Skip to main content

Worksheet Problem Set M6

Instructions: You may type up or handwrite your work, but it must be neat, professional, and organized and it must be saved as a PDF file and uploaded to the appropriate Gradescope assignment. Use a scanner or scanning app to convert handwritten work on paper to PDF. I encourage you to type your work using the provided template.
All tasks below must have a complete solution that represents a good-faith attempt at being right to receive engagement credits. If your submission is complete and turned in on time, you will receive full engagement credits for the assignment. All other submissions will receive zero engagement credit. Read the guidelines at Grading Specifications carefully.
To abide by the class’ academic honesty policy, your work should represent your own understanding in your own words. If you work with other students, you must clearly indicate who you worked with in your submission. The same is true for using tools like generative AI although I strongly discourage you from using such tools since you need to build your own understanding here to do well on exams.

1. Minimal Polynomials.

(a)

Let \(a(x)\) be a monic irreducible polynomial over a finite field \(F\text{.}\) Show that \(a(x)\) is the minimal polynomial of the element \(\xi\) in the extension field \(F[\xi]/a(\xi).\)
Solution.
In this extension field the element \(\xi\) satisfies the equation \(a(\xi)=0\text{,}\) i.e. is a root of \(a(x)\text{.}\) So the minimal polynomial \(m_{\xi}(x)\) over \(F\) divides \(a(x)\text{,}\) but since \(a(x)\) is a monic irreducible over \(F\) we must have \(m_{\xi}(x)=a(x)\text{.}\)

(b)

Let \(\alpha\) be a nonzero element in an extension field \(E\) of \(F=\GF(q)\text{.}\) Show that the conjugacy classes of \(\alpha\) and \(\alpha^{-1}\) (with respect to \(F\)) are the same size.
Solution.
Let \(t=|C_{\alpha}|\text{,}\) so that \(\alpha^{q^t}=\alpha\) and no smaller positive power of \(q\) satisfies \(\alpha^{q^d}=\alpha\text{.}\) Then
\begin{equation*} (\alpha^{-1})^{q^t}=(\alpha^{q^t})^{-1}=\alpha^{-1} \end{equation*}
and by nearly the same computation if a smaller power of \(q\) satisfies
\begin{equation*} (\alpha^{-1})^{q^d}=\alpha^{-1} \end{equation*}
then it will also satisfy \(\alpha^{q^d}=\alpha\text{,}\) so \(t\) must be the smallest positive power possible and thus is the size of \(C_{\alpha^{-1}}\text{.}\)

(c)

Let \(M_{\alpha}(x)=\sum_{i=0}^m a_i x^i\) be the minimal polynomial of \(\alpha\) as in (b) (with respect to \(F\)). Show that the minimal polynomial of \(\alpha^{-1}\) is given by
\begin{equation*} M_{\alpha^{-1}}(x)=\frac{1}{a_0} \sum_{i=0}^m a_{m-i}x^i = \frac{1}{a_0} M_{\alpha}(x)^{[-1]}\text{.} \end{equation*}
That is, up to scaling by some nonzero element of \(F\text{,}\) the minimal polynomial of \(\alpha^{-1}\) is the reciprocal polynomial obtained by reversing the order of coefficients of the minimal polynomial of \(\alpha\text{.}\)
Solution.
From the previous part we know that \(m=\deg(M_{\alpha}(x))=\deg(M_{\alpha^{-1}}(x))\text{.}\) We will show that
\begin{equation*} \frac{1}{a_0}\sum_{i=0}^m a_{m-i}x^i = \prod_{i=0}^m (x-(\alpha^{-1})^{q^i}) \end{equation*}
which proves the claim. To do this, first note that both polynomials are monic, so it suffices to show that they have the same roots. Since the conjugates of a root of a polynomial over \(F\) are also roots, it suffices to show that \(\alpha^{-1}\) is a root of the left-hand polynomial.
\begin{align*} \frac{1}{a_0}\sum_{i=0}^m a_{m-i}(\alpha^{-1})^i \amp = \frac{1}{a_0}\sum_{i=0}^m a_{i}(\alpha^{-1})^{m-i}\\ \amp = \frac{1}{a_0}\sum_{i=0}^m a_{i}\alpha^{i-m}\\ \amp = \frac{1}{a_0\alpha^m}\sum_{i=0}^m a_{i}\alpha^{i}\\ \amp = \frac{1}{a_0\alpha^m}m_{\alpha}(\alpha)\\ \amp =0 \text{.} \end{align*}
So the claim holds.

2. All Finite Field Automorphisms are Frobenius.

Let \(p\) be a prime and let \(E\) and \(K\) be extension fields of \(F=\GF(p)\) with the same extension degree \(n\text{.}\) In all parts below, you may use without proof the fact that any isomorphism of an extension field of \(\GF(p)\) must fix \(\GF(p)\text{.}\)

(a)

Let \(\varphi: K \to E\) be an isomorphism (a multiplicative, additive, bijective map). Show that the minimal polynomial (with respect to \(F\)) of every element \(\beta \in K\) is the same as the minimal polynomial of the element \(\varphi(\beta)\in E\text{.}\)
Solution.
Let \(\beta \in K\) and \(m_{\beta}(x)=\sum_{i=0}^m a_i x^i\) be its minimal polynomial over \(\GF(p)\text{.}\) Note that \(a_i\in F\) so \(\varphi(a_i)=a_i\) for all \(i\text{.}\) Then we have
\begin{align*} 0 \amp =\varphi(0) \\ \amp =\varphi(m_{\beta}(\beta))\\ \amp =\varphi\left(\sum_{i=0}^m a_i \beta^i\right)\\ \amp =\sum_{i=0}^m \varphi(a_i) \varphi(\beta)^i\\ \amp =\sum_{i=0}^m a_i \varphi(\beta)^i\\ \amp =m_{\beta}(\varphi(\beta)) \end{align*}
Since \(m_{\beta}(x)\) is a monic irreducible over \(F\) it is the minimal polynomial for all of its roots; in particular for \(\varphi(\beta)\in E\text{.}\)

(b)

Show that every isomorphism \(\varphi: K \to E\) is completely defined by the value of \(\varphi\) at an element \(\beta\in K\) that does not belong to any proper subfield of \(K\text{.}\)
Solution.
Let \(\beta \in K\) be an element that does not belong to any proper subfield of \(K\text{.}\) Then the minimal polynomial of \(\beta\) has degree \(n\) and correspondingly the set \(\{1,\beta,\beta^2,\dots,\beta^{n-1}\) forms an \(F\)-vector space basis of \(K\) over \(F\text{.}\) Since \(\varphi\) fixes \(F\text{,}\) we have that for any other element \(\alpha \in K\text{:}\)
\begin{align*} \varphi(\alpha) \amp = \varphi\left(\sum_{i=0}^{n-1} a_i \beta^i\right) \\ \amp \sum_{i=0}^{n-1} \varphi(a_i) \varphi(\beta)^i \\ \amp \sum_{i=0}^{n-1} a_i \varphi(\beta)^i \end{align*}
by the additive and multiplicative properties of an isomorphism. So the action on any other \(\alpha\) is determined by the value of \(\varphi(\beta)\text{.}\) Since \(\beta\) was an arbitrary element of \(K\) not belonging to any of its proper subfields, the result follows.

(c)

Show that there are exactly \(n\) distinct isomorphisms \(\varphi:K \to E\text{.}\)
Hint.
Given an element \(\beta \in K\) that does not belong to any proper subfield of \(K\text{,}\) what are the possible values that \(\varphi(\beta)\) may take?
Solution.
Let \(\varphi: K \to E\) be an isomorphism, \(\beta\in K\) be an element not belonging to a proper subfield of \(K\text{,}\) and \(m_{\beta}(x)\) be the corresponding minimal polynomial over \(F\text{.}\) From (a) we know that \(\varphi(\beta)\) is a root of \(m_{\beta}(x)\) in \(E\text{.}\) The argument of part (a) applies equally well to all \(F\)-conjugates of \(\beta\text{,}\) so all \(n\) conjugate roots are present in \(E\text{.}\) From (b) any choice of a value of \(\varphi(\beta)\) fully determines \(\varphi\) so there are \(n\) isomorphisms \(\varphi:K\to E\) given by the \(n\) maps determined by taking
\begin{equation*} \varphi(\beta)=\gamma^{q^i} \end{equation*}
for some element \(\gamma\in E\) a root of \(a(x)\) and \(i=0,\dots,n-1\) and extending to an isomorphism. Each such map is distinct since the image of \(\beta\) is different for each.

(d)

Show that the automorphisms \(\varphi: K \to K\) are the Frobenius mappings \(f_m: x \mapsto x^{p^m}\text{.}\)
Solution.
This follows immediately from part (c) since we saw that \(\varphi\) must take an element of a conjugacy class of an element in \(K\) but not any proper subfield to another root of its minimal polynomial, i.e. to another element in the same conjugacy class. The other elements of that conjugacy class are exactly the powers \(\beta,\beta^p,\dots,\beta^{p^{n-1}}\) and these choices for \(\beta\) result in the Frobenius maps because \(f_m(\beta)=\beta^{p^m}\text{.}\)

3. Burst Error Correction and Cyclic Codes.

Let \(C\) be a binary \([15,9]\) cyclic code generated by \(g(x)=1+x^3+x^4+x^5+x^6\text{.}\) Suppose that we know \(C\) can correct any burst of length 3 or less. Decode each of the following vectors.

(a)

\(\vec{y}_1=(10101\, 11010\, 11100)\)
Solution.
For each of these parts we will implement Error Trapping for Cyclic Burst-Error Codes. So our first step is to find the syndrome \(s_0(x)\) by long division. We have
\begin{equation*} x^{12}+x^{11}+x^{10}+x^8+x^6+x^5+x^4+x^2+1 = (x^{6} + x^{3} + x)g(x)+x^{3} + x^{2} + x + 1 \end{equation*}
so \(s_0(x)=x^{3} + x^{2} + x + 1\) and \(\vec{s_0}=111100\text{.}\) This is a burst of length \(4\) so we begin shifting.
The first two shifts do not affect the burst length and we get immediately \(\vec{s_1}=011110\) and \(\vec{s_2}=001111\text{.}\) Then we get
\begin{equation*} \vec{s_3}: x s_2(x)-g(x) \mapsto 100000\text{.} \end{equation*}
Since this is a burst of length 3 or less, we stop and decode. The error is \(e(x)=x^{12}(1)\) so the corrected codeword is \(\vec{c}=(10101\, 11010\, 11000)\text{.}\)

(b)

\(\vec{y}_2=(11000\, 01110\, 10011)\)
Solution.
Again our first step is to find the syndrome \(s_0(x)\) by long division. We have
\begin{equation*} x^{14} + x^{13} + x^{10} + x^8 + x^7 + x^6 + x + 1 = (x^{8} + x^{6} + x^{3} + x^{2} + x + 1)g(x)+x^{5} + x^{2} \end{equation*}
so \(s_0(x)=x^{5} + x^{2}\) and \(\vec{s_0}=001001\text{.}\) Again this is a burst of length \(4\) so we begin shifting.
We get
\begin{align*} s_1 \amp = 100011 \amp \amp\text{(a burst of length 6)}\\ s_2 \amp = 110110 \amp \amp\text{(a burst of length 5)}\\ s_3 \amp = 011011 \amp \amp\text{(a burst of length 5)}\\ s_4 \amp = 101010 \amp \amp\text{(a burst of length 5)}\\ s_5 \amp = 010101 \amp \amp\text{(a burst of length 5)}\\ s_6 \amp = 101101 \amp \amp\text{(a burst of length 6)}\\ s_7 \amp = 110001 \amp \amp\text{(a burst of length 6)}\\ s_8 \amp = 111111 \amp \amp\text{(a burst of length 6)}\\ s_9 \amp = 111000 \amp \amp\text{(a burst of length 3)}\text{.} \end{align*}
So the error is \(e(x)=x^6(1+x+x^2)\) and the corrected codeword is \(\vec{c}=(11000\, 00000\, 10011)\text{.}\)

(c)

\(\vec{y}_3=(01000\, 00010\, 11111)\)
Solution.
Again our first step is to find the syndrome \(s_0(x)\) by long division. We have
\begin{equation*} x^{14} + x^{13} + x^{12} + x^{11} + x^{10} + x^{8} + x = (x^{8} + x^{4} + x^{3} + 1)g(x)+x^{5} + x + 1 \end{equation*}
so \(s_0(x)=x^{5} + x + 1\) and \(\vec{s_0}=110001\text{.}\) This is already a cyclic burst of length \(3\) but not a non-cyclic one, so we have to continue.
We get
\begin{align*} s_1 \amp = 111111 \amp \amp\text{(a burst of length 6)}\\ s_2 \amp = 111000\amp \amp\text{(a burst of length 3)} \text{.} \end{align*}
Notice we get the same run of error bits as part (b), but at a different position since this time \(e(x)=x^{13}(x^2+x+1) \pmod{x^{15}-1}=x^{14}+x^{13}+1\text{.}\) So the corrected codeword is \(\vec{c}=(11000\,00010\,11100)\text{.}\)

4. BCH Codes Again.

Let \(\alpha\) be a primitive element in \(E=\GF(2^m)\) and let \(C_{\BCH}\) be a narrow-sense primitive BCH code (of length \(n=2^m-1\)) over \(F=\GF(2)\) whose parity-check matrix is obtained by representing each entry in
\begin{equation*} \begin{bmatrix} 1 \amp \alpha \amp \alpha^2 \amp \dots \amp \alpha^{n-1} \\ 1 \amp \alpha^3 \amp \alpha^6 \amp \dots \amp \alpha^{3(n-1)} \end{bmatrix} \end{equation*}
as a column vector in \(F^m \text{.}\) For the given code \(C_{\BCH}\text{,}\) let \(C_{\RS}\) be an \([N=n,K,D]\) code over \(E\) that satisfies
\begin{equation*} C_{\BCH}=C_{\RS} \cap F^N \end{equation*}

(a)

Show that \(\alpha^2,\alpha^4,\) and \(\alpha^6\) are roots of \(C_{\BCH}\text{.}\)
Solution.
This follows from the linearity of the 2-Frobenius and the fact that \(\alpha\) and \(\alpha^3\) are given to be roots.

(b)

Write a parity-check matrix of \(C_{\RS}\text{,}\) assuming that \(D=5\) and that \(\alpha^3\) is a root of \(C_{\RS}\text{.}\)
Solution.
If \(D=5\) then we need four consecutive roots for \(C_{\RS}\) and we have to avoid introducing any more roots to \(C_{\BCH}\text{.}\) So the roots must be \(\alpha,\alpha^2,\alpha^3,\alpha^4\) (as shifting either direction would add either \(1\) or \(\alpha^5\) as a root). Hence a parity-check matrix of \(C_{\RS}\) is
\begin{equation*} H=\begin{bmatrix} 1 \amp \alpha \amp \alpha^2 \amp \dots \amp \alpha^{n-1} \\ 1 \amp \alpha^2 \amp \alpha^4 \amp \dots \amp \alpha^{2(n-1)}\\ 1 \amp \alpha^3 \amp \alpha^6 \amp \dots \amp \alpha^{3(n-1)}\\ 1 \amp \alpha^4 \amp \alpha^8 \amp \dots \amp \alpha^{4(n-1)}\\ \end{bmatrix}\text{.} \end{equation*}

(c)

Repeat part (b), except that now \(D=4\) (and \(\alpha^3\) is still a root of \(C_{\RS}\)). Is the solution unique?
Solution.
Now we only need three consecutive roots, including \(\alpha^3\text{,}\) and not introducing extraneous roots. In this case there are two possible solutions: \(\{\alpha,\alpha^2,\alpha^3\}\) and \(\{\alpha^2,\alpha^3,\alpha^4\}\) producing respective parity-check matrices
\begin{equation*} H_1=\begin{bmatrix} 1 \amp \alpha \amp \alpha^2 \amp \dots \amp \alpha^{n-1} \\ 1 \amp \alpha^2 \amp \alpha^4 \amp \dots \amp \alpha^{2(n-1)}\\ 1 \amp \alpha^3 \amp \alpha^6 \amp \dots \amp \alpha^{3(n-1)}\\ \end{bmatrix} \end{equation*}
and
\begin{equation*} H_2=\begin{bmatrix} 1 \amp \alpha^2 \amp \alpha^4 \amp \dots \amp \alpha^{2(n-1)}\\ 1 \amp \alpha^3 \amp \alpha^6 \amp \dots \amp \alpha^{3(n-1)}\\ 1 \amp \alpha^4 \amp \alpha^8 \amp \dots \amp \alpha^{4(n-1)}\\ \end{bmatrix}\text{.} \end{equation*}

(d)

Find the largest possible dimension of any RS code \(C_{\RS}\) over \(E\) that satisfies \(C_{\BCH}=C_{\RS}\cap \F^N\text{.}\)
Solution.
To maximize the dimension of the RS code, we must minimize its distance, since we have \(K=N-D+1\text{.}\) So we must minimize the required consecutive root sequence. We saw above that \(D=4\) is possible; can we do better?
Yes, we can. We need a root from each cyclotomic coset of the roots of \(C_{\BCH}\text{;}\) one from \(\{\alpha,\alpha^2,\dots,\}\) and one from \(\{\alpha^3,\alpha^6,\dots\}\text{.}\) It follows that \(D=2\) (one root) is not possible. But \(D=3\) is possible: take \(C_{\RS}\) to have root sequence \(\alpha^2,\alpha^3\) or \(\alpha^3,\alpha^4\text{.}\) So the largest possible dimension is \(N-3+1=N-2\text{.}\)