Linear algebra problem sheet 1

 
    In Q1-7 you can assume the vector spaces are finite-dimensional.
  1. Let $F$ be a field and $f(x)$ an irreducible polynomial in $F[x]$. Show that the quotient ring $F[x]/⟨f(x)⟩$ is a field.
    Proof.
    Since $F[x]$ is a commutative ring with identity, so is $F[x]/⟨f(x)⟩$. To show $F[x]/⟨f(x)⟩$ is a field only needs to show that nonzero elements are invertible.
    A nonzero element of $F[x]/⟨f(x)⟩$ has the form $a(x) + \langle f(x) \rangle$ for $a(x) \in F[x]∖⟨f(x)⟩$, so $f(x)∤a(x)$, by irreducibility of $f(x)$, $\gcd(a(x),f(x))=1$, then $∃m(x),n(x) ∈ F[x]$ , such that $a(x) m(x) + f(x) n(x) = 1$. Then\begin{aligned} a(x) \cdot m(x)+ f(x) \cdot n(x)+ \langle f(x) \rangle & = 1 + \langle f(x) \rangle \\ a(x) \cdot m(x) + \langle f(x) \rangle & = 1 + \langle f(x) \rangle \\ \left(a(x) + \langle f(x) \rangle\right) \left(m(x) + \langle f(x) \rangle\right) & = 1 + \langle f(x) \rangle \end{aligned}This shows that $m(x) + \langle f(x) \rangle$ is the multiplicative inverse of $a(x) + \langle f(x) \rangle$.
  2. A rational function over a field $F$ is a quotient $\frac{f(t)}{g(t)}$ where $f, g$ are polynomials over $F$ and $g$ is not identically zero. (Obviously we identify $\frac{f_1}{g_1}$ and $\frac{f_2}{g_2}$ if $f_1g_2=f_2g_1$ as polynomials).
    Show that the set $F(t)$ of rational functions form a field and use this to produce an example of an infinite field of positive characteristic.
    Solution.
    For $\frac{f_1}{g_1}$ and $\frac{f_2}{g_2}$, define $\frac{f_1}{g_1}+\frac{f_2}{g_2}=\frac{f_1g_2+f_2g_1}{g_1g_2},\frac{f_1}{g_1}⋅\frac{f_2}{g_2}=\frac{f_1f_2}{g_1g_2}$.
    (1) Addition and multiplication are well-defined.
    Suppose $\frac{a}{b} = \frac{a'}{ b'}$ and $\frac{c}{d} = \frac{c'}{ d'}$. Then $\frac{a}{b} + \frac{c}{d} = \frac{a d + b c}{ b d}$ and $\frac{a'}{ b'} + \frac{c'}{ d'} = \frac{a' d' + b' c'}{ b' d'}$ and $\frac{a}{b} \frac{c}{d} = \frac{a c}{ b d}$ and $\frac{a'}{ b'} \frac{c'}{ d'} = \frac{a' c'}{ b' d'}$
    $(a d + b c) b' d' = a b' d d' + b b' c d' = a' b d d' + b b' c' d = (a' d' + b' c') b d ⇒ \frac{a d + b c}{ b d} = \frac{a' d' + b' c'}{ b' d'}$
    $(a c)(b' d') = a b' c d' = a' b c' d = (a' c')(b d) ⇒ \frac{a c}{ b d} = \frac{a' c'}{ b' d'}$

    (2) Addition is associative. $$\left(\frac{a}{b} + \frac{c}{d}\right) + \frac{e}{f} = \frac{a d + b c}{ b d} + \frac{e}{f} = \frac{a d f + b c f + b d e}{ b d f},$$ $$\frac{a}{b} + \left(\frac{c}{d} + \frac{e}{f}\right) = \frac{a}{b} + \frac{c f + d e}{ d f} = \frac{a d f + b c f + b d e}{ b d f}.$$ (3) Addition is commutative. $$\frac{a}{b} + \frac{c}{d} = \frac{a d + b c}{ b d} \quad\hbox{and}\quad \frac{c}{d} + \frac{a}{b} = \frac{b c + a d}{ b d}.$$ (4) $\frac{0}{1}$ is the additive identity. $$\frac{a}{b} + \frac{0}{1} = \frac{a \cdot 1 + b \cdot 0}{ b} = \frac{a}{b}.$$ (5) $-\frac{a}{b} = \frac{-a}{ b}$ . $$\frac{a}{b} + \frac{-a}{ b} = \frac{a b - a b}{ b^2} = \frac{0}{ b^2}.$$ Finally, $\frac{0}{ b^2} = \frac{0}{1}$ , since $0\cdot 1 = b^2\cdot 0$ .

    (6) Multiplication is associative. $$\left(\frac{a}{b} \frac{c}{d}\right) \frac{e}{f} = \frac{a c e}{ b d f} = \frac{a}{b} \left(\frac{c}{d} \frac{e}{f}\right).$$ (7) Multiplication is commutative. $$\frac{a}{b} \frac{c}{d} = \frac{a c}{ b d} = \frac{c}{d} \frac{a}{b}.$$ (8) $\frac{1}{1}$ is the multiplicative identity. $$\frac{a}{b} \frac{1}{1} = \frac{a}{b}.$$ (9) Multiplication distributes over addition. By commutativity of multiplication, it suffices to check this on one side. $$\frac{a}{b}\left(\frac{c}{d} + \frac{e}{f}\right) = \frac{a}{b} \frac{c f + d e}{ d f} = \frac{a c f + a d e}{ b d f},$$ $$\frac{a}{b} \frac{c}{d} + \frac{a}{b} \frac{e}{f} = \frac{a c}{ b d} + \frac{a e}{ b f} = \frac{a b c f + a b d e}{ b^2 d f}.$$ Finally, $$(a c f + a d e) b^2 d f = a b^2 c d f^2 + a b^2 d^2 e f \quad\hbox{and}\quad (a b c f + a b d e) b d f = a b^2 c d f^2 + a b^2 d^2 e f.$$ Therefore, $\frac{a c f + a d e}{ b d f} = \frac{a b c f +a b d e}{ b^2 d f}$.

    (10) Nonzero elements have multiplicative inverses.
    Suppose $\frac{a}{b} \ne \frac{0}{1}$ , so $a \ne 0$ . Then using $a b \cdot 1 = 1 \cdot a b$ , I have $\frac{a}{b} \frac{b}{a} = \frac{a b}{ a b} = \frac{1}{1}$. Hence, $\frac{b}{a} $ is the inverse of $ \frac{a}{b}$ .
    This completes the verification that $F(t)$ is a field.
    When $F=𝔽_p,F(t)$ is an infinite field of characteristic $p$.
    The field of rational functions on $F(t)$, that is $(F(t))(t')$, is isomorphic to the field of rational functions of two variables $t,t'$. By symmetry, this is isomorphic to $(F(t'))(t)$.
  3. Show that $\mathbb{Z}$ is a principal ideal domain, ie. every ideal is of the form $\langle m\rangle=m \mathbb{Z}$ for some $m \in \mathbb{Z}$.
    Discuss how to prove this result for $F[x]$, the ring of polynomials with coefficients in a field $F$.
    Proof for ℤ.
    Let $I$ be an ideal of $\mathbb Z$. If $I=\{0\}$ then $I=⟨0⟩$ and we are done.
    Suppose $I\neq\{0\}$, let $I^+=I∩ℤ^+$ and $a=\min I^+$. We will prove $⟨a⟩=I$.
    Since $I$ is an ideal, $∀r∈ℤ, ar\in I$, so $⟨a⟩⊂I$.
    Clearly $0∈⟨a⟩$. For any $b∈I∖\{0\}$, by Division Algorithm we have $b=aq+r$ for some $q,r∈ℤ$ and $0≤r< a$.
    Since $b,aq \in I$, $r=b-aq \in I$. This implies $r=0$ since $a$ is the smallest element in $I^+$. So $b∈⟨a⟩$. So $⟨a⟩⊃I$. So $⟨a⟩=I$.
    Proof for $F[x]$.
    Let $I$ be an ideal of $F[x]$. If $I=\{0\}$ then $I=⟨0⟩$ and we are done.
    Suppose $I\neq\{0\}$, let $a$ be a polynomial of the lowest degree in $I∖\{0\} $. We will prove $⟨a⟩=I$.
    Since $I$ is an ideal, $∀r∈F[x], ar\in I$, so $⟨a⟩⊂I$.
    Clearly $0∈⟨a⟩$. For any $b∈I∖\{0\}$, by Division Algorithm we have $b=aq+r$ for some $q,r∈F[x]$ and $\deg r<\deg a$.
    Since $b,aq \in I$, $r=b-aq \in I$. This implies $r=0$ since $a$ has the lowest degree in $I∖\{0\}$. So $b∈⟨a⟩$. So $⟨a⟩⊃I$. So $⟨a⟩=I$.
  4. Let $P: V → V$ be a projection, that is, $P^2=P$. Show that $V=\operatorname{im}P ⊕ \ker P$, and deduce that there is a basis in which $P$ is a block matrix $$ P=\left(\begin{array}c I_r & 0 \\ 0 & 0 \end{array}\right) $$ where $r$ is the rank of $P$.
    What are the minimum and characteristic polynomials of $P$ ?
    Solution.
    For any $v∈V$, $v=P(v)+(v-P(v))$, since $P(v)∈\operatorname{im}P$ and $P(v-P(v))=P(v)-P^2(v)=0⇒v-P(v)∈\ker P$, we have $V=\operatorname{im}P + \ker P$.
    Suppose $w∈\operatorname{im}P ∩ \ker P$, we have $w=P(v)$ for some $v∈V$, then $0=P(w)=P^2(v)=P(v)=w⇒\operatorname{im}P ∩ \ker P=\{0\}⇒V=\operatorname{im}P ⊕ \ker P$.
    Suppose $w∈\operatorname{im}P$, we have $w=P(v)$ for some $v∈V$, then $P(w)=P^2(v)=P(v)=w$. So the matrix of $P|_{\operatorname{im}P}$ is identity matrix. Also $P|_{\operatorname{im}P}$ is zero matrix.
    Since $V=\operatorname{im}P ⊕ \ker P$, the union of a basis of $\operatorname{im}P$ and a basis of $\operatorname{im}P$ is a basis for $V$. In this basis $P$ is a block matrix $ \left(\begin{array}c I_r & 0 \\ 0 & 0 \end{array}\right) $, where $r$ is the rank of $P$.
    Let $n=\dim V$, then $m_P(t)=t^2-t,χ_P(t)=|P-tI_n|=\begin{vmatrix} (1-t)I_r & 0 \\ 0 & -tI_{n-r} \end{vmatrix}=(1-t)^r(-t)^{n-r}$.
  5. Show that a block triangular matrix $$ X=\left(\begin{array}{cc} A & B \\ 0 & D \end{array}\right) $$ has determinant $\det X=\det A \det D$. (One way to do this is to look for a factorisation $X=X_1X_2X_3$ where $X_2$ is block triangular and $X_1,X_3$ are block diagonal, and some of the diagonal blocks are the identity).
    Deduce the equality of characteristic polynomials $χ_X(t)=χ_A(t)χ_D(t)$.
    Proof.
    If $\det A=0$, columns of $A$ are linearly dependent, hence the columns of $X$ are linearly dependent, hence $\det X=0$. If $\det A≠0$, we have\begin{align*} &\det\pmatrix{A&B\\0&D}\\ =&\det\pmatrix{A&0\\0&D}⋅\det\pmatrix{I&A^{-1}B\\0&I}\\ =&\hskip3.47px\det A\det D\hskip3.47px⋅\hskip47.3px1&\text{by Laplace expansion} \end{align*} $χ_X(t)=\det\pmatrix{A-tI&B\\0&D-tI}=\det(A-tI)⋅\det(D-tI)=χ_A(t)χ_D(t)$
    We proved the same question using Leibniz formula for determinants in Q3 sheet 1 in M1: Linear Algebra Ⅱ.
  6. Prove that $T: V → V$ is invertible if and only if $x$ does not divide the minimal polynomial $m_T(x)$.
    Solution.
    $T$ is invertible ⇔ $\ker T=\{0\}$ ⇔ 0 is not an eigenvalue of $T$ ⇔ 0 is not a root of $m_T(x)$ ⇔ $x$ does not divide $m_T(x)$.
  7. Let $T: V → V$ be a linear transformation. Assume that $v, T v, T^2 v, …$ span $V$ for some $v \in V$. Show that
    (i) there exists a $k$ such that $v, T v, \ldots, T^{k-1} v$ are linearly independent and for some $a_i$$$T^kv=a_0v+a_1T v+\ldots+a_{k-1} T^{k-1} v$$ (ii) the set $v, T v, \ldots, T^{k-1} v$ forms a basis for $V$;
    (iii) its minimal polynomial is given by $m_T(x)=x^k-a_{k-1} x^{k-1}-\ldots-a_0$;
    (iv) What is the characteristic polynomial $χ_T(x)$?
    Computation of minimal polynomial by Cyclic vector
    Companion matrix of a polynomial
    Solution.
    (i) Let $n=\dim V$, for $k=0$, the single element $v$ is linearly independent; for $k>n$, $v, T v, \ldots, T^{k-1} v$ are linearly dependent. So there exists $0< k≤n$ such that $v, T v, \ldots, T^{k-1} v$ are linearly independent but $v, T v, \ldots, T^k v$ are linearly dependent, so $T^kv$ can be written as linear combination of the rest.
    (ii) Since $v, T v, \ldots$ span $V$ and $T^mv(m≥k)$ can be written as linear combination of $v, T v, \ldots, T^{k-1} v$, we have $v, T v, …, T^{k-1} v$ span $V$. Also they are linearly independent, so they form a basis.
    (iii) Proof 1: $x^k-a_{k-1} x^{k-1}-…-a_0$ is the minimal polynomial of $T$ in $⟨v⟩$, so $x^k-a_{k-1} x^{k-1}-…-a_0∣m_T(x)$.
    By Cayley-Hamilton, $\deg m_T(x)≤\deg V=k$, so they are equal.
    Proof 2: In general if any polynomial in $T$ annihilates a vector $v$, then it also annihilates $T^iv$ (just apply $T^i$ to the equation $p(T)v=0$).
    $V=⟨v,Tv,⋯,T^{k-1}v⟩⇒T^k-a_{k-1} T^{k-1}-…-a_0=0_V⇒m_T(x)∣x^k-a_{k-1}x^{k-1}-…-a_0$.
    But $v,Tv,⋯,T^{k-1}v$ are linearly independent$⇒\deg m_T(x)≥k⇒m_T(x)=x^k-a_{k-1}x^{k-1}-…-a_0$.
    (iv) The matrix of $T$ in the basis $v, T v, …, T^{k-1} v$ is $\pmatrix{0&0&0&⋯&0&a_0\\1 & 0 & 0&⋯&0&a_1\\ 0 & 1 & 0&⋯&0&a_2\\ 0 & 0 &1&⋯&0&a_3\\⋮&⋮&⋮&\ddots&⋮&⋮\\0&0&0&⋯&1&a_{k-1}}⇒χ_T(x)=\left|T-xI\right|=\begin{vmatrix}-x&0&0&⋯&0&a_0\\1 & -x & 0&⋯&0&a_1\\ 0 & 1 & -x&⋯&0&a_2\\ 0 & 0 &1&⋯&0&a_3\\⋮&⋮&⋮&\ddots&⋮&⋮\\0&0&0&⋯&1&a_{k-1}-x\end{vmatrix}$
    Laplace expansion along last column $χ_T(x)=(-1)^{k-1}\sum_{i=0}^ka_i(-x)^i$.
    Alternatively, since $m_T(x)∣χ_T(x)$ and $\deg m_T(x)=\deg χ_T(x)$, they are same up to a constant. Comparing coefficients of $x^k$ we find $χ_T(x)=(-1)^km_T(x)$.
    $T$ is singular$⇔\det T=0⇔a_0=0$. Including the case that $T$ is nilpotent, $m_T(x)=x^k$.
  8. Let $𝒫=F[x]$ be the vector space of polynomials over the field $F$. Determine whether or not $𝒫 / ℳ$ is finite dimensional when $ℳ$ is
    (i) the subspace $𝒫_n$ of polynomial of degree less or equal $n$;
    (ii) the subspace $ℰ$ of even polynomials;
    (iii) the subspace $x^n 𝒫$ of all polynomials divisible by $x^n$.
    Solution.
    (i) $1,x,…,x^n$ is a basis for $𝒫_n$ which extends to a basis $1,x,…,x^n,x^{n+1},…$ for $𝒫$. So $x^{n+1}+𝒫_n,…$ is a basis for $𝒫 /𝒫_n$. So $𝒫 /𝒫_n$ is infinite dimensional.
    (ii) $1,x^2,…$ is a basis for $ℰ$ which extends to a basis $1,x,x^2,x^3,…$ for $𝒫$. So $x+ℰ,x^3+ℰ,…$ is a basis for $𝒫 /ℰ$. So $𝒫 /ℰ$ is infinite dimensional.
    (iii) $x^n,x^{n+1}…$ is a basis for $x^n 𝒫$ which extends to a basis $1,x,…,x^{n-1},x^n,x^{n+1}…$ for $𝒫$. So $1+x^n 𝒫,x+x^n 𝒫,…,x^{n-1}+x^n 𝒫$ is a basis for $𝒫 /x^n 𝒫$. So $𝒫 /x^n 𝒫$ is finite dimensional.
  9. Let $L: 𝒫 → 𝒫$ be given by$$L: f(x) ↦ x^2 f(x)$$ In each of the examples of the preceding question, decide whether $L$ induces a map of quotients $\bar{L}: 𝒫 / ℳ → 𝒫 / ℳ$. When it does, find a matrix representation of $\bar{L}$ with respect to a convenient basis of the quotient space.
    Solution.
    (i) $L(x^n)=x^{n+2}∉ℳ$, so ℳ is not invariant under $L$, so $L$ does not induce a map of quotients.
    (ii) a matrix representation of $\bar{L}$ with the basis $x+ℰ,x^3+ℰ,…$ is$$\pmatrix{0&0&0\\1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 &⋱}$$ (iii) a matrix representation of $\bar{L}$ with the basis $1+x^n 𝒫,x+x^n 𝒫,…,x^{n-1}+x^n 𝒫$ is$$\pmatrix{0&0&0&⋯&0&0&0\\0&0&0&⋯&0&0&0\\1 & 0 & 0&⋯&0&0&0\\ 0 & 1 & 0&⋯&0&0&0\\ 0 & 0 &1&⋯&0&0&0\\⋮&⋮&⋮&\ddots&⋮&⋮&⋮\\0&0&0&⋯&1&0&0}$$