目录

Linear Algebra Done Right 4th 习题的参考解答及思考(第3章)

第3章

章节3A

练习3A.1

题目:

Suppose \( b, c \in \mathbf{R} \). Define \( T: \mathbf{R}^3 \to \mathbf{R}^2 \) by \( T(x, y, z) = (2x - 4y + 3z + b, 6x + cxyz) \). Show that \( T \) is linear if and only if \( b = c = 0 \).

证明:

必要性:

如果\( T \)为线性映射,则\( \forall u = (u_1, u_2, u_3), v = (v_1, v_2, v_3) \in \mathbf{R}^3 \),有\( T(u + v) = T(u) + T(v) \),可得\( 2(u_1 + v_1) - 4(u_2 + v_2) + 3(u_3 + v_3) + b, 6(u_1 + v_1) + c(u_1 + v_1)(u_2 + v_2)(u_3 + v_3) = (2u_1 - 4u_2 + 3u_3 + b, 6u_1 + cu_1u_2u_3) + (2v_1 - 4v_2 + 3v_3 + b, 6v_1 + cv_1v_2v_3) \),可得\( 2b = b, c(u_1u_2u_3 + v_1v_2v_3) = c(u_1 + v_1)(u_2 + v_2)(u_3 + v_3) \),由\( 2b = b \),可得\( b = 0 \),假设\( c \neq 0 \),则由\( c(u_1u_2u_3 + v_1v_2v_3) = c(u_1 + v_1)(u_2 + v_2)(u_3 + v_3) \),可得\( (u_1u_2u_3 + v_1v_2v_3) = (u_1 + v_1)(u_2 + v_2)(u_3 + v_3) \)对所有 \( u = (u_1, u_2, u_3), v = (v_1, v_2, v_3) \in \mathbf{R}^3 \)都成立,特别的,我们令\( u = (1, 0, 0), v = (0, 1, 0) \),则\( (u_1u_2u_3 + v_1v_2v_3) = 2 \neq 0 = (u_1 + v_1)(u_2 + v_2)(u_3 + v_3) \)矛盾,因此假设不成立,可得\( c = 0 \)。

充分性:

如果\( b = c = 0 \),则\( T(x, y, z) = (2x - 4y + 3z, 6x) \),此时\( \forall u = (u_1, u_2, u_3), v = (v_1, v_2, v_3) \in \mathbf{R}^3 \),可得\( T(u + v) = 2(u_1 + v_1) - 4(u_2 + v_2) + 3(u_3 + v_3), 6(u_1 + v_1) \),而\( T(u) + T(v) = (2u_1 - 4u_2 + 3u_3, 6u_1) + (2v_1 - 4v_2 + 3v_3, 6v_1) = 2(u_1 + v_1) - 4(u_2 + v_2) + 3(u_3 + v_3), 6(u_1 + v_1) = T(u + v) \),可得\( T(u + v) = T(u) + T(v) \),类似的,易得\( \forall \lambda \in \mathbf{R}, v \in \mathbf{R}^3 \),有\( T(\lambda v) = \lambda T(v) \),综上,有\( T \)为线性映射。

证毕。

练习3A.2

题目:

Suppose \( b, c \in \mathbf{R} \). Define \( T: \mathcal{P}(\mathbf{R}) \to \mathbf{R}^2 \) by \[ Tp = (3p(4) + 5p'(6) + bp(1)p(2), \int_{-1}^{2} x^3p(x) \ dx + c \sin p(0)). \] Show that \( T \) is linear if and only if \( b = c = 0 \).

证明:

必要性:

如果\( T \)为线性映射,则\( \forall p, q \in \mathcal{P}(\mathbf{R}) \), \( T(p + q) = T(p) + T(q) \),可得\( 3(p + q)(4) + 5(p' + q')(6) + b((p + q)(1))((p + q)(2)) = (3p(4) + 5p'(6) + bp(1)p(2)) + (3q(4) + 5q'(6) + bq(1)q(2)) \) 以及\( \int_{-1}^{2} x^3(p + q)(x) \ dx + c \sin (p + q)(0) = (\int_{-1}^{2} x^3p(x) \ dx + c \sin p(0)) + (\int_{-1}^{2} x^3q(x) \ dx + c \sin q(0)) \)。

由前者解得\( b((p + q)(1))((p + q)(2)) = bp(1)p(2) + bq(1)q(2) = b(p(1)p(2) + q(1)q(2)) \),假设\( b \neq 0 \),则\( ((p + q)(1))((p + q)(2)) = p(1)p(2) + q(1)q(2) \),举两个特别的\( p, q \),易得矛盾,因此假设不成立,可得\( b = 0 \)。

由后者解得\( c \sin (p + q)(0) = c \sin p(0) + c \sin q(0) = c(\sin p(0) + \sin q(0)) \),同理,在假设\( c \neq 0 \)的情况下举反例易得矛盾,因此假设不成立,可得\( c = 0 \)。

充分性:

如果\( b = c = 0 \),则\( Tp = (3p(4) + 5p'(6), \int_{-1}^{2} x^3p(x) \ dx) \),易验证\( \forall p, q \in \mathcal{P}(\mathbf{R}), \forall \lambda \in \mathbf{R} \),有\( T(p + q) = Tp + Tq, T(\lambda p) = \lambda Tp \),可得\( T \)为线性映射。

证毕。

练习3A.3

Suppose that \( T \in \mathcal{L}(\mathbf{F}^n, \mathbf{F}^m) \). Show that there exist scalars \( A_{j, k} \in \mathbf{F} \) for \( j = 1, \dots, m \) and \( k = 1, \dots, n \) such that \[ T(x_1, \dots, x_n) = (A_{1, 1}x_1 + \dots + A_{1, n}x_n, \dots, A_{m, 1}x_1 + \dots + A_{m, n}x_n) \] for every \( (x_1, \dots, x_n) \in \mathbf{F}^n \). (This exercise shows that the linear map has the form promised in the second to last item of Example 3.3.)

证明:

一个线性映射由它对所有基向量的映射唯一决定,因此我们只需要关注对基向量的映射即可。

令\( u_1, \dots, u_n \)为\( \mathbf{F}^n \)的标准基,令\( v_1, \dots, v_m \)为\( \mathbf{F}^m \)的标准基, \( \forall 1 \leq i \leq n \),令\( T(u_i) = w_i \),则\( w_i \in \mathbf{F}^m \),可得\( \exists A_{1, i}, \dots, A_{m, i} \in \mathbf{F} \),使得\( T(u_i) = w_i = A_{1, i}v_1 + \dots + A_{m, i}v_m \),进而\( \forall (x_1, \dots, x_n) \in \mathbf{F}^n \),由\( T \)为线性映射,可得\( T((x_1, \dots, x_n)) = x_1 T(u_1) + \dots + x_n T(u_n) = x_1(A_{1, 1}v_1 + \dots + A_{m, 1}v_m) + \dots + x_n(A_{1, n}v_1 + \dots + A_{m, n}v_m) = (A_{1, 1}x_1 + \dots + A_{1, n}x_n, \dots, A_{m, 1}x_1 + \dots + A_{m, n}x_n) \)。

证毕。

练习3A.4

Suppose \( T \in \mathcal(V, W) \) and \( v_1, \dots, v_m \) is a list of vectors in \( V \) such that \( Tv_1, \dots, Tv_m \) is a linearly independent list in \( W \). Prove that \( v_1, \dots, v_m \) is linearly independent.

证明:

假设\( v_1, \dots, v_m \)线性相关,则存在不全为\( 0 \)的系数\( a_1, \dots, a_m \)满足 \( a_1v_1 + \dots + a_mv_m = 0 \),任取其中一个不为\( 0 \)的系数\( a_k \),则\( a_k v_k = -a_1v_1 - \dots - a_{k - 1}v_{k - 1} - a_{k + 1}v_{k + 1} - \dots - a_mv_{m} \),由\( T \)为线性映射,可得\( a_k Tv_k = -a_1Tv_1 - \dots - a_{k - 1}Tv_{k - 1} - a_{k + 1}Tv_{k + 1} - \dots - a_mTv_{m} \),移项可得\( a_1Tv_1 + \dots + a_mTv_m = 0 \),但这里\( a_k \neq 0 \),这意味着\( Tv_1, \dots, Tv_m \)线性相关,矛盾,因此假设不成立,有\( v_1, \dots, v_m \)线性相关。

证毕。

练习3A.5

题目:

Prove that \( \mathcal{L}(V, W) \) is a vector space, as was asserted in 3.6.

证明:

看着定义1.20一个个验证过去。

证明满足交换律:

\( \forall S, T \in \mathcal{L}(V, W) \),有\( \forall v \in V, (S + V)(v) = Sv + Tv = Tv + Sv = (T + S)v \),可得\( S + T = T + S \)。

证明满足结合律:

\( \forall T_1, T_2, T_3 \in \mathcal{L}(V, W) \),有\( \forall v \in V, ((T_1 + T_2) + T_3)(v) = (T_1v + T_2v) + T_3v = T_1v + (T_2v + T_3v) = (T_1 + (T_2 + T_3))(v) \),可得\( (T_1 + T_2) + T_3 = T_1 + (T_2 + T_3) \)。

证明加法单位元的存在性:

构造\( 0 \in \mathcal{L}(V, W) \),\( \forall v \in V \),令\( 0(v) = 0 \),则易得\( \forall S \in \mathcal{L}(V, W) \), \( S + 0 = S \)。

证明加法逆元的存在性:

\( \forall S \in \mathcal{L}(V, W) \),构造\( T \in \mathcal{L}(V, W) \), \( \forall v \in V \),令\( T(v) = -S(v) \),易得\( S + T = 0 \)。

证明乘法单位元的存在性:

\( \forall S \in \mathcal{L}(V, W) \),有\( \forall v \in V \),\( (1S)(v) = 1(Sv) = Sv \),可得\( 1S = S \)。

证明满足分配律:

\( \forall S, T \in \mathcal{L}(V, W), \forall a, b \in \mathbf{F} \)。

\( \forall v \in V \),有\( (a(S + T))(v) = a((S + T)(v)) = a(Sv + Tv) = aSv + aTv \),可得\( a(S + T) = aS + aT \)。

\( \forall v \in V \),有\( ((a + b)S)(v) = (a + b)(Sv) = aSv + bSv \),可得\( (a + b)S = aS + bS \)。

综上,根据定义1.20,可得\( \mathcal{L}(V, W) \)为向量空间。

证毕。

练习3A.6

题目:

Prove that multiplication of linear maps has the associative, identity, and distributive properties asserted in 3.8.

3.8的内容:

associativity

\( (T_1T_2)T_3 = T_1(T_2T_3) \) whenever \( T_1, T_2 \), and \( T_3 \) are linear maps such that the products make sense (meaning \( T_3 \) maps into the domain of \( T_2 \) and \( T_2 \) maps into the domain of \( T_1 \)).

identity

\( TI = IT = T \) whenever \( T \in \mathcal{L}({V, W}) \); here the first \( I \) is the identity operator on \( V \), and the second \( I \) is the identity operator on \( W \).

distributive properties

\( (S_1 + S_2)T = S_1T + S_2T \) and \( S(T_1 + T_2) = ST_1 + ST_2 \) whenever \( T, T_1, T_2 \in \mathcal{L}(U, V) \) and \( S, S_1, S_2 \in \mathcal{L}(V, W) \).

证明:

证明满足结合律:

线性映射的乘法就是函数的复合,而我们知道函数的复合是复合结合律的,简单写的话,到这里就可以说结合律成立了,详细点的话,如下:

\( \forall T_1 \in \mathcal{L}(V_1, V_2), T_2 \in \mathcal{L}(V_2, V_3), T_3 \in \mathcal{L}(V_3, V_4) \),我们有\( \forall v \in V_1, ((T_1T_2)T_3)(x) = (T_1T_2)(T_3x) = T_1(T_2(T_3x)) = T_1((T_2T_3)(x)) = (T_1(T_2T_3))(x) \),综上,有\( (T_1T_2)T_3 = T_1(T_2T_3) \)。

证明乘法单位元的存在性:

构造映射\( I_1 \in \mathcal{L}(V, V), \forall v \in V, I_1(v) = v \),构造映射\( I_2 \in \mathcal{L}(W, W), \forall w \in W, I_2(w) = w \),易得\( \forall v \in \mathcal{L}(V, W), TI = IT = T \)。

证明满足分配律:

\( \forall T, T_1, T_2 \in \mathcal{L}(U, V), \forall S, S_1, S_2 \in \mathcal{L}(V, W) \):

\( \forall u \in U, ((S_1 + S_2)T)(u) = (S_1 + S_2)(Tu) = S_1(Tu) + S_2(Tu) = (S_1T)(u) + (S_2T)(u) \),综上,有\( (S_1 + S_2)T = S_1T + S_2T \)。

\( \forall u \in U, S(T_1 + T_2)(u) = S((T_1 + T_2)(u)) = S(T_1u + T_2u) = S(T_1u) + S(T_2u) = (ST_1)(u) + (ST_2)(u) \),综上,有\( S(T_1 + T_2) = ST_1 + ST_2 \)。

证毕。

练习3A.7

题目:

Show that every linear map from a one-dimensional vector space to itself is multiplication by some scalar. More precisely, prove that if \( \dim V = 1 \) and \( T \in \mathcal{L}(V) \), then there exists \( \lambda \in \mathbf{F} \) such that \( Tv = \lambda v \) for all \( v \in V \).

证明:

给定任意满足题述条件的线性映射\( T \),因为\( \dim V = 1 \),可得\( V \)的基中的向量数量为\( 1 \),任取\( V \)的基\( w \),则\( \forall v \in V \),存在\( \alpha \in \mathbf{F} \),使得\( v = \alpha w \),因为\( T \in \mathcal{L}(V) \),有\( Tw \in V \),可得存在\( \lambda \in \mathbf{F} \),使得\( Tw = \lambda w \),进而可得\( Tv = T(\alpha w) = \alpha (Tw) = \alpha (\lambda w) = \lambda (\alpha w) = \lambda v \)。

证毕。

练习3A.8

题目:

Give an example of a function \( \varphi: \mathbf{R}^2 \to \mathbf{R} \) such that \( \varphi(av) = a \varphi(v) \) for all \( a \in \mathbf{R} \) and all \( v \in \mathbf{R}^2 \) but \( \varphi \) is not linear. (This exercise and the next exercise show that neither homogeneity nor additivity alone is enough to imply that a function is a linear map.)

解答:

构造\( \varphi: \mathbf{R}^2 \to \mathbf{R} \), \( \forall (x, y) \in \mathbf{R}^2 \),如果\( y \neq 0 \),则令\( \varphi((x, y)) = \dfrac{x^2}{y} \),如果\( y = 0 \),则令\( \varphi((x, y)) = 0 \),于是\( \forall a \in \mathbf{R}, (x, y) \in \mathbf{R}^2, \varphi(a(x, y)) = \varphi((ax, ay)) = \dfrac{a^2x^2}{ay} = a\dfrac{x^2}{y} = a \varphi((x, y)) \),但是\( \varphi((1, 2) + (3, 4)) = \varphi((4, 6)) = \dfrac{16}{6}, \varphi((1, 2)) = \dfrac{1}{2}, \varphi((3, 4)) = \dfrac{9}{4} \),可得\( \varphi((1, 2)) + \varphi((3, 4)) = \dfrac{11}{4} = \dfrac{33}{12} \neq \dfrac{32}{12} = \dfrac{16}{6} = \varphi((1, 2) + (3, 4)) \),这意味着\( \varphi \)非线性。

练习3A.9

题目:

Give an example of a function \( \varphi: \mathbf{C} \to \mathbf{C} \) such that \( \varphi(w + z) = \varphi(w) + \varphi(z) \) for all \( w, z \in \mathbf{C} \) but \( \varphi \) is not linear. (Here C is thought of as a complex vector space. There also exists a function \( \varphi: \mathbf{R} \to \mathbf{R} \) such that \( \varphi \) satisfies the additivity condition above but \( \varphi \) is not linear. However, showing the existence of such a function involves considerably more advanced tools.)

解答:

构造\( \varphi: \mathbf{C} \to \mathbf{C} \), \( \forall a + bi \in \mathbf{C} \),令\( \varphi(a + bi) = a \),则\( \forall w = (a + bi), z = (c + di) \in \mathbf{C} \),有\( \varphi(w + z) = \varphi((a + c) + (b + d)i) = a + c = \varphi(w) + \varphi(z) \),但是\( i \varphi(1) = i \neq \varphi(i \times 1) = \varphi(i) = 0 \),这意味着\( \varphi \)非线性。

练习3A.10

题目:

Prove or give a counterexample: If \( q \in \mathcal{P}(\mathbf{R}) \) and \( T: \mathcal{P}({\mathbf{R}}) \to \mathcal{P}({\mathbf{R}}) \) is defined by \( Tp = q \circ p \), then \( T \) is a linear map. (The function \( T \) defined here differs from the function \( T \) defined in the last bullet point of 3.3 by the order of the functions in the compositions.)

Last bullet point of 3.3的内容:

composition

Fix a polynomial \( q \in \mathcal{P}({\mathbf{R}}) \). Define a linear map \( T \in \mathcal{L}(\mathcal{P}({\mathbf{R}})) \) by \( (Tp)(x) = p(q(x)) \).

解答:

不成立,令\( q(x) = x^2 \),\( p(x) = x \),则\( (T(2p))(x) = q(2p(x)) = 4x^2 \),而\( (2T(p))(x) = 2q(p(x)) = 2x^2 \),可得\( T(2p) \neq 2Tp \),这说明\( T \)非线性映射。

练习3A.11

题目:

Suppose \( V \) is finite-dimensional and \( T \in \mathcal{L}(V) \). Prove that \( T \) is a scalar multiple of the identity if and only if \( ST = TS \) for every \( S \in \mathcal{L}(V) \).

证明:

必要性:

如果\( T \)为恒等函数的标量倍,则\( \exists \lambda \in F \),使得\( \forall v \in V, T(v) = \lambda v \),此时\( \forall S \in \mathcal{L}(V) \),有\( (ST)(v) = S(T(v)) = S(\lambda v) = \lambda S(v) \),且\( (TS)(v) = T(S(v)) = \lambda S(v) \),可得\( ST = TS \)。

充分性:

注: 这部分直接用了xen的回答,Robert Israel回答中的解法更简洁,但是依赖于存在\( V \)上的非平凡线性函数映射(linear functional),但是书中到该练习所在页数为止,并没有定理支持该函数的存在性,所以这里用了xen的解法。

令\( n = \dim V \),令\( e_1, \dots, e_n \)为\( V \)的基。

\( \forall 1 \leq i \leq n \),构造\( S_i \in \mathcal{L}(V) \), \( \forall a_1 e_1 + \dots + a_n e_n \in V \),令\( S_i(a_1 e_1 + \dots + a_n e_n) = a_i e_i \),易验证\( S_i \in \mathcal{L}(V) \),则 \( \forall 1 \leq i \leq n, T e_i = T S_i e_i = S_i T e_i = S_i (b_{i, 1} e_1 + \dots + b_{i, n} e_n) = b_{i, 1} S_i e_1 + \dots + b_{i, n} S_i e_n = b_{i, i} e_n \),我们接下来需要证明\( \forall 1 \leq i, j \leq n, i \neq j, b_{i, i} = b_{j, j} \)。

\( \forall 1 \leq i, j \leq n, i \neq j \),构造\( S_{i, j} \in \mathcal{L}(V) \), \( \forall a_1 e_1 + \dots + a_n e_n \in V \),令\( S_{i, j}(a_1 e_1 + \dots + a_n e_n) = a_j e_i + a_i e_j \),易验证\( S_{i, j} \in \mathcal{L}(V) \),则 \( TS(e_i + e_j) = T(e_i + e_j) = Te_i + Te_j = b_{i, i} e_i + b_{j, j} e_j \), \( ST(e_i + e_j) = S(b_{i, i} e_i + b_{j, j} e_j) = b_{j, j} e_j + b_{i, i} e_i \),而\( TS(e_i + e_j) = ST(e_i + e_j) \),可得\( b_{i, i} e_i + b_{j, j} e_j = b_{j, j} e_j + b_{i, i} e_i \),于是有\( (b_{i, i} - b_{j, j}) e_i + (b_{j, j} - b_{i, i}) e_j = 0 \),由于\( e_i, e_j \)线性无关,可得\( b_{i, i} - b_{j, j} = b_{j, j} - b_{i, i} = 0 \),有\( b_{i, i} = b_{j, j} \)。

综上,我们有\( \forall v = a_1 e_1 + \dots + a_n e_n \in V, Tv = a_1 Te_1 + \dots + a_n Te_n = b_{1, 1} a_1 e_1 + \dots + b_{n, n} a_n e_n = b_{1, 1} a_1 e_1 + \dots + b_{1, 1} a_n e_n = b_{1, 1}(a_1 e_1 + \dots + a_n e_n) = b_{1, 1} v \),这意味着\( T \)为恒等函数的标量倍。

证毕。

练习3A.12

题目:

Suppose \( U \) is a subspace of \( V \) with \( U \neq V \). Suppose \( S \in \mathcal{L}(U, W) \) and \( S \neq 0 \) (which means that \( Su \neq 0 \) for some \( u \in U \)). Define \( T: V \to W \) by \[ Tv = \begin{cases} Sv \quad \text{if } v \in U, \\
0 \quad \text{if } v \in V \text{ and } v \notin U. \end{cases} \] Prove that \( T \) is not a linear map on \( V \).

证明:

由于\( U \neq V \),因此\( \exists w \in V \setminus U \)(注:向量空间一定非空),由于\( S \neq 0 \),故\( \exists v \in U, Sv \neq 0 \),我们断言\( v + w \notin U \),假设\( v + w \in U \),则\( (v + w) - v = w \in U \),这和\( w \notin U \)矛盾,因此假设不成立,可得\( v + w \notin U \),此时考虑\( T(v + w) \),假设\( T \)为线性映射,则\( T(v + w) = T(v) + T(w) = Sv + 0 = Sv \neq 0 \),但由于\( v + w \notin U \),故根据\( T \)的定义,实际上\( T(v + w) = 0 \),这和\( T(v + w) = Sv \neq 0 \)矛盾,因此假设不成立,有\( T \)非线性映射。

证毕。

练习3A.13

题目:

Suppose \( V \) is finite-dimensional. Prove that every linear map on a subspace of \( V \) can be extended to a linear map on \( V \). In other words, show that if \( U \) is subspace of \( V \) and \( S \in \mathcal{L}(U, W) \), then there exists \( T \in \mathcal{L}(V, W) \) such that \( Tu = Su \) for all \( u \in U \). (The result in this exercise is used in the proof of 3.125.)

证明:

根据2.33,可得存在\( V \)的子空间\( U' \),使得\( V = U \oplus U' \),取\( \mathcal{L}(U', W) \)中的零元\( 0 \),构造\( T \in \mathcal{L}(V, W) \),\( \forall v \in V \),由\( V = U \oplus U' \) 可得存在 唯一的 \( u \in U, u' \in U' \),使得\( v = u + u' \),我们令\( Tv = S(u) + 0(u') \)(注:这里直和表示的唯一性很重要,不然我们函数的定义可能会有歧义,只是可能有歧义,因为即使表示不唯一,如果所有表示的映射值都一样,那函数的定义也没有问题),我们验证下\( T \)为线性映射:

验证加性:

\( \forall v_1 = u_1 + u'_1, v_2 = u_2 + u'_2 \in V \),有\( T(v_1 + v_2) = T((u_1 + u_2) + (u'_1 + u'_2)) = S(u_1 + u_2) + 0(u'_1 + u'_2) = S(u_1) + S(u_2) + 0(u'_1) + 0(u'_2) = (S(u_1) + 0(u'_1)) + (S(u_2) + 0(u'_2)) = T(v_1) + T(v_2) \)。

验证齐次性:

\( \forall \lambda \in \mathbf{F}, v = u + u' \in V \),有\( T(\lambda v) = T(\lambda (u + u')) = T(\lambda u + \lambda u') = S(\lambda u) + 0(\lambda u') = \lambda S(u) + \lambda 0(u') = \lambda (S(u) + 0(u')) = \lambda T(v) \)。

综上,\( T \)为线性映射。

证毕。

练习3A.14

题目:

Suppose \( V \) is finite-dimensional with \( \dim V > 0 \), and suppose \( W \) is infinite-dimensional. Prove that \( \mathcal{L}(V, W) \) is infinite-dimensional.

证明:

我们使用练习2A.17的结论来证明。

记\( \dim V = n, v_1, \dots, v_n \)为\( V \)的基,由\( W \)无限维以及练习2A.17,可得存在序列\( w_1, w_2, \dots \)满足\( \forall i \geq 1, w_i \in W \)且 \( \forall m \geq 1, w_1, \dots, w_m \)线性无关。针对\( \forall m \geq 1 \),根据3.4,可得存在线性映射\( S_m \in \mathcal{L}(V, W) \),满足\( \forall 1 \leq i \leq n, S_m v_i = w_m \),我们接下来证明\( \forall m \geq 1 \),\( S_1, \dots, S_m \)线性无关,进一步使用练习2A.17就可以说明\( \mathcal{L}(V, W) \)无限维了: \( \forall m \geq 1 \),\( \forall a_1, \dots, a_m \in \mathbf{F} \)满足\( a_1S_1 + \dots + a_mS_m = 0 \) (注:这里右边的\( 0 \)是个线性映射,不是标量),我们要证明\( a_1 = \dots = a_m = 0 \),这里考虑\( a_1S_1 + \dots + a_mS_m \)对\( v_1 \)的映射就行了(注意:\( \dim V > 0 \),所以\( V \)的基的数量一定\( > 0 \),进而\( v_1 \)是存在的),我们有\( (a_1S_1 + \dots + a_mS_m)(v_1) = a_1w_1 + \dots + a_mw_m = 0 \),由\( w_1, \dots, w_m \)线性无关,可得\( a_1 = \dots = a_m = 0 \),至此可得,\( \forall m \geq 1 \),\( S_1, \dots, S_m \)线性无关,进而根据练习2A.17,可得\( \mathcal{L}(V, W) \)无限维。

证毕。

练习3A.15

题目:

Suppose \( v_1, \dots, v_m \) is linearly dependent list of vectors in \( V \). Suppose also that \( W \neq \{ 0 \} \). Prove that there exist \( w_1, \dots, w_m \in W \) such that no \( T \in \mathcal{L}(V, W) \) satisfies \( Tv_k = w_k \) for each \( k = 1, \dots, m \).

注:

根据3.4,我们知道如果\( v_1, \dots, v_m \)线性无关,则我们可以随便给\( v_i \)取对应的映射值\( w_i \),永远唯一存在一个线性映射满足我们对映射关系的选择,本练习则说明,如果\( v_1, \dots, v_m \)线性相关,则我们无法随便给\( v_i \)取对应的映射值\( w_i \),有可能会产生冲突的映射关系。

证明:

由\( v_1, \dots, v_m \)线性相关,可得存在不全为\( 0 \)的系数\( a_1, \dots, a_m \in \mathbf{F} \),使得\( a_1v_1 + \dots + a_mv_m = 0 \),这里记该不为\( 0 \)的系数为\( a_k \),则\( v_k = \dfrac{\sum_{i \in \{ 1, \dots, m \} \setminus \{ k \}} -a_iv_i}{a_k} \),针对\( \forall 1 \leq i \leq m, i \neq k \),我们任取\( w_i \in W \),此时考虑任意\( T \in \mathcal{L}(V, W) \)满足\( \forall 1 \leq i \leq m, i \neq k, T v_i = w_i \) (注意,我们这里没有要求是否\( T v_k = w_k \),因为到这里,我们还没有选择好\( w_k \)),针对该线性映射\( T \),有\( Tv_k = \dfrac{\sum_{i \in \{ 1, \dots, m \} \setminus \{ k \}} -a_i T v_i}{a_k} \),由于\( W \neq \{ 0 \} \),所以它至少有两个元素,故一定存在元素\( w' \in W \)满足 \( w' \neq Tv_k \),我们令\( w_k = w' \),则\( Tv_k \neq w_k \),综上,任意满足了\( \forall 1 \leq i \leq m, i \neq k, T v_i = w_i \)的线性映射\( T \),它一定无法进一步满足\( Tv_k = w_k \),故满足\( \forall 1 \leq i \leq m, T v_i = w_i \)的线性映射\( T \) 不存在。

证毕。

练习3A.16

题目:

Suppose \( V \) is finite-dimensional with \( \dim V > 1 \). Prove that there exist \( S, T \in \mathcal{L}(V) \) such that \( ST \neq TS \).

证明:

令\( v_1, \dots, v_n \)为\( V \)的基,由\( \dim V > 1 \),可得\( n \geq 2 \),构造\( S \in \mathcal{L}(V) \),令\( Sv_1 = v_2, Sv_2 = v_1 \),针对\( \forall i > 2 \),则令\( Sv_i = v_i \) (根据3.4,可得存在唯一的线性映射\( S \)满足我们指定的映射关系,下面也类似,不再赘述),构造\( T \in \mathcal{L}(V) \),令\( Tv_1 = v_1, Tv_2 = 2v_2 \),针对\( \forall i > 2 \),则令\( Tv_i = v_i \),此时可得\( S(T(v_1)) = v_2, T(S(v_1)) = 2v_2 \),由于\( v_2 \neq 0 \),因此\( 2v_2 \neq v_2 \),于是\( S(T(v_1)) \neq T(S(v_1)) \),这意味着\( ST \neq TS \)。

证毕。

练习3A.17

题目:

Suppose \( V \) is finite-dimensional. Show that the only two-sided ideals of \( \mathcal{L}(V) \) are \( \{ 0 \} \) and \( \mathcal{L}(V) \). (A subspace \( \varepsilon \) of \( \mathcal{L}(V) \) is called a two-sided ideal of \( \mathcal{L}(V) \) if \( TE \in \varepsilon \) and \( ET \in \varepsilon \) for all \( E \in \varepsilon \) and all \( T \in \mathcal{L}(V) \).)

证明:

本答案参考了nehc0的解法。

我们首先验证下\( \{ 0 \}, \mathcal{L}(V) \)为\( \mathcal{L}(V) \)的two-sided ideal。

验证\( \{ 0 \} \)为two-sided ideal: \( \forall E \in \{ 0 \} \),有\( E = 0 \),进而易得\( \forall T \in \mathcal{L}(V), TE = 0 \in \{ 0 \}, ET = 0 \in \{ 0 \} \)。

针对\( \mathcal{L}(V) \),则明显它为two-sided ideal。

最后我们要证明任意\( \mathcal{L}(V) \)的子空间\( \varepsilon \),如果\( \varepsilon \neq \{ 0 \}, \varepsilon \neq \mathcal{L}(V) \),则\( \varepsilon \)不为two-sided ideal:

记\( n = \dim V \),对\( n \)进行数学归纳,当\( n = 0 \)时,\( \varepsilon = \{ 0 \} \),这和\( \varepsilon \neq \{ 0 \} \)矛盾,因此这种情况不用考虑,归纳假设当\( n = k \)时,命题成立,当\( n = k + 1 \)时,我们记\( V \)的基为\( v_1, \dots, v_n \), \( \forall i, j \in \{ 1, \dots, n \} \),我们定义\( T_{ij} \in \mathcal{L}(V) \), \( \forall v = a_1v_1 + \dots + a_nv_n \in V \),令\( T_{ij}(v) = a_jv_i \),容易验证\( T_{ij} \in \mathcal{L}(V) \),现在用反证法,假设\( \varepsilon \)为two-sided ideal,由\( \varepsilon \neq \{ 0 \} \),可得\( \exists E \in \varepsilon \)满足 \( \exists i \in \{ 1, \dots, n \}, Ev_i \neq 0 \),记\( Ev_i = \sum_{j \in L} b_jv_j, L \subseteq \{ 1, \dots, n \}, b_j \in \mathbf{F}, b_j \neq 0 \), \( Ev_i \)一定可以表达成这种形式,这是因为\( v_1, \dots, v_n \)为\( V \)的基,而\( Ev_i \in V \),故\( Ev_i \)一定能表达成\( v_1, \dots, v_n \)的线性组合,进一步过滤掉为\( 0 \)的系数就可以做到了,这里我们希望从\( L \)中随便取个元素,但是\( L \)本身可能为空,故我们先讨论它为空的情况,如果它为空,则说明\( v_i = 0 \),但这意味着\( Ev_i = 0 \),和\( Ev_i \neq 0 \)矛盾,因此\( L \)不可能为空,我们任取\( j \in L \),\( \forall k \in \{ 1, \dots, n \} \),定义\( S_k = \dfrac{1}{b_j} T_{kj} E T_{ik} \),则\( \forall v = a_1v_1 + \dots + a_nv_n \in V \), \( S_k(v) = (\dfrac{1}{b_j} T_{kj} E)(T_{ik}(v)) = (\dfrac{1}{b_j} T_{kj} E)(a_kv_i) = (\dfrac{1}{b_j} T_{kj} )(a_kEv_i) = (\dfrac{a_k}{b_j} T_{kj})(\sum_{j \in L} b_jv_j) = (\dfrac{a_k}{b_j})(b_jv_k) = a_kv_k \),根据two-sided ideal的定义,可得\( T_{kj} E T_{ik} \in \varepsilon \),而\( S_k = \dfrac{1}{b_j} T_{kj} E T_{ik} \),还差一个标量乘以\( \dfrac{1}{b_j} \),它可以通过线性映射来实现,具体的,构造线性映射\( O \in \mathcal{L}(V) \),令\( O(v) = \dfrac{1}{b_j}v \),易验证\( O \)为线性映射,进而根据two-sided ideal的定义,可得\( S_k = \dfrac{1}{b_j} T_{kj} E T_{ik} = O T_{kj} E T_{ik} \in \varepsilon \),由\( S_k(v) = a_kv_k \),易得\( S_1 + \dots + S_n = I \),而\( \varepsilon \)为向量空间,可得\( I \in \varepsilon \),由于\( \varepsilon \neq \mathcal{L}(V) \),可得存在\( T \in \mathcal{L}(V), T \notin \varepsilon \),但由two-sided ideal的定义,可得\( IT = T \in \varepsilon \),矛盾,因此一开始假设不成立,\( \varepsilon \)不为two-sided ideal。

证毕。

参考资料

未完待续,慢慢更新