Wozencraft ensemble

From testwiki
Jump to navigation Jump to search

Template:Refimprove In coding theory, the Wozencraft ensemble is a set of linear codes in which most of codes satisfy the Gilbert-Varshamov bound. It is named after John Wozencraft, who proved its existence. The ensemble is described by Template:Harvtxt, who attributes it to Wozencraft. Template:Harvtxt used the Wozencraft ensemble as the inner codes in his construction of strongly explicit asymptotically good code.

Existence theorem

Theorem: Let ε>0. For a large enough k, there exists an ensemble of inner codes Cin1,⋯,CinN of rate 12, where N=qk−1, such that for at least (1−ε)N values of i,Cini has relative distance ⩾Hq−1(12−ε).

Here relative distance is the ratio of minimum distance to block length. And Hq is the q-ary entropy function defined as follows:

Hq(x)=xlogq(q−1)−xlogqx−(1−x)logq(1−x).

In fact, to show the existence of this set of linear codes, we will specify this ensemble explicitly as follows: for α∈𝔽qk−{0}, define the inner code

{Cinα:𝔽qk→𝔽q2kCinα(x)=(x,αx)

Here we can notice that x∈𝔽qk and α∈𝔽qk. We can do the multiplication αx since 𝔽qk is isomorphic to 𝔽qk.

This ensemble is due to Wozencraft and is called the Wozencraft ensemble.

For all x,y∈𝔽qk, we have the following facts:

  1. Cinα(x)+Cinα(y)=(x,αx)+(y,αy)=(x+y,α(x+y))=Cinα(x+y)
  2. For any a∈𝔽q,aCinα(x)=a(x,αx)=(ax,α(ax))=Cinα(ax)

So Cinα is a linear code for every α∈𝔽qk−{0}.

Now we know that Wozencraft ensemble contains linear codes with rate 12. In the following proof, we will show that there are at least (1−ε)N those linear codes having the relative distance ⩾Hq−1(12−ε), i.e. they meet the Gilbert-Varshamov bound.

Proof

To prove that there are at least (1−ε)N number of linear codes in the Wozencraft ensemble having relative distance ⩾Hq−1(12−ε), we will prove that there are at most εN number of linear codes having relative distance <Hq−1(12−ε) i.e., having distance <Hq−1(12−ε)⋅2k.

Notice that in a linear code, the distance is equal to the minimum weight of all codewords of that code. This fact is the property of linear code. So if one non-zero codeword has weight <Hq−1(12−ε)⋅2k, then that code has distance <Hq−1(12−ε)⋅2k.

Let P be the set of linear codes having distance <Hq−1(12−ε)⋅2k. Then there are |P| linear codes having some codeword that has weight <Hq−1(12−ε)⋅2k.

Lemma. Two linear codes Cinα1 and Cinα2 with α1,α2∈𝔽qk distinct and non-zero, do not share any non-zero codeword.
Proof. Suppose there exist distinct non-zero elements α1,α2∈𝔽qk such that the linear codes Cinα1 and Cinα2 contain the same non-zero codeword y. Now since y∈Cinα1,y=(y1,α1y1) for some y1∈𝔽qk and similarly y=(y2,α2y2) for some y2∈𝔽qk. Moreover since y is non-zero we have y1,y2≠0. Therefore (y1,α1y1)=(y2,α2y2), then y1=y2≠0 and α1y1=α2y2. This implies α1=α2, which is a contradiction.

Any linear code having distance <Hq−1(12−ε)⋅2k has some codeword of weight <Hq−1(12−ε)⋅2k. Now the Lemma implies that we have at least |P| different y such that wt(y)<Hq−1(12−ε)⋅2k (one such codeword y for each linear code). Here wt(y) denotes the weight of codeword y, which is the number of non-zero positions of y.

Denote

S={y : wt(y)<Hq−1(12−ε)⋅2k}

Then:[1]

|P|⩽|S|⩽Volq(Hq−1(12−ε)⋅2k,2k)Volq(r,n) is the volume of Hamming ball of radius r in [q]n⩽qHq(Hq−1(12−ε))⋅2kVolq(pn,n)⩽qHq(p)n=q(12−ε)⋅2k=qk(1−2ε)<ε(qk−1) for k large enough =εN

So |P|<εN, therefore the set of linear codes having the relative distance ⩾Hq−1(12−ε)⋅2k has at least N−εN=(1−ε)N elements.

See also

References

Template:Reflist

  1. ↑ For the upper bound of the volume of Hamming ball check Bounds on the Volume of a Hamming ball