Section 8B

Generalized Eigenspace Decomposition

[23/0%]
Show
LaTeX
Problem 8B.1

Define T∈L(C2)T \in \mathcal{L}\left(\mathbb {C}^{2}\right) by T(w,z)=(−z,w)T(w, z) = (-z, w). Find the generalized eigenspaces corresponding to the distinct eigenvalues of TT.

?
Problem 8B.2

Suppose T∈L(V)T \in \mathcal{L}(V) is invertible. Prove that G(λ,T)=G(1λ,T−1)G(\lambda , T) = G\left(\frac{1}{\lambda }, T^{-1}\right) for every λ∈F\lambda \in \mathbf{F} with λ≠0\lambda \neq 0.

?
Problem 8B.3

Suppose T∈L(V)T \in \mathcal{L}(V). Suppose S∈L(V)S \in \mathcal{L}(V) is invertible. Prove that TT and S−1TSS^{-1} T S have the same eigenvalues with the same multiplicities.

?
Problem 8B.4

Suppose dim⁡V≥2\operatorname {dim} V \geq 2 and T∈L(V)T \in \mathcal{L}(V) is such that null⁡Tdim⁡V−2≠null⁡Tdim⁡V−1\operatorname {null}T^{\operatorname {dim} V-2} \neq \operatorname {null} T^{\operatorname {dim} V-1}. Prove that TT has at most two distinct eigenvalues.

?
Problem 8B.5

Suppose T∈L(V)T \in \mathcal{L}(V) and 3 and 8 are eigenvalues of TT. Let n=dim⁡Vn = \operatorname {dim} V. Prove that V=(null⁡Tn−2)⊕(range⁡Tn−2)V = \left(\operatorname {null} T^{n-2}\right) \oplus \left(\operatorname {range} T^{n-2}\right).

?
Problem 8B.6

Suppose T∈L(V)T \in \mathcal{L}(V) and λ\lambda is an eigenvalue of TT. Explain why the exponent of z−λz-\lambda in the factorization of the minimal polynomial of TT is the smallest positive integer mm such that (T−λI)m∣G(λ,T)=0\left.(T-\lambda I)^{m}\right|_{G(\lambda , T)} = 0.

?
Problem 8B.7

Suppose T∈L(V)T \in \mathcal{L}(V) and λ\lambda is an eigenvalue of TT with multiplicity dd. Prove that G(λ,T)=null⁡(T−λI)dG(\lambda , T) = \operatorname {null}(T-\lambda I)^{d}.

?
Remark.
?

If d<dim⁡Vd < \operatorname {dim} V, then this exercise improves 8.20. Theorem 8.20 (description of generalized eigenspaces): suppose T∈L(V)T \in \mathcal{L}(V) and λ∈F\lambda \in \mathbf{F}. Then G(λ,T)=null⁡(T−λI)dim⁡VG(\lambda , T) = \operatorname {null}(T-\lambda I)^{\operatorname {dim} V}.

Problem 8B.8

Suppose T∈L(V)T \in \mathcal{L}(V) and λ1,…,λm\lambda_{1}, \ldots , \lambda_{m} are the distinct eigenvalues of TT. Prove that

V=G(λ1,T)⊕⋯⊕G(λm,T) V = G\left(\lambda _{1}, T\right) \oplus \cdots \oplus G\left(\lambda _{m}, T\right)

if and only if the minimal polynomial of TT equals (z−λ1)k1…(z−λm)km\left(z-\lambda_{1}\right)^{k_{1}} \ldots \left(z-\lambda_{m}\right)^{k_{m}} for some positive integers k1,…,kmk_{1}, \ldots , k_{m}.

?
Remark.
?

The case F=C\mathbf{F} = \mathbb {C} follows immediately from 5.27(b)5.27(b) and the generalized eigenspace decomposition (8.22); thus this exercise is interesting only when F=R\mathrm{F} = \mathbb {R}. Theorem 5.27(b): if VV is a finite-dimensional complex vector space and T∈L(V)T \in \mathcal{L}(V), then the minimal polynomial of TT has the form (z−λ1)⋯(z−λm)(z-\lambda_{1}) \cdots (z-\lambda_{m}), where λ1,…,λm\lambda_{1}, \ldots , \lambda_{m} is a list of all eigenvalues of TT, possibly with repetitions. Theorem 8.22 (generalized eigenspace decomposition): suppose F=C\mathbf{F} = \mathbb {C} and T∈L(V)T \in \mathcal{L}(V), and let λ1,…,λm\lambda_{1}, \ldots , \lambda_{m} be the distinct eigenvalues of TT. Then G(λk,T)G(\lambda_{k}, T) is invariant under TT and (T−λkI)∣G(λk,T)(T-\lambda_{k} I)|_{G(\lambda_{k}, T)} is nilpotent for each k=1,…,mk = 1, \ldots , m, and V=G(λ1,T)⊕⋯⊕G(λm,T)V = G(\lambda_{1}, T) \oplus \cdots \oplus G(\lambda_{m}, T).

Problem 8B.9

Suppose F=C\mathbf{F} = \mathbb {C} and T∈L(V)T \in \mathcal{L}(V). Prove that there exist D,N∈L(V)D, N \in \mathcal{L}(V) such that T=D+NT = D+N, the operator DD is diagonalizable, NN is nilpotent, and DN=NDD N = N D.

?
Problem 8B.10

Suppose VV is a complex inner product space, e1,…,ene_{1}, \ldots , e_{n} is an orthonormal basis of TT, and T∈L(V)T \in \mathcal{L}(V). Let λ1,…,λn\lambda_{1}, \ldots , \lambda_{n} be the eigenvalues of TT, each included as many times as its multiplicity. Prove that

∣λ1∣2+⋯+∣λn∣2≤∥Te1∥2+⋯+∥Ten∥2. \left|\lambda _{1}\right|^{2}+\cdots +\left|\lambda _{n}\right|^{2} \leq \left\| T e_{1}\right\| ^{2}+\cdots +\left\| T e_{n}\right\| ^{2}.

See the comment after Exercise 5 in Section 7A. Exercise 7A.5: suppose T∈L(V,W)T \in \mathcal{L}(V, W); suppose e1,…,ene_{1}, \ldots , e_{n} is an orthonormal basis of VV and f1,…,fmf_{1}, \ldots , f_{m} is an orthonormal basis of WW. Prove that ∥Te1∥2+⋯+∥Ten∥2=∥T∗f1∥2+⋯+∥T∗fm∥2\left\| Te_{1}\right\|^{2}+\cdots +\left\| Te_{n}\right\|^{2} = \left\| T^{*}f_{1}\right\|^{2}+\cdots +\left\| T^{*}f_{m}\right\|^{2}. The comment after that exercise notes: the numbers ∥Te1∥2,…,∥Ten∥2\left\| Te_{1}\right\|^{2}, \ldots , \left\| Te_{n}\right\|^{2} in that equation depend on the orthonormal basis e1,…,ene_{1}, \ldots , e_{n}, but the right side does not depend on e1,…,ene_{1}, \ldots , e_{n}; thus the equation shows that the sum on the left side does not depend on which orthonormal basis e1,…,ene_{1}, \ldots , e_{n} is used.

?
Problem 8B.11

Give an example of an operator on C4\mathbb {C}^{4} whose characteristic polynomial equals (z−7)2(z−8)2(z-7)^{2}(z-8)^{2}.

?
Problem 8B.12

Give an example of an operator on C4\mathbb {C}^{4} whose characteristic polynomial equals (z−1)(z−5)3(z-1)(z-5)^{3} and whose minimal polynomial equals (z−1)(z−5)2(z-1)(z-5)^{2}.

?
Problem 8B.13

Give an example of an operator on C4\mathbb {C}^{4} whose characteristic and minimal polynomials both equal z(z−1)2(z−3)z(z-1)^{2}(z-3).

?
Problem 8B.14

Give an example of an operator on C4\mathbb {C}^{4} whose characteristic polynomial equals z(z−1)2(z−3)z(z-1)^{2}(z-3) and whose minimal polynomial equals z(z−1)(z−3)z(z-1)(z-3).

?
Problem 8B.15

Let TT be the operator on C4\mathbb {C}^{4} defined by T(z1,z2,z3,z4)=(0,z1,z2,z3)T\left(z_{1}, z_{2}, z_{3}, z_{4}\right) = \left(0, z_{1}, z_{2}, z_{3}\right). Find the characteristic polynomial and the minimal polynomial of TT.

?
Problem 8B.16

Let TT be the operator on C6\mathbb {C}^{6} defined by

T(z1,z2,z3,z4,z5,z6)=(0,z1,z2,0,z4,0). T\left(z_{1}, z_{2}, z_{3}, z_{4}, z_{5}, z_{6}\right) = \left(0, z_{1}, z_{2}, 0, z_{4}, 0\right).

Find the characteristic polynomial and the minimal polynomial of TT.

?
Problem 8B.17

Suppose F=C\mathbf{F} = \mathbb {C} and P∈L(V)P \in \mathcal{L}(V) is such that P2=PP^{2} = P. Prove that the characteristic polynomial of PP is zm(z−1)nz^{m}(z-1)^{n}, where m=dim⁡null⁡Pm = \operatorname {dim}\operatorname {null}P and n=dim⁡n = \operatorname {dim} range PP.

?
Problem 8B.18

Suppose T∈L(V)T \in \mathcal{L}(V) and λ\lambda is an eigenvalue of TT. Explain why the following four numbers equal each other.

?
(a)

The exponent of z−λz-\lambda in the factorization of the minimal polynomial of TT.

(b)

The smallest positive integer mm such that (T−λI)m∣G(λ,T)=0\left.(T-\lambda I)^{m}\right|_{G(\lambda , T)} = 0.

(c)

The smallest positive integer mm such that

null⁡(T−λI)m=null⁡(T−λI)m+1 \operatorname {null}(T-\lambda I)^{m} = \operatorname {null}(T-\lambda I)^{m+1}
(d)

The smallest positive integer mm such that

range⁡(T−λI)m=range⁡(T−λI)m+1 \operatorname {range}(T-\lambda I)^{m} = \operatorname {range}(T-\lambda I)^{m+1}
Problem 8B.19

Suppose F=C\mathbf{F} = \mathbb {C} and S∈L(V)S \in \mathcal{L}(V) is a unitary operator. Prove that the constant term in the characteristic polynomial of SS has absolute value 1.

?
Problem 8B.20

Suppose that F=C\mathbf{F} = \mathbb {C} and V1,…,VmV_{1}, \ldots , V_{m} are nonzero subspaces of VV such that

V=V1⊕⋯⊕Vm V = V_{1} \oplus \cdots \oplus V_{m}

Suppose T∈L(V)T \in \mathcal{L}(V) and each VkV_{k} is invariant under TT. For each kk, let pkp_{k} denote the characteristic polynomial of T∣Vk\left.T\right|_{V_{k}}. Prove that the characteristic polynomial of TT equals p1⋯pmp_{1} \cdots p_{m}.

?
Problem 8B.21

Suppose p,q∈P(C)p, q \in \mathcal{P}(\mathbb {C}) are monic polynomials with the same zeros and qq is a polynomial multiple of pp. Prove that there exists T∈L(Cdeg⁡q)T \in \mathcal{L}\left(\mathbb {C}^{\operatorname {deg} q}\right) such that the characteristic polynomial of TT is qq and the minimal polynomial of TT is pp.

?
Remark.
?

This exercise implies that every monic polynomial is the characteristic polynomial of some operator.

Problem 8B.22

Suppose AA and BB are block diagonal matrices of the form

A=[A10⋱0Am],B=[B10⋱0Bm] A =\left[\begin{smallmatrix} A_{1} & & 0 \\ & \ddots & \\ 0 & & A_{m} \end{smallmatrix}\right], \quad B =\left[\begin{smallmatrix} B_{1} & & 0 \\ & \ddots & \\ 0 & & B_{m} \end{smallmatrix}\right]

where AkA_{k} and BkB_{k} are square matrices of the same size for each k=1,…,mk = 1, \ldots , m. Show that ABA B is a block diagonal matrix of the form

AB=[A1B10⋱0AmBm] A B =\left[\begin{smallmatrix} A_{1} B_{1} & & 0 \\ & \ddots & \\ 0 & & A_{m} B_{m} \end{smallmatrix}\right]
?
Problem 8B.23

Suppose F=R,T∈L(V)\mathbf{F} = \mathbb {R}, T \in \mathcal{L}(V), and λ∈C\lambda \in \mathbb {C}.

?
(a)

Show that u+iv∈G(λ,TC)u+i v \in G\left(\lambda , T_{\mathrm{C}}\right) if and only if u−iv∈G(λˉ,TC)u-i v \in G\left(\bar{\lambda }, T_{\mathrm{C}}\right).

(b)

Show that the multiplicity of λ\lambda as an eigenvalue of TCT_{\mathrm{C}} equals the multiplicity of λˉ\bar{\lambda } as an eigenvalue of TCT_{\mathrm{C}}.

(c)

Use (b) and the result about the sum of the multiplicities (8.25) to show that if dim⁡V\operatorname {dim} V is an odd number, then TCT_{\mathrm{C}} has a real eigenvalue.

Theorem 8.25 (sum of the multiplicities equals dim⁡V\operatorname {dim} V): suppose F=C\mathbf{F} = \mathbb {C} and T∈L(V)T \in \mathcal{L}(V). Then the sum of the multiplicities of all eigenvalues of TT equals dim⁡V\operatorname {dim} V.

(d)

Use (c) and the result about real eigenvalues of TCT_{\mathrm{C}} (Exercise 17 in Section 5A) to show that if dim⁡V\operatorname {dim} V is an odd number, then TT has an eigenvalue (thus giving an alternative proof of 5.34).

Exercise 17 in Section 5A (Exercise 5A.17): if F=R\mathbf{F} = \mathbb {R}, T∈L(V)T \in \mathcal{L}(V), and λ∈R\lambda \in \mathbb {R}, then λ\lambda is an eigenvalue of TT if and only if λ\lambda is an eigenvalue of the complexification TCT_{\mathrm{C}}.

Remark.
?

The complexification TCT_{\mathrm{C}} was defined in Exercise 33 in Section 3B: suppose VV is a real vector space and T∈L(V)T \in \mathcal{L}(V); the complexification TC:VC→VCT_{\mathrm{C}}: V_{\mathrm{C}} \rightarrow V_{\mathrm{C}} is defined by TC(u+iv)=Tu+iTvT_{\mathrm{C}}(u+iv) = Tu+iTv for all u,v∈Vu, v \in V, where VCV_{\mathrm{C}} is the complexification of VV from Exercise 8 in Section 1B.