Quantum Convolutional BCH Codes

Salah A. Aly1 , Markus Grassl2 , Andreas Klappenecker1, Martin Rötteler3 , Pradeep Kiran Sarvepalli1
1
Department of Computer Science, Texas A&M University, College Station, TX 77843-3112, USA
2
Institut für Algorithmen und Kognitive Systeme, Universität Karlsruhe (TH), D-76128 Karlsruhe, Germany
3
NEC Laboratories America, Inc., 4 Independence Way, Suite 200, Princeton, NJ 08540, USA

Abstract— Quantum convolutional codes can be used to protect q elements. A convolutional code C of length n and dimension
a sequence of qubits of arbitrary length against decoherence. k over Fq is a free module of rank k that is a direct summand
We introduce two new families of quantum convolutional codes. of Fq [D]n . A matrix G in Fq [D]k×n such that C = im G =
arXiv:quant-ph/0703113v2 9 Apr 2007

Our construction is based on an algebraic method which allows

to construct classical convolutional codes from block codes, {uG | u ∈ Fq [D]k } is called a basic generator matrix of C,
in particular BCH codes. These codes have the property that and a matrix H ∈ Fq [D](n−k)×n such that C = ker H t =
they contain their Euclidean, respectively Hermitian, dual codes. {v | v ∈ Fq [D]n , vH t = 0} is called a basic parity check
Hence, they can be used to define quantum convolutional codes matrix of C.
by the stabilizer code construction. We compute BCH-like bounds The existence of a convolutional code C is equivalent
on the free distances which can be controlled as in the case of
block codes, and establish that the codes have non-catastrophic to the existence of four matrices G ∈ Fq [D]k×n , H ∈
encoders. Fq [D](n−k)×n , K ∈ Fq [D]n×k , and L ∈ Fq [D]n×(n−k) such
that C = im G = ker H t , GK = 1Fq [D]k , and Lt H t =
I. I NTRODUCTION 1Fq [D]n−k = HL.
Quantum convolutional codes provide an alternative to Let νi denote the maximum of the degrees among the
quantum block codes to protect quantum information for polynomials in the ith row of a basic generator matrix G,
reliable quantum communication. Ollivier and Tillich launched and let the memory m be the maximal value of νi . A basic
the stabilizer framework for quantum convolutional codes [11]. generator matrix of a convolutional code C is called reduced
Using this stabilizer framework Forney et al. constructed rate if the overall constraint length ν = ν1 + · · · + νk has the
(n−2)/n quantum convolutional codes [3]. Recently, two of us smallest value among all basic generator matrices of C. It
constructed quantum convolutional codes from product codes is often convenient to express the generator matrix as G =
[5] and derived an algorithm to construct non-catastrophic G0 + G1 D + · · · + Gm Dm , where Gi ∈ Fqk×n .
encoders and encoder inverses [6]. In [1], a generalized Let Fq ((D)) be the field of LaurentP series consisting of
i
Singleton bound for a class of quantum convolutional codes elements of the form v(D) = i vi D for vi ∈ Fq and
has been established, together with a family of codes based vi = 0 for i ≤ r for some r ∈ Z. We associate with a con-
on generalized Reed-Solomon codes meeting this bound. volutional code C another module C ∞ = {u(D)G | u(D) ∈
Unit memory convolutional codes are an important class Fq ((D))k }, The entries of a generator matrix G of C ∞ might
of codes that appeared in a paper by Lee [10]. He also be rational functions. Let v(D) P = (v1 (D), . . . , vn (D)) ∈
showed that these codes have large free distance df among Fq ((D))n where vi (D) = j vij Dj . Then we can identify
P
other codes (multi-memory) with the same rate. Convolutional v(D) with an element in Fnq ((D)) as j vj Dj , where vj =
codes are often designed heuristically. However, classes of (v1j , . . . , vnj ) P∈ Fnq . We define the weight of v(D) as
unit memory codes were constructed algebraically by Piret wt(v(D)) = i∈Z wt(vi ). A generator matrix G is called
based on Reed-Solomon codes [12] and by Hole based on catastrophic if there exists a u(D) ∈ Fq ((D))k of infinite
BCH codes [8]. In a recent paper, doubly-cyclic convolutional Hamming weight such that u(D)G ∈ C ∞ has finite Hamming
codes are investigated which include codes derived from Reed- weight. The free distance df of C is defined as
Solomon and BCH codes [4]. These codes are related, but not
df = min{wt(v(D)) | v(D) ∈ C, v(D) 6= 0}. (1)
identical to the codes defined in this paper.
The main results of this paper are: (a) a method to construct A rate k/n convolutional code with memory m, overall
convolutional codes from block codes (b) a new class of constraint length ν, and free distance df is denoted by
convolutional stabilizer codes based on BCH codes. These (n, k, ν; m, df )q . Sometimes a shorter notation (n, k, ν)q is
codes have non-catastrophic dual encoders making it possible also used.
to derive non-catastrophic encoders for the quantum convolu-
PThe Euclidean inner Pproduct of two n-tuples u(D) =
i j n
tional codes. u
i i D and v(D)P = j vi D in Fq [D] is defined as
hu(D)|v(D)i = u
i i · vi . The Euclidean dual of a con-
II. BACKGROUND
volutional code C is denoted by C ⊥ = {u(D) ∈ Fq [D]n |
A. Convolutional Codes hu(D)|v(D)i = 0 for all v(D) ∈ C}. Note that H(D), the
We briefly recall the basic facts about classical convolutional parity check matrix of C, does not generate the Euclidean
codes relevant for our discussion. Let Fq be a finite field with dual in general. Instead, one has to reverse the order of
the coefficients of the polynomials in H(D), i.e. consider A. Convolutional Codes from Block Codes
⊥
the matrix Dm H(1/D), where m⊥ is the memory of the Given an [n, k, d]q block code with parity check matrix H, it
code generated by H(D). For codes over FqP 2 , we define the
is possible to split the matrix H into m+1 disjoint submatrices
Hermitian inner product as hu(D)|v(D)ih = i ui ·viq , where Hi , each of which has n columns, such that
ui , vi ∈ Fnq2 and viq = (v1i
q q
, . . . , vni ). The Hermitian dual of  
⊥h H0
C is then C = {u(D) ∈ Fq2 [D]n | hu(D)|v(D)ih =  H1 
0 for all v(D) ∈ C}.  
H =  . . (3)
 .. 
B. Quantum Convolutional Codes Hm
We briefly describe the stabilizer framework for quantum Then we can form the polynomial matrix
convolutional codes, see also [1], [7], [11]. The stabilizer is
e0 + H
G(D) = H e1 D + H
e2 D2 + . . . + H
em Dm , (4)
given by a matrix
where the number of rows of G(D) equals the maximal
S(D) = (X(D)|Z(D)) ∈ Fq [D](n−k)×2n . (2) ei
number κ of rows among the matrices Hi . The matrices H
which satisfies the symplectic orthogonality condition 0 = are obtained from the matrices Hi by adding zero-rows at the
bottom such that the matrix H e i has κ rows in total. Then
X(D)Z(1/D)t − Z(D)X(1/D)t . Let C be a quantum convo-
lutional code defined by a stabilizer matrix as in eq. (2). Then G(D) generates a convolutional code. The fact that the Hi
n is called the frame size, k the number of logical qudits per come from a common block code allows us to characterize
frame, and k/n the rate of C. It can be used to encode a the parameters of the convolutional code and its dual using
sequence of blocks with k qudits in each block (that is, each the techniques of block codes. Our first result concerns a non-
element in the sequence consists of k quantum systems each catastrophic encoder for the code generated by G(D).
of which is q-dimensional) into a sequence of blocks with n Theorem 3: Let C ⊆ Fnq be an [n, k, d]q linear code with
(n−k)×n
qudits. parity check matrix H ∈ Fq . Assume that H is
The memory of the quantum convolutional code is defined partitioned into submatrices H0 , H1 , . . . , Hm as in eq. (3) such
as m = max1≤i≤n−k,1≤j≤n (max(deg Xij (D), deg Zij (D))). that κ = rk H0 and rk Hi ≤ κ for 1 ≤ i ≤ m. Define the
We use the notation [(n, k, m)]q to denote a quantum con- polynomial matrix G(D) as in eq. (4). Then we have:
volutional code with the above parameters. We can identify (a) The matrix G(D) is a reduced basic generator matrix.
S(D) with the generator matrix of a self-orthogonal classical (b) If the code C contains its Euclidean dual C ⊥ , respectively
convolutional code over Fq or Fq2 , which gives us a means its Hermitian dual C ⊥h , then the convolutional code V =
to construct convolutional stabilizer codes. Analogous to the {v(D) = u(D)G(D) | u(D) ∈ Fqn−k [D]} is contained
classical codes we can define the free distance, df and the in its dual V ⊥ , respectively its Hermitian dual V ⊥h .
degree ν, prompting an extended notation [(n, k, m; ν, df )]q . (c) Let df and d⊥ f respectively denote the free distances of V
All the parameters of the quantum convolutional code can and V ⊥ . Let di be the minimum distance of the code Ci =
be related to the associated classical code as the following {v ∈ Fnq | vH e t = 0}, and let d⊥ denote the minimum
i
⊥
propositions will show. For proof and further details see [1]1 . distance of C . Then the free distances are bounded by
Proposition 1: Let (n, (n − k)/2, ν; m)q be a convolutional min{d0 + dm , d} ≤ d⊥ f ≤ d and df ≥ d .
⊥

code such that C ≤ C ⊥ , where the dimension of C ⊥ is Proof: To prove the claim (a), it suffices to show that (i)
νi
given by (n + k)/2. Then an [(n, k, m; ν, df )]q convolutional G(0) has full rank κ, (ii) (coeff(G(D)P ij , D ))1≤i≤κ,1≤j≤n ,
i
stabilizer code exists whose free distance is given by df = has full rank κ, where for f (D) = i≥0 ai D we define
wt(C ⊥ \C), which is said to be pure if df = wt(C ⊥ ). coeff(f (D), Di ) = ai , and (iii) G(D) is non-catastrophic;
Proposition 2: Let C be an (n, (n − k)/2, ν; m)q2 con- cf. [12, Theorem 2.16 and Theorem 2.24].
volutional code such that C ⊆ C ⊥h . Then there exists an By definition, G(0) = H e 0 has rank κ, so (i) is satisfied.
[(n, k, m; ν, df )]q convolutional stabilizer code, where df = Condition (ii) is satisfied, since the rows of H are linearly
wt(C ⊥h \ C). independent; thus, the rows of the highest degree coefficient
matrix are independent as well.
III. A C ONSTRUCTION OF C ONVOLUTIONAL C ODES It remains to prove (iii). Seeking a contradiction, we assume
that the generator matrix G(D) is catastrophic.
P Then there
In this section, we give a method to construct convolutional i
exists an input sequence u(D) = u
i i D ∈ Fq ((D))κ
codes from block codes. This generalizes an earlier construc-
with infinite Hamming weight P that is mapped to an output
tion by Piret [13] to construct convolutional codes from block
sequence v(D) = u(D)G = i vi Di ∈ Fq ((D))n with finite
codes. One benefit of this method is that we can easily bound
Hamming weight, i.e. vi = 0 for all i ≥ i0 . We have
the free distance using the techniques for block codes. Another
benefit is that we can derive non-catastrophic encoders. e 0 + ui+m−1 H
vi+m = ui+m H em,
e 1 + . . . + ui H (5)
1 A small difference exists between the notion of memory defined here and where vi+m ∈ Fnq and uj ∈ Fκq . By construction, the vector
the one used in [1]. spaces generated by the rows of the matrices Hi intersect
trivially. Hence vi = 0 for i ≥ i0 implies that ui−j H ej = 0 every entry in the matrix Hδ,b as a column vector over some
e
for j = 0, . . . , m. The matrix H0 has full rank. This implies Fq -basis of Fqr , and removing any dependent rows. Let
that ui = 0 for i ≥ i0 , contradicting the fact that u(D) has B = {b1 , . . . , br } denote a basis of Fqr over Fq . Suppose
infinite Hamming weight; thus, the claim P (a) holds. that w = (w1 , . . . , wn ) is a vector in Fnqr , then we can
i write wj = wj,1 b1 + · · · + wj,r br for 1 ≤ j ≤ n. Let
P To prove the claim (b), let v(D) = i vi D , w(D) =
i n
i wi D be any two codewords in V ⊆ Fq [D]. Then from w(i) = (w1,i , . . . , wn,i ) be vectors in Fnq with 1 ≤ i ≤ r,
eq. (5), we see that vi and wj are in the rowspan of H i.e. For a vector v in Fnq , we have v · w = 0 if and only if
they are elements of C ⊥ , for any i, j ∈ Z. Since C ⊥ ⊆ C = v · w(i) = 0 for all 1 ≤ i ≤ r.
(C ⊥ )⊥ , it follows that vi · wP j = 0, for any i, j ∈ Z which For a matrix M over Fqr , let exB (M ) denote the matrix
implies that hv(D)|w(D)i = i∈Z vi · wi = 0. Hence V ⊆ that is obtained by expanding each row into r rows over Fq
V ⊥ . Similarly, we can show that if C ⊥h ⊆ C, then V ⊆ V ⊥h . with respect to the basis B, and deleting all but the first rows
For the claim (c),Pwithout loss of generality assume that the that generate the rowspan of the expanded matrix. Then H =
codeword c(D) = ℓi=0 ci Di is in V ⊥ , with c0 6= 0 6= cℓ . exB (Hδ,b ).
It follows that hDi c(D)|Dl Gj (D)i = 0 for i, l ≥ 0, where It is well known that the minimum distance of a BCH code
Gj (D) denotes the jth row of G(D). In particular we have is greater than or equal to its designed distance δ, which is
c0 Hemt
= 0 and cℓ H e 0t = 0. It follows that c0 ∈ Cm and cℓ ∈ very useful in constructing codes [9]. Before we can construct
C0 . If ℓ > 0, then wt(c0 ) ≥ dm and wt(cℓ ) ≥ d0 implying convolutional BCH codes we need the following result on the
wt(c(D)) ≥ d0 + dm . If ℓ = 0, then hDi c0 |Gj (D)i = 0 distance of cyclic codes.
implies c0 H e t = 0 for 0 ≤ i ≤ m, whence c0 H t = 0 and Lemma 4: Let gcd(n, q) = 1 and 2 ≤ α ≤ β < n. Let
i
c0 ∈ C, implying that wt(c0 ) ≥ d. It follows that wt(c(D)) ≥ C ⊆ Fnq be a cyclic code with defining set
min{d0 + dm , d}, giving the lower bound on d⊥ f .
For the upper bound note that if c0 is a codeword of C, Z = {z | z ∈ Cx , α ≤ x ≤ β, x 6≡ 0 mod q}. (6)
then c0 Hit = 0. From c0 we can construct a codeword c(D)
The minimum distance ∆(α, β) of C is lower bounded as
by padding with zeros. Now, hDi c(D)|Dl Gj (D)i = 0 and
hence c(D) ∈ V ⊥ . Since wt(c(D)) = wt(c0 ) we obtain that (
d⊥f ≤ d. P
q + ⌊(β − α + 3)/q⌋ − 2, if β − α ≥ 2q − 3;
∆(α, β) ≥
Finally, let c(D) = i ci Di be a non-zero codeword in V . ⌊(β − α + 3)/2⌋ , otherwise.
We saw earlier in the proof of (b) that every ci is in C ⊥ . Thus
Proof: Our goal is to bound the distance of C using the
df ≥ min{wt(ci ) | ci 6= 0} ≥ d⊥ .
A special case of our claim (a) has been established by a Hartmann-Tzeng bound (for instance, see [9]). Suppose that
different method in [8, Proposition 1]. there exists a such that A = {z, z + 1, . . . , z + a − 2} ⊆ Z.
Suppose further, that there exists b, where gcd(b, q) < a and
IV. C ONVOLUTIONAL BCH C ODES A + jb = {z + jb, z + 1 + jb, . . . , z + a − 2 + jb} ⊆ Z for
One of the attractive features of BCH codes is that they all 0 ≤ j ≤ s. Then by [9, Theorem 4.5.6], the minimum
allow us to design codes with desired distance. There have distance of C is ∆(α, β) ≥ a + s.
been prior approaches to construct convolutional BCH codes, We choose b = q, so that gcd(n, q) = 1 < a is satisfied for
see [8], [14], and most notably [4], where one can control any a > 1. Next we choose A ⊆ Z such that |A| = q − 1
the free distance of the convolutional code. Here we focus and A + jb ⊆ Z for 0 ≤ j ≤ s, with s as large as possible.
on codes with unit memory. Our codes have better distance Now two cases can arise. If β − α + 1 < 2q − 2, then there
parameters as compared to Hole’s construction [8] and are may not always exist a set A such that |A| = q − 1. In this
easier to construct compared to [14]. case we relax the constraint that |A| = q − 1 and choose A
as the set of maximum number of consecutive elements. Then
A. Unit Memory Convolutional BCH Codes
|A| = a − 1 ≥ ⌊(β − α + 1)/2⌋ and s ≥ 0 giving the distance
Let Fq be a finite field with q elements, n be a positive ∆(α, β) ≥ ⌊(β − α + 1)/2⌋ + 1 = ⌊β − α + 3)/2⌋.
integer such that gcd(n, q) = 1. Let α be a primitive nth root If (β − α + 1) ≥ 2q − 2, then we can always choose a
of unity. A BCH code C of designed distance δ and length n is set A ⊆ {z | α ≤ z ≤ α + 2q − 3, z 6≡ 0 mod q} such that
a cyclic code with generator polynomial g(x) in Fq [x]/hxn − |A| = q − 1. As we want to make s as large as possible, the
1i whose defining set is given by Z = Cb ∪Cb+1 ∪· · ·∪Cb+δ−2 , worst case arises when A = {α+ q − 1, . . . , α+ 2q − 3}. Since
where Cx = {xq i mod n | i ∈ Z, i ≥ 0}. Let A+jb ⊆ Z holds for 0 ≤ j ≤ s, it follows α+2q−3+sq ≤ β.
  Thus s ≤ ⌊(β − α + 3)/q⌋ − 2. Thus the distance ∆(α, β) ≥
1 αb α2b ··· αb(n−1)
 1 αb+1 α2(b+1) · · · α(b+1)(n−1)  q + ⌊(α − β + 3)/q⌋ − 2.
  Theorem 5 (Convolutional BCH codes): Let n be a posi-
Hδ,b =  . .. .. .. .. .
 .. . . . .  tive integer such that gcd(n, q) = 1, r = ordn (q) and
(b+δ−2) 2(b+δ−2) (b+δ−2)(n−1)
1 α α ··· α 2 ≤ 2δ < δmax , where
 
Then C = {v ∈ Fnq | vHδ,b t
= 0}. If r = ordn (q), n
δmax = r (q ⌈r/2⌉ − 1 − (q − 2)[r odd]) .
then a parity check matrix H for C is given by writing q −1
Then there exists a unit memory rate k/n convolutional BCH code contains its dual. From Proposition 1 we can conclude
code with free distance df ≥ δ+1+∆(δ+1, 2δ) and k = n−κ, that there exists a convolutional code with the parameters
where κ = r ⌈δ(1 − 1/q)⌉. The free distance of the dual is [(n, n − 2κ, 1)]q . By Theorem 5 the free distance of the dual
≥ δmax + 1. is d′ ≥ δmax + 1, also implying its purity.
Proof: Let C ⊆ Fnq be a narrow-sense BCH code of Another useful method to construct quantum codes makes
designed distance 2δ + 1 and let B a basis of Fqr over Fq . use of codes over Fq2 .  
Recall that a parity check matrix for C is given by H = Theorem 7: Let 2 ≤ 2δ < n(q r − 1)/(q 2r − 1) , where
exB (H2δ+1,1 ). Further, let H0 = exB (Hδ+1,1 ), then from and r = ordn (q 2 ). Then there exist quantum convolutional
  codes with parameters [(n, n− 2κ, 1)] df ≥
Hδ+1,1  q and free distance

H2δ+1,1 = , (7) δ + 1 + ∆(δ + 1, 2δ), where κ = r δ(1 − 1/q 2 ) .
Hδ+1,δ+1
Proof: By Theorem 5 there exists an (n, n − κ, 1)q2
t convolutional BCH code with the polynomial parity check
it follows that H = [H0t , H1t ] , where H1 is obtained
from exB (Hδ+1,δ+1 ) by removing all rows common to matrix as in eq. (8). The parent BCH code has design distance
exB (Hδ+1,1 ). The code C0 with parity check matrix H0 = 2δ + 1 and given the range of δ, we know by [2, Theorem 14]
exB (Hδ+1,1 ) coincides with the narrow-sense BCH code of that it contains its Hermitian dual. By Theorem 3(b), the
length n and designed distance δ + 1. convolutional code also contains its Hermitian dual. By Propo-
By [2, Theorem 10], we have dim C = n− r ⌈2δ(1 − 1/q)⌉ sition 2, we can conclude that there exists an [(n, n − 2κ, 1)]q
and dim C0 = n − r ⌈δ(1 − 1/q)⌉ which implies rk H = code with df ≥ δ + 1 + ∆(δ + 1, 2δ).
r ⌈2δ(1 − 1/q)⌉, rk H0 = r ⌈δ(1 − 1/q)⌉, and rk H1 = We conclude by noting that the convolutional codes in The-
rk H − rk H0 = r ⌈2δ(1 − 1/q)⌉ − r ⌈δ(1 − 1/q)⌉. For x > 0, orems 6 and 7 have non-catastrophic encoders and encoder
we have ⌈x⌉ ≥ ⌈2x⌉ − ⌈x⌉; therefore, κ = rk H0 ≥ rk H1 . inverses. This follows directly from the fact that G(D) in
By Theorem 3(a), the matrix H defines a reduced basic eq. (8) is a basic generator matrix (cf. [6], [7]).
generator matrix ACKNOWLEDGMENT
e0 + DH
G(D) = H e1 (8) We would like to thank one of the referees for drawing our
attention to [4]. This research was supported by NSF CAREER
of a convolutional code of dimension κ, while its dual which award CCF 0347310, NSF grant CCF 0622201, and a Texas
we refer to as a convolutional BCH code is of dimension n−κ. A&M TITF initiative.
Now H1 is the parity check matrix of a cyclic code, C1 R EFERENCES
of the form given in Lemma 4, i.e. the defining set of C1 is [1] S. A. Aly, A. Klappenecker, and P. K. Sarvepalli, “On quantum and
Z1 as defined in (6) with α = δ + 1 and β = 2δ. Since H1 classical BCH codes,” in Proc. 2007 IEEE Intl. Symp. Inform. Theory,
is linearly independent of H0 we have x 6≡ 0 mod q in the Nice, France, 2007, (to appear), quant-ph/701037v1.
[2] ——, “On quantum and classical BCH codes,” IEEE Trans. Inform.
definition of Z1 . Theory, vol. 53, no. 3, pp. 1183–1188, 2007.
By Theorem 3(c), the free distance of the convolutional [3] G. D. Forney, Jr., M. Grassl, and S. Guha, “Convolutional and tail-biting
BCH code is bounded as min{d0 + d1 , d} ≤ df ≤ d. quantum error-correcting codes,” IEEE Trans. Inform. Theory, vol. 53,
no. 3, pp. 865–880, 2007.
By Lemma 4, d1 ≥ ∆(δ + 1, 2δ) and by the BCH bound [4] H. Gluesing-Luerssen and W. Schmale, “On doubly-cyclic convolutional
d0 ≥ δ + 1. Thus df ≥ δ + 1 + ∆(δ + 1, 2δ). The dual free codes,” Applicable Algebra in Engineering, Communication and Com-
distance also follows from Theorem 3(c) as d⊥ ⊥
f ≥ d . But
puting, vol. 17, no. 2, pp. 151–170, 2006.
⊥ [5] M. Grassl and M. Rötteler, “Quantum block and convolutional codes
d ≥ δmax + 1 by [2, Lemma 12]. from self-orthogonal product codes,” in Proc. 2005 IEEE Intl. Symp.
Inform. Theory, Adelaide, Australia, 2005, pp. 1018–1022.
V. C ONSTRUCTING Q UANTUM C ONVOLUTIONAL C ODES [6] ——, “Non-catastrophic encoders and encoder inverses for quantum
Under some restrictions on the designed free distance, we convolutional codes,” in Proc. 2006 IEEE Intl. Symp. Inform. Theory,
Seattle, USA, 2006, pp. 1109–1113.
can use convolutional codes derived in the previous section to [7] ——, “Constructions of quantum convolutional codes,” in Proc. 2007
construct quantum convolutional codes. IEEE Intl. Symp. Inform. Theory, Nice, France, 2007, (to appear),
Theorem 6: Assume the same notation as in Theorem 5. quant-ph/0703182.
[8] K. J. Hole, “On classes of convolutional codes that are not aymptotically
Then there exists a quantum convolutional code C with pa- catastrophic,” IEEE Trans. Inform. Theory, vol. 46, no. 2, pp. 663–669,
rameters [(n, n − 2κ, 1)]q , where κ = r ⌈δ(1 − 1/q)⌉. For the 2000.
free distance of C the bound df ≥ δ + 1 + ∆(δ + 1, 2δ) holds [9] W. C. Huffman and V. Pless, Fundamentals of Error-Correcting Codes.
Cambridge: University Press, 2003.
and it is pure to d′ ≥ δmax + 1. [10] L. Lee, “Short unit-memory byte-oriented binary convolutional codes
Proof: We construct a unit memory (n, n − κ)q classical having maximal free distance,” IEEE Trans. Inform. Theory, vol. 22,
convolutional BCH code as per Theorem 5. Its polynomial no. 3, pp. 349–352, 1976.
[11] H. Ollivier and J.-P. Tillich, “Quantum convolutional codes: Fundamen-
parity check matrix G(D) is as given in eq. (8). Using the tals,” 2004, quant-ph/0401134.
notation as in the proof of Theorem 5, we see that the [12] P. Piret, Convolutional Codes: An Algebraic Approach. Cambridge,
code contains its dual if H is self-orthogonal. But given Massachusetts: The MIT Press, 1988.
[13] ——, “A convolutional equivalent to Reed-Solomon codes,” Philips J.
the restrictions on the designed distance, we know from [2, Res., vol. 43, pp. 441–458, 1988.
Theorem 3] that the BCH block code defined by H contains its [14] J. Rosenthal and E. York, “BCH convolutional codes,” IEEE Trans.
dual. It follows from Theorem 3(b) that the convolutional BCH Inform. Theory, vol. 45, no. 6, pp. 1833–1844, 1999.