Jump to content

Endomorphism/Jordan normal form/Introduction/Section

From Wikiversity


Let K be a field and λ∈K. A Jordan matrix (to the eigenvalue λ) is a square matrix of the form[1]

(λ10⋯⋯00λ10⋯0⋮⋱⋱⋱⋱⋮0⋯0λ100⋯⋯0λ10⋯⋯⋯0λ).

If we consider such a Jordan matrix as a linear mapping φ on the standard space Kn, then

φ(e1)=λe1 and φ(ek)=λek+ek−1 for all k≥2.

In particular, e1 is a eigenvector to the eigenvalue λ. A simple observation shows that there is no further eigenvector linearly independent to e1 (see exercise). The property on the right is equivalent with the condition[2]

ek−1=(φ−λ⋅Id⁡)(ek)

for k≥2. The eigenvector e1 is a generating element of the kernel of the mapping ψ:=φ−λId, and the other standard vectors ek arise successively as preimages of ek−1 under ψ.


A square matrix of the form

(J10⋯⋯00J20⋯0⋮⋱⋱⋱⋮0⋯0Jk−100⋯⋯0Jk),

where each Ji is a Jordan matrix, is called a matrix in

Jordan normal form.

The Jordan matrices occurring here are called the Jordan blocks of the matrix. Their eigenvalues might be different or equal. In the matrix

(210000020000004100000410000040000002),

there are three Jordan blocks,

(2102),(410041004) and (2)

with eigenvalues 2,4 and again 2.

We state and prove now the theorem about the Jordan normal form for trigonalizable endomorphisms.


Theorem

For every trigonalizable endomorphism

φ:V⟶V

on a finite-dimensional K-vector space V, there exists a basis, such that the describing matrix is in

Jordan normal form.

Proof  

Since φ is trigonalizable, we can apply fact. Hence, there exists a direct sum decomposition

V=GeEigλ1(φ)⊕⋯⊕GeEigλm(φ),

where the generalized eigenspaces are φ-invariant. Looking at the situation for each generalized eigenspace, we may assume that φ has only one eigenvalue λ, and that

V=GeEigλ(φ)

holds. Then,

ψ=φ−λIdV

is nilpotent. Therefore, because of fact, there exists a basis such that ψ is described by a matrix of the form

(0c10⋯⋯000c20⋯0⋮⋱⋱⋱⋱⋮0⋯00cn−200⋯⋯00cn−10⋯⋯⋯00),

where the ci equal 0 or equal 1. With respect to this basis,

φ=ψ+λIdV

has the form

(λc10⋯⋯00λc20⋯0⋮⋱⋱⋱⋱⋮0⋯0λcn−200⋯⋯0λcn−10⋯⋯⋯0λ).
◻


Hence, every upper triangular matrix is similar to a matrix in Jordan normal form. Over the complex numbers, every matrix can be brought to Jordan normal form. If a matrix has Jordan normal form, then we can read of directly the diagonalizable and the nilpotent part in the sense of fact: The diagonal yields the diagonalizable part, and the entries which are strictly above the diagonal, yield the nilpotent part (in general, this is not true for upper triangular matrices).


We describe how to find to a linear trigonalizable mapping φ:V→V a basis, such that describing matrix with respect to this basis is in Jordan normal form. For this, we determine, for every eigenvalue λ∈K, the minimal exponent s with

kern⁡(φ−λId)s=kern⁡(φ−λId)s+1.

This kernel is the generalized eigenspace to λ. We set

Vi=kern⁡(φ−λId)i⊆GeEigλ(φ)

for i=1,…,s. This yields the chain

V1=Eig(λ)⊆V2⊂⋯⊂Vs−1⊂Vs=GeEigλ(φ).

Now, we choose a vector u from Vs∖Vs−1. The vectors

u,(φ−λId⁡)(u),(φ−λId⁡)2(u),…,(φ−λId⁡)s−1(u)

form a basis for a Jordan block. If this basis generates the generalized eigenspace, then we are done. Otherwise, we look in Vs∖Vs−1 for another vector which is linearly independent to u and to Vs−1. Again, we add this vector and all its successive images. If Vs∖Vs−1 is exhausted, then we look whether Vs−1∖Vs−2 is already covered, and so on. If the generalized eigenspace to λ is covered, then we continue with the next eigenvalue.

Under certain circumstances, we can also start with a basis of the eigenspace. If, for example, the eigenspace to λ is one-dimensional, then we can choose an eigenvector v for λ, and we can find successively preimages under φ−λIdV of the vectors, that is, we have to solve the equation

v=(φ−λIdV)(v′),

then

v′=(φ−λIdV)(v″),

etc.

If, for example, the eigenspace is k-dimensional and the generalized eigenspace is (k+1)-dimensional, then we only have to find a preimage for one eigenvector under φ−λIdV.


We consider the matrix

M=(221023002),

and want to bring it to Jordan normal form. bringen. The vector u=e1=(100) is an eigenvector to the eigenvalue 2. We have

A:=M−2E3=(021003000),

so that there exists no further linearly independent eigenvector. We look at the linear system e1=Av. This imples (looking at the second row) v3=0 and so 2v2=1 (we can choose v1 freely to be 0). Hence, we set v=(0120). Finally, we need a solution for v=Aw. This yields the equation w=(0−11216). The matrix M acts as

Mu=2u,Mv=2v+u,Mw=2w+v,

so that the mapping is described with respect to the basis u,v,w by

(210021002).

This matrix is a Jordan matrix and, in particular, in Jordan normal form.


We consider the matrix

M=(200023002)

and want to bring it to Jordan normal form. The vectors u=e1=(100) and v=e2=(010) are linearly independent eigenvectors to the eigenvalue 2. We have

A:=M−2E3=(000003000),

so that u and v span this eigenspace. An eigenvector must be the image of some vector under the matrix A. In fact, the linear system

e2=Aw

has the solution w=(0013). Therefore, the matrix M acts in the following way

Mu=2u,Mv=2v,Mw=2w+v.

Hence, the mapping is described, with respect to the basis u,v,w, by the matrix

(200021002).

This matrix is in Jordan normal form with the Jordan blocks (2) and (2102).


We consider the matrix

M=(31040−12100−100003)

and want to bring in in Jordan normal for. Here, we have two eigenvalues and, therefore, two two-dimensional generalized eigenspaces, which we treat separately. We have

M−3E4=(01040−42100−400000),

therefore, (1000) belongs to the kernel. The determinant of the upper right submatrix is not 0, so the rank of the matrix is 3, and its kernel is one-dimensional. The second power is

(01040−42100−400000)2=(01040−42100−400000)(01040−42100−400000)=(0−421016−16−4001600000),

a new element of the kernel is (0104). Thus, we have

GeEig3(M)=⟨(1000),(0104)⟩.

Because of

(01040−42100−400000)(0104)=(17000),

we can use the vectors (17000) and (0104) to establish the first Jordan block.

We have

M+1E4=(4104002100000004),

therefore, (1−400) belongs to the kernel. The rank of this matrix is again 3, and the kernel has dimension one. The second power is

(4104002100000004)2=(4104002100000004)(4104002100000004)=(1642330004000000016),

a new element in the kernel is (01−20). Thus, we have

GeEig−1(M)=⟨(1−400),(01−20)⟩.

Because of

(4104002100000004)(01−20)=(1−400),

we can use the vectors (1−400) and (01−20) to establish the second Jordan block. Altogether, the linear mapping defined by M has, with respect the basis

(17000),(0104),(1−400),(01−20),

the Jordan normal form

(3100030000−11000−1).
  1. ↑ Some authors define a Jordan matrix one where the 1 are below the diagonal.
  2. ↑ In the context of trigonalizable mappings and in order to find the Jordan normal form, it is useful to work with φ−λ⋅Id instead of λ⋅Id⁡−φ.