t s a t tm
p
m
n
, where pt's is the one but highest coefficient of xin the irreducible polynomial Pt's( )x which is a factor ofx
n
s while mtn',s is its degree, and where'
( , )
t
t k s
if cs is self adjoint, whereast
'
t k
( ,
s
)(
t k n s
( ,
))
if cs is not self adjoint.(iii) The coefficients
st in (ii) can also be written as s / s s t t s a st st m p m n
, where the index
st
hasto be taken with respect to the zero
s ofx
n
s.(iv) The inproduct of a pair of primitive idempotents
t and
u is equal to(
t,
u)
m
t t u,
. (v) If t ( ) st s( ) s S x c x
and u ( ) su s( ) s S x c x
are two primitive idempotent polynomials, then s st us t t u, s S n w m
, with as s s w m
.(vi) If s and r are both elements of
S
, then s 1 st rt s, rt T t nw m
. Proof(i) This follows immediately from eq. (7). (ii) We know that
t( i) is equal to 1 if
i is a zero of the irreducible polynomial Pt( )x , whileit is equal to 0 otherwise. Moreover, cs(
i)(
i s) (
i sq) .... is constant fori
-values from the same constacyclotomic coset, since ifi
and i' are both from Ctn q, ,
, then there exists an exponent
j
such that
i'
(
i q)
j . Hence, 1 * * * 0 ( , ) ( ) ( ) ( ) n i i t s t s t t s i c
c
m c
. Firstly, weassume that cs( )x is self adjoint, and so cs*( )x cs( )x . Applying (7) and [2, Theorem 13 (v)], we can write (cs*,
t)= m m pt s tn', s /mtn', s , witht
' :
t k s
( , )
. So, ', s / ', s s t t n n s a t t m p m n
, with [ /2] ( ms 1) / 2 sa s q . Next, we assume that cs( )x is not self adjoint, and so cs*( )x cn s ( )x . In this case we find by applying [2, Theorem 13 (v)] that (cs*,
t) m mt n s ptn',s /mtn',s, with'
( ,
)
t
t k n s
. Since mn s ms, it follows that ', s / ', s n s t t n n s a t t m p m n
, with an s 1, since n sc is not self adjoint, just like cs. (iii) The s-power of any zero of Pt( ) ( )x is a zero of the irreducible polynomial Pst(s)( )x which is a
divisor of
x
n
s. The result now follows immediately from (ii). (iv) We have 1 , 0 ( , ) ( ) ( ) n i i t u t u t t u i m
9
(v) Substituting the given expressions in the inproduct yields ' '
, ' ( t , u ) st su( s, s) s s c c
= ( , ) as t u t u s s s s s s s s s c c n m
, and the relation follows by applying (iv). (vi) By using (4) and (6), we can write1 , 0
(
)
(
)
(
)
(
)
s n a t t i i t s r s r s s r i t Tm c
c
c
c
nm
. Substitution in the lhs of this expression ( t) as t
t s s s m c
nm
and ( t) ar t t r r r m c
nm
yields , r a t t r s r s r t T tnm
m
, and so the relation (v) follows. □Example 2
For
n
6
andq
5
we have the following factorizations into irreducible polynomials and next into factorsx
i where
is a zero ofx
2
2
x
1
which is of order 24:6 2 2
1 (
1)(
1)(
1)(
1)
x
x
x
x
x
x
x
=(
x
0)(
x
4)(
x
20)(
x
8)(
x
16)(
x
12)
; 6 2 2 2 1 5 9 21 13 172
(
2)(
2)(
2)
(
)(
)(
)(
)(
)(
)
x
x
x
x
x
x
x
x
x
x
x
x
; 6 2 2 2 3 15 7 11 19 233 (
2)(
2
2)(
2
2)
(
)(
)(
)(
)(
)(
);
x
x
x
x
x
x
x
x
x
x
x
x
6 2 2 2 10 6 14 22 184
(
2
1)(
2)(
2
1)(
2)
(
)(
)(
)(
)(
)(
).
x
x
x
x
x
x
x
x
x
x
x
x
x
In [2, Theorem 5 (iii)] we stated that for any
j
,0
j
k
1
, the n zeros ofx
n
j can be written as
j ik ,0
i
n
1
, where
is a zero of orderkn
. By defining
:
k, one can equally well write the n zeros as
j i,0
i
n
1
. The following example provides us with the primitive idempotents of the cyclic codes generated by the irreducible factors ofx
6
1
.1
,k
1
,kn
6
Let
be a primitive6
th root of unity, e.g. a zero ofx
2
x
2
. Then it follows from the relations atthe bottom of the previous page (with
4) that x6 1 P0(1)( )x P1(1)( )x P2(1)( )x P3(1)( )x , with(1) 0 0 ( ) 1 P x x x
, P1(1)( )x x2 x 1 (x
1)(x
5), (1) 2 2 4 2 ( ) 1 ( )( ) P x x x x
x
, P3(1)( )x x 1 x
3.We define
S
1
{0,1, 2,3}
andT
1
{0,1, 2,3}
. All (consta)cyclonomials are self adjoint, and since1
k , we have ds (kn s, ) / ( , ) 1n s for all
s
S
6,5,1. In this case the expression for the coefficients
st in
t1( )x simplifies to',1 ',1 n t t t s n t
m p
n m
, which reduces to 1 st t stp
m
m
inGF
(5)
. Applying this to
01( )x gives1 0 0 0 0 0 0 1 2 3 0 0
p
m
m
1. 1
= 1, and hence, using the expression for (consta)cyclonomials in Theorem 1 (ii) , we obtain
1 1 2 3 4 5
0( )x 1 x x x x x
10 Similarly, we find for
11( )x the coefficients1 1 0 0 1 0
2. 1
2
p
m
m
, 1 1 1 1 1 11
2
1
2
p
m
m
, 1 1 2 2 1 2p
m
m
=2
1
2
= 1, 1 1 3 3 1 3p
m
m
=2
1
2
1
. So,
11( )x 2 x1 x22x3x4x5.In the same way, omitting the explicit calculations, we obtain for
21( )x its coefficients
02 2,2
1 1
,
22 1,
32 2, and so
21( )x 2 x1 x22x3x4x5, while for
31( )x we get3
0 1
,
13 2,
23 2,
33 1, and so
31( )x 1 x1 x2x3x4x5. As for these and similar calculations one should keep in mind that products like st have to be carried in the ringZ
6. The above expressions for
t1( )x satisfy the idempotent relation1 2 1
( ) ( )
t x t x
, and turn out to coincide with the expressions obtained by applying the general formula for idempotent generating polynomialse x
( )
(
n
)
1xh x g x
'( ) ( )
(cf. [1,2]). E.g. by taking h x( )P1(1)( )x x2 x 1 and6 2 4 3
( )
(
1) /
1
1
g x
x
x
x
x
x
x
, we get againxh x g x
'( ) ( )
x
5
x
42
x
3
x
2x
2
. Instead of writing the zeros as
i,0
i
5
, one can also represent them by
i,0
i
5
, where
is some zero ofx
6
1
and
some primitive6
th root of unity. E.g. one could take
, in which case the indexation of the irreducible polynomials is: P5(1)( )x x 1, P0(1)( )x x2 x 1,(1)
1 ( )
P x
x
2
x
1
, P2(1)( )x x 1. Actually, this last way of representing the zeros is a special case of the convention of taking
k, with
a zero of orderkn
. □Next, we shall try to do the same for
1
and the same values for n and q.
Example 3
Take
1
, hencek
2
andkn
12
. Let
(
2)
be a zero of order 12, and choose
2. Then we have the following factorizationinto irreducible polynomials x6 1 P0( 1) ( )x P1( 1) ( )x P3( 1) ( )x P4( 1) ( )x , with
( 1) 2 1 5 0 ( ) 2 1 ( )( ) P x x x x
x
(
x
0)(
x
2)
, ( 1 3 1 1 ( ) 2 P x x x
x
, ( 1) 2 3 5 3 ( ) 2 1 ( )( ) P x x x x
x
, ( 1) 4 4 ( ) 2 P x x x
.Here, we defined
S
1
{0,1, 2,3}
andT
1
{0,1,3, 4}
. However, when applying the expression for s
in Theorem 1 (ii), also polynomials Pst(1)( )x will play a role having as zeros the s-powers of thezeros of Pt( 1) ( )x . Therefore, the index st does not have its usual meaning with respect to
, but to s11
a subset of
[0,
kn
1]
. To this end, one has to reformulate a couple of properties of the constacyclonomials cs( )x
in terms of
-powers. □Theorem 2
Let
be a zero ofx
n
of order kn.(i) For the inner product
( , )
p q
,p q
,
C
n q, ,, one can write1 1 1 0 ( , ) ( ) ( ) n ik ik i p q p
q
. Moregenerally, for
p q
,
C
n q, ,j, one can write1 0 ( , ) s ( ) ( ) n s ik s ik i p q p
q
.(ii) The cyclonomials
c c
j,
kC
n q, ,
satisfy the orthogonality relation ( *, ) aj ,
j k j j k
c c nm
.(iii) For any
s
0
and for any j, one has1 0 ( ) 0 n j ik s i c
, while 1 0 0 ( ) n j ik i c
n
.(iv) For any
s
S
, for anyi
[0,
n
1]
and for anyj
[0,
k
1]
, one has(
)
.
js js j ik is s s is
p
c
m
m
(v) Let
e x
( )
be some polynomial inC
n q, ,, then one can write( )
s s( )
s S
e x
c
x
with( )
sGF q
and ( *, ) / as s cs e nms
.(vi) If
t( )x ,t
T
, is the primitive idempotent generator of the constacyclic code( ) \ ( ) i i T t P x
, then t ( ) st s( ) s S x c x
with js js s t t st s a st m p n m
. ProofAll relations follow from [3, Theorem 5 (v), (ix), (x), (xi)] and from Theorem 1. □
Example 4
Again
n
6
,q
5
,
1
and sok
2
,kn
12
. We now shall determine the idempotents1
( )
t x
,
t
T
1, by applying Theorem 2 (iv). Again we take for a primitive12
th root of unity
a zero ofx
2
2
x
1
, which gives rise to the following factorization of the irreducible polynomials:12
We now apply for the above parameter values the expression
1 ( 1) ( 1)
( 1)
s s s s s t t st t st s a a st stm
p
m
p
n
m
m
,where we omitted the *-symbol, since all irreducible polynomials are self adjoint.
13
2
s
,m
2
2
a
2
2
, 2911
1
11
,3
s
,m
3
1
a
3
1
, 391
2
2
1 1
, 1 1 5 2 4 3 9 ( )x 1 2(x x ) (x x ) 2x
.In order to check the correctness of these results, we also compute the primitive idempotents by the general equations (cf. [1, Theorem 5])
t1( )x 1
t1( )x ,
t1( )x (n
)1xh x g x'( ) ( ), g x( )Pt( 1) ( )x . We find the following polynomials:( 1) 2 1 ( ) 2 1 P x x x ,
h x
( )
x
4
2
x
3
2
x
1
,h x
'( )
4
x
3
x
22
, 1 1 ( )x
2 3 4 52 2
x
x
x
x
2
x
, ( 1) 3 ( ) 2 P x x ,h x
( )
x
5
2
x
4
x
32
x
2
x
2
,h x
'( )
2
x
3
2
x
2
x
1
, 1 2 3 4 5 3 ( )x 1 2x x 2x x 2x
, ( 1) 2 7 ( ) 2 1 P x x x ,h x
( )
x
4
2
x
3
2
x
1
,h x
'( )
x
3x
22
, 1 2 3 4 5 7 ( )x 2 2x x x x 2x
, ( 1) 9 ( ) 2 P x x ,h x
( )
x
5
2
x
4
x
32
x
2
x
2
,h x
'( )
2
x
3
2
x
2
x
1
, 1 2 3 4 5 9 ( )x 1 2x x 2x x 2x
.So, there is fully agreement between the two methods in this case.
The matrix which has as columns the representing vectors with respect to the basis of negacyclonomials cs1( )x ,
s
{0,1, 2,3}
t 1 3 7 9 mt1 2 1 2 1 sm
s ws 2
1
2
1
2
2
2
2
1
1
1
1
1
2
1
2
0 1 2 3 1 2 2 1 1 3 2 1 In order to check Theorem 1 (iv) and (v), we determine the weight factors : as14 7 9 s s s s w
1.2.1 ( 2).2.2 2.1.( 1) ( 1).( 1).( 2)
0
, 9 9 s s s s w
2 2 2 21.1
( 2)2
2.( 1)
( 1)( 2)
1
.Both results satisfy the relation of Theorem 1 (iv). Notice that m911. We also present a few examples to illustrate Theorem 1 (v):
0 1 1 1 3.2( 2) 1.1.( 2) 3.2.2 1.1.2 0 t t t mt
, 0 2 1 1 3.2.1 1.1.( 1) 3.2.1 1.1.( 1) 10 0 t t t mt
. 1 1 1 1 1 1 1 3.( 2)( 2) 1.( 2)( 2) 3.2.2 1.2.2 32 2 3 t t t mt nw
, 3 3 1 3 1 1 1 3.1,1 1.2.2 3.( 1)( 1) 1.( 2).( 2) 14 1 1 t t t mt nw
. □ Example 5Take
n
6
,q
5
,
2
, hencek
4
andkn
24
. We choose as primitive 24th root of unity again a zero ofx
2
x
2
, which yields the followingindexing of the irreducible polynomials contained in
x
24
1
:(1) 0 ( ) 1 P x x , P4(1)( )x x2 x 1, P8(1)( )x x2 x 1, P12(1)( )x x 1, (2) 2 1 ( ) 2 P x x x , P9(2)( )x x22, P13(2)( )x x2 x 2, (3) 2 3 ( ) 2 P x x , P7(3)( )x x2 2x2, P19(3)( )x x2 2x2, (4) 2 2 ( ) 2 1 P x x x , P6(4)( )x x 2, P14(4)( )x x22x1, P18(4)( )x x 2. As for the set
T
2, we defineT
2
{1,9,13}
.Next, we shall compute the primitive idempotent polynomials
15
1
s
,m
1
2
,a
1
1
, 2 2 1 1 1 1 1 2 12 1
3
2
2 2
m p
m
,2
s
,m
2
2
,a
2
2
, 2 4 1 1 2 2 2 4 22 2
3
2
4 2
m p
m
, 2 2 4 5 1 ( )x 2 2x 2x x 2x
.The equations
12( )x 1
12( )x ,
12( )x (6.2)1xh x g x'( ) ( ),g x
( )
x
2
x
2
and4 3 2
( )
2
1
h x
x
x
x
x
give the same result. Similarly, we obtain:9
t
, 2 1 9 9 0 0 0 1 02 1
2
2
1 1
m p
m
, 2 2 9 9 9 1 1 2 92 0
0
2
2 2
m p
m
, 2 4 9 9 18 2 2 4 182 2
1
2
4 1
m p
m
, 2 2 4 9( )x 2 x 2x
.13
t
, 2 1 13 13 0 0 0 1 02 1
2
2
1 1
m
p
m
, 2 2 13 13 13 1 1 2 132 1
2
2
2 2
m
p
m
, 2 4 13 13 2 2 2 4 22 2
2
2
4 2
m
p
m
,
132( )x 2 2x2x2x42x5.We verified the relation
i2( )x 2
i2( )x for all three values of i.Finally, we present the three primitive idempotents as the colunns of a
3 3
-marix: t 1 9 13 sm
sw
s
2 2 2 2 0 2 2 1 2 0
1
2
1
2
2
1
4
3
One can easily check that all relations of Theorem (iv) and (v) are satisfied in this case, e.g.
3 3 2 3 1 1 1 3.( 2).( 2) 3.( 1).( 1) 3.( 2).( 2) 27 2 3 t t t mt nw
. □In order to put the concept of the matrix with orthogonal columns and rows of the previous examples in a general form, we now define
, : as t s / s
s t n s m pt st mst
,s
S
,t
T
, (8)16
Definition 3
The matrix Mn q, ,, shortly M , is defined as the
n
0
n
0-matrix with elements s t, stt n M m
,(
/
as)
s tm
t
and the
-adjoint matrixM
* as the n0n0-matrix with elementsM
t s*,
w
s
ts, for alls
S
andt
T
.Theorem 4
The matrix
M
has the following properties.(i) The matrix
M
* is the inverse of the matrixM
, i.e. u s*, s t, u t,s S
M M
and s t, t r*, s r, t TM M
.(ii) If e x( )Cn q, , is the idempotent generator of the code
1 2 ( ) ( ) ( )
( )
( )....
( )
l i i iP
x P
x
P
x
and if one writes , ,( )
( )
n q s s s Se x
c x
, then the coefficients
s are uniquely determined by the matrix equationM
where
is the column vector of length no with zeros on the positions i1, i2,...., il and oneselsewhere, while
is the column vector 00 1 2
( ,
,....,
)
T( )
n nGF q
. Proof(i) This is nothing else as the matrix form of the orthogonality properties of Theorem 1 (v) and (vi). By applying these properties, we derive u s*, s t, s us st
s s t n M M w m
, t u s s s t u s tn
w
m
and * , , , 1 t t t t s t t r s r r r s r s r t t t t t n M M w nw m
m
. (ii) The columns ofM
are the representation vectors
stof the primitive idempotents
t( )x for all, ,
n q
t
T
. So, when taking
: t (0,....,1,...., 0) with all elements equal to 0 except the element on positiont
we obtain precisely the vector
st. The statement in (ii) now follows immediately. □Remarks
As for the notation in Theorem 3 and in its proof, one should keep in mind that the sets S and
T
need not always be identical, and even that it is not always possible to choose their elements such that they become identical. E.g. if sS and sT for some value of s, one has to take fort
in
s t, theinteger t'T with tCtn q', . The conclusion in the above proof that the matrix
nM
* is a right inverse ofM
can also be drawnfrom Theorem 1 (iv). As for the variables
s t, introduced in (8), we can write
s t, g pst sts, where: / s
t
s t st
g m m is equal to the quotient of the sizes of the constacyclotomic cosets Ct and Csts, or equivalently of the degrees of the irreducible polynomial
P
t( )( )
x
and of Pst(s)( )x which has as zeros the s-powers of the zeros of Pt( ) ( )x . Consequently, the value of
s t, is equal to the sum of the s-powers of the zeros of
P
t( )( )
x
. In general we have , as ts t
n
s17 ms
s t, mt
t s, , [4, Th. 11 (i)] s r, r t, s, t r Tn
, [4, Th. 11 (ii)] s s t, s u, t t, u s Sm
nm
, [4, Th. 11(iii)] t 1 s t, r t, s 1 s, r t Tm
nm
. [4, Th. 11 (iv)] Actually, Ctn q, ,1Ctn q,1 which is not identical to Ctn q, , but since this identity holds for all tT we areentitled to replace (m1t)1 by mt1 in the derivations we made and also in Definition 3. Notice that in the case
1 the setsS
andT
can be chosen identical. The last two relations are special forms of the general orthogonality relations in terms of the
-variables as follows from Theorem 1 (v) and (vi) 1 , , , s s s t s u t u a s S t m n m
, 1 , , , s s s t r t s r a t T t m n m
.As an alternative for the matrix
M
in Definition 3, we now introduce the matrix
with entries,
:
,s t
s t
. According to the above formulas,
is an orthogonal matrix with weight factorss s a m
and 1 tm for the orthogonality of respectively its rows and its columns.
The orthogonality relations for the primitive idempotents represented by the column vectors
t,t
T
, with components
st,s
S
, or equivalently, by the column vectors
t, show a certain similarity with the orthogonality relations of the characters
sjof the irreducible representations( j
D G) of a finite group G. Within this context cyclotonomials cs( )x are to be considered as the counterparts of the classes Cs of conjugated elements in the group. The irreducible polynomials
( )
( ) t
P x ,
t
T
, which take care of the labelling of the idempotents, seem to have no obvious counterpart in character theory. Usually, an irreducible character (and the corresponding matrix representation) is indicated by some ad hoc parameterj
or by its dimension
j(1)
. Since1
0,
(
0 0)
t
t
g p
is equal to the sum of the zero-powers of the zeros of Pt( ) ( )x , we may conclude that0,t
m
t
. So, the row of
with label 0 contains the ‘sizes’ mt of the irreducible polynomials, just like the row in a character table labelled by 1 (from the class C1 (1)) contains the ‘sizes’ of theirreducible representations
j(1). For
1 the resemblance between idempotent tables and character tables is even stronger. Since0 1
,0
:
0s
g p
s
is equal to the sum of the s-powers of the zeros ofP
0(1)( )
x
x
1
,
s,0 1 for alls
S
. So, the idempotent column vector
0 can be seen as the counterpart of the trivial character1
, in the literature also denoted by 1G. Furthermore, a cyclonomial is self adjoint if and only if, ,
( ) ( )
n q n q
s n s
18
definition of adjoint irreducible polynomial it follows immediately that Pt(1)( )x is self adjoint if and
only if Ctn q, Cn tn q, . For
1, the index setsS
n q, ,1 andT
n q, ,1 can be chosen identical. So,in the case1
, the number of self adjoint cyclonomials and the number of self adjoint irreducible polynomials, and hence also the number of self adjoint primitive idempotent generators, are equal for fixed values of nandq
. This equality resembles the property that the number of real characters, i.e. characters of self adjoint or self conjugated irreducible representations over ₵. is equal to the number of self inverse classes of conjugated elements in the group. We can also prove this property by exploiting the orthogonality relations like is done in character theory. From the first orthogonalty relation for
1it follows that a column
t is self adjoint if and only if 1 s s t, s t, 1 st m
nm
, while the rhs is equal to 0 if
t is not self adjoint. Hence, the total number of self adjoint columns of
is equal to, , , 1 s s t s t s t t m
n
m
. Similarly, the second orthogonality relation gives that a row
s of
is self adjoint if and only if s 1 . , 1s t s t
t t
m
n
m
, and if not then the rhs equals 0. Hence, the total number of self adjoint rows of
also equals , ,, 1 s s t s t s t t m n
m
.3. Conjugated constacyclotomic cosets
In this section we shall introduce the notion of conjugated constacyclotomic coset with respect to certain transformations. Unlike as in [1-4], in this report we shall consider cyclotomic and
constacyclotomic cosets as ordered sets. Furthermore, we study sets of integers mod n of the form
, ,
n q t
aC
r
. It appears that for special values of r, such a set is again a constacyclotomic coset, which of course has to be labelled as Cat rn q, ,, since atr is an element of it. In the next we shall describe this process as the mapping Ctn q, , Cat r induced by aCtn q, , r, or shortly by, , , ,
n q n q
t at r
C C . Like on earlier occasions, we also shall drop the parameters n and
q
.Theorem 5 Let 0 1 1 ( , ,...., ) t t m
C c c c ,
c
0
t
, be some constacyclotomic coset, and let the notation aCt rstand for the ordered set 0 1
1
( , ,...., )
t m
ac r ac r ac r . (i) If a is a fixed integer and if there is an integer
r
0 such thata
1
kr
0
0
modn
/ ( , )
l n
, thenthere are
( , )
l n
elementskr
Z
n such thata
1
kr
0
modn
/ ( , )
l n
, and they can be written as0
/ ( , )
kr
kr
jn
l n
,0
j
( , )
l n
. For all r- values satisfying the above relation one has aCtr*
t at r atm
C
m
, where the factor tat
m
m
denotes the number of times that each element of Cat r
occurs in
the right hand side.
(ii) For any
r
jn
/ (
q
1, )
n
,0
j
(
q
1, )
n
, the mapping Ct Ct r defines a permutation onthe family of constacyclotomic cosets