In information theory, the bar product of two linear codes C2 ⊆ C1 is defined as
C1\midC2=\{(c1\midc1+c2):c1\inC1,c2\inC2\},
where (a | b) denotes the concatenation of a and b. If the code words in C1 are of length n, then the code words in C1 | C2 are of length 2n.
The bar product is an especially convenient way of expressing the Reed–Muller RM (d, r) code in terms of the Reed–Muller codes RM (d - 1, r) and RM (d - 1, r - 1).
The bar product is also referred to as the | u | u+v | construction[1] or (u | u + v) construction.[2]
The rank of the bar product is the sum of the two ranks:
\operatorname{rank}(C1\midC2)=\operatorname{rank}(C1)+\operatorname{rank}(C2)
Let
\{x1,\ldots,xk\}
C1
\{y1,\ldots,yl\}
C2
\{(xi\midxi)\mid1\leqi\leqk\}\cup\{(0\midyj)\mid1\leqj\leql\}
is a basis for the bar product
C1\midC2
The Hamming weight w of the bar product is the lesser of (a) twice the weight of C1, and (b) the weight of C2:
w(C1\midC2)=min\{2w(C1),w(C2)\}.
For all
c1\inC1
(c1\midc1+0)\inC1\midC2
which has weight
2w(c1)
(0\midc2)\inC1\midC2
for all
c2\inC2
w(c2)
c1\inC1,c2\inC2
w(C1\midC2)\leqmin\{2w(C1),w(C2)\}
Now let
c1\inC1
c2\inC2
c2\not=0
\begin{align} w(c1\midc1+c2)&=w(c1)+w(c1+c2)\\ &\geqw(c1+c1+c2)\\ &=w(c2)\\ &\geqw(C2) \end{align}
If
c2=0
\begin{align} w(c1\midc1+c2)&=2w(c1)\\ &\geq2w(C1) \end{align}
so
w(C1\midC2)\geqmin\{2w(C1),w(C2)\}
. F.J. MacWilliams . Jessie MacWilliams . N.J.A. Sloane . The Theory of Error-Correcting Codes . registration . North-Holland . 1977 . 0-444-85193-3 . 76 .