SIMILARITY: JORDAN FORM.
Lecturer: Askarbekkyzy Aknur
Jordan Block
Definition. A Jordan block with eigenvalue 𝜆 is a square upper
triangular matrix whose entries are all 𝜆 on the main diagonal, all
1 on the entries immediately above the main diagonal and 0
elsewhere:
𝜆 1 0 ⋯ 0 0
0 𝜆 1 ⋯ 0 0
𝐽= 0 0 𝜆 ⋯ 0 0
⋮ ⋮ ⋮ ⋱ ⋮ ⋮
0 0 0 ⋯ 𝜆 1
0 0 0 ⋯ 0 𝜆
Examples of Jordan Blocks
𝜆 1 0 0
𝜆 1 0
𝜆 1 0 𝜆 1 0
𝜆 0 𝜆 1
0 𝜆 0 0 𝜆 1
0 0 𝜆 0 0 0 𝜆
Jordan Canonical form
Definition. A Jordan (canonical) form matrix is a block
diagonal matrix whose blocks are all Jordan blocks.
Theorem. Let 𝐴 be a 𝑛 × 𝑛 matrix. Then there is a Jordan
form matrix 𝐽 that is similar to 𝐴.
Jordan Canonical Form
Theorem. Let 𝐴 be a matrix with characteristic and minimal
polynomials, respectively,
∆𝐴 (𝑡) = 𝑡 − 𝜆1 𝑛1 · · · 𝑡 − 𝜆𝑟 𝑛𝑟
and
𝑚𝐴 𝑡 = 𝑡 − 𝜆1 𝑚1 · · · 𝑡 − 𝜆𝑟 𝑚𝑟 ,
where the 𝜆𝑖 are distinct scalars. Then 𝐴 has a Jordan matrix
𝐽 in which each diagonal block is a Jordan block 𝐽𝑖𝑗 =
𝐽(𝜆𝑖 ).
For each 𝜆𝑖𝑗 , the corresponding 𝐽𝑖𝑗 have the following
properties:
1. There is at least one 𝐽𝑖𝑗 of size 𝑚𝑖 ; all other 𝐽𝑖𝑗 are
of size ≤ 𝑚𝑖 .
2. The sum of the sizes of the 𝐽𝑖𝑗 is 𝑛𝑖 .
3. The number of 𝐽𝑖𝑗 equals the geometric
multiplicity of 𝜆𝑖 .
4. The number of 𝐽𝑖𝑗 of each possible size is uniquely
determined by 𝐴.
Jordan Canonical Form
Thus, in order to construct Jordan form matrix J of a
square matrix A, it is sufficient to have following
information about A:
1. characteristic polynomial of A and its roots;
2. algebraic multiplicities of eigenvalues;
3. minimal polynomial of A;
4. geometric multiplicities of eigenvalues.
Jordan Canonical Form
As a result we have a matrix J which is very close to
diagonal matrix and similar to A.
We note that in case the size of A is 2×2 or 3×3 if we
know minimal polynomial of A we do not need
geometric multiplicities of eigenvalues of A and
conversely.
Example 1. Suppose A has
◦ ∆(t) = (t−3)4(t−1)2 and
◦m(t) = (t−3)2(t−1).
Now we will find all possible Jordan form matrices of A (since
A is not given explicitly) step by step.
Step 1. The size of A is 6×6, because the degree of its
characteristic polynomial is 6. Consequently, J is the matrix of
size 6 × 6.
Step 2. The 3 and 1 are eigenvalues of A with the algebraic
multiplicities 4 and 2 respectively. By Property 2 of the
Theorem we have
where the symbols ∗ are either 1 or 0.
Step 3. By Property 1 of the Theorem we obtain that
there is
1. at least one J1j (3) of size 2 (from the minimal
polynomial (t − 3)2);
2. all other J1j (3) are of size less than or equal to 2.
Similarly, there is
1. at least one J2j(1) of size 1 ( from the minimal
polynomial (t − 1)1 );
2. all other J2j (1) are of size less than or equal to 1.
Thus,
where the symbols ∗ stands for 1 or 0.
Step 4. By Property 3 the geometric multiplicity of the
eigenvalue 3 is either 2 or 3. Depending on them there are two
possible Jordan form matrices:
We have found two possible Jordan form matrices of A.
Step 5. To determine which is the correct Jordan form
matrix of A, according to Property 4 we need to know the
dimension of 𝜀𝐴 3 .
In the end, we have our Jordan form matrix J which is
similar to A.
Example 2. Let Let us find its Jordan
form matrix.
∆A(λ)=|λI3 −A|=(λ+1)2(λ−3).
Then −1 and 3 are eigenvalues of A. According to the
algebraic multiplicities of the eigenvalues we have Jordan
form matrix J of the form:
To obtain the explicit Jordan form matrix of A, we calculate
the geometric multiplicity of the eigenvalue −1. Then we
consider
(A − λI3)X = O
where is an eigenvector belonging to eigenvalue λ = −1.
We have
We solve the system of linear equations and have
So, there is only one Jordan block with −1. Therefore, a
Jordan form matrix of A is
A Jordan base
Whenever we find out A is similar to some Jordan form matrix,
there should exist P such that
J = P−1AP.
◦In case A is a diagonalizable matrix, P is a matrix formed just from
eigenvectors belonging to eigenvalues of A.
◦If A is not diagonalizable, then we will not have enough
eigenvectors. So P can not be formed only from eigenvectors. We
need some other vectors, so-called a Jordan basis determine the
matrix P.
A Jordan base
A Jordan base
A Jordan base
A Jordan base
A Jordan base
A Jordan base
Calculation of f(J)
Below we consider question how to calculate 𝑓 𝐽 for
Jordan block J if 𝑓 𝑥 is a differentiable function with
respect to a variable 𝑥.
𝑘
We write 𝑓 𝑥 for 𝑘-th derivative of 𝑓 𝑥 with
respect to 𝑥.
When 𝑘 = 1 or 2, simply write 𝑓 ′ 𝑥 and 𝑓 ′′ 𝑥 .
Example
Calculating f(A)
