Академический Документы
Профессиональный Документы
Культура Документы
TECHNICAL SCIENCES
Vol. 56, No. 4, 2008
1
The School of Banking and Management, 4 Armii Krajowej St., 30-150 Krakw, Poland
Faculty of Management, AGH University of Science and Technology, 30 Mickiewicza St., 30-059 Krakw, Poland
Abstract. This paper gives the simple algorithm for calculation of the degree and coefficients of the minimal polynomial for the complex
matrix A = [aij ]nn .
Key words: matrix, minimal polynomial, characteristic polynomial.
1. Introduction
Ak =
(k)
we have:
(k)
(k)
(k)
a22
(k = 0, 1, 2, . . .),
a12
a22
1 a11
0 a
12
B1 = [a(0) a(1) ] =
.
0 a21
1
a11
a21
f () = an n + an1 n1 + + a1 + a0 ,
(k)
[aij ]
"
sbialas@uci.agh.edu.pl
391
1) rank Bk = k + 1 (k = 0, 1, . . . , k0 1),
2) rank Bk = k0 (k k0 ),
3) deg () = k0 ,
0 + 1 + + k0 1 k0 1 + k0 ,
It denotes that
0 I + 1 A + + k0 1 Ak0 1 + Ak0 = 0 Mn
k0
392
Bk0 1 = a(k0 ) ,
(3)
Bk1 = a(k) ,
with the unknown = [0 1 . . . k1 ]T C k , has not the
solutions.
This denotes that the polynomial (3) is the minimal polynomial of the matrix A and deg () = k0 .
Now, we give the algorithm for the calculation of the degree and coefficients of the minimal polynomial of the matrix
A = [aij ] Mn .
Consider the matrix
Bn = [a(0) a(1) . . . a(n) ] Mn2 ,n+1 ,
which is defined in (1).
The elements of the matrix Bn are denoted by bij , therefore Bn = [bij ] Mn2 ,n+1 , where b11 = 1, b12 =
(1)
(n)
(n)
a11 , . . . , b1,n+1 = a11 , . . . , bn2 ,n+1 = ann .
We will calculate the rank of the matrix Bn by Gaussian
elimination, except interchange and cancel of the null
columns.
We obtain
1
b12 . . . b1,n+1
(1)
(1)
0
b22
. . . b2,n+1
(1)
(1)
rankBn = rank
0
b
.
.
.
b
32
3,n+1 ,
(1)
(1)
bn2 ,2
(1)
. . . bn2 ,n+1
(1)
(1)
...
rankBn = rank 0
...
that r n and
b12
b13
... ...
...
...
...
...
b1,n+1
b12
(1)
b23
(1)
... ...
...
...
...
...
b2,n+1
b33
(2)
... ...
...
...
...
...
b3,n+1
...
...
... ...
...
...
...
...
(r1)
br,n+1
...
...
br+1,r+2
...
br+2,r+2
...
...
...
...
...
...
...
bn2 ,r+2
(r1)
...
bn2 ,n+1
... ...
...
Bk0 1 = a(k0 ) ,
(4)
T
k0
C , has only
0 + 1 A + + k0 1 Ak0 1 + Ak0 = 0 Mn .
Therefore 0 , 1 , . . . , k0 1 , 1 are the coefficients of the
minimal polynomial of the matrix A. The set of Eq. (4) is
equivalent to the set of equations
= b,
B
where
b11
0 b(1)
22
B= 0
0
... ...
0
0
...
...
b1r
...
...
b2r
b33
(2)
...
b3r
...
...
...
(r1)
brr
...
(1)
(2)
= [0 1 . . . k0 1 ]T , r = k0 .
Example 2. We will calculate
matrix
A = 1
1
b1,r+1
(1)
b2,r+1
, b =
..
(r1)
br,r+1
3 2
5 2 .
3
0
10 18 12
2
A = 6 22 12 ,
6 18
8
...
(2)
(i1)
...
(r1)
br,r+1 . . .
(1)
(r1)
brr
where bii
6= 0 (i = 1, 2, . . . , r).
From this it follows that rankBj = j (j = 1, 2, . . . , n),
rankBr1 = r, rankBr = r.
Therefore k0 = min K = r and deg () = r = k0 .
Thus, by Gaussian elimination we can compute the degree
of the minimal polynomial of the matrix A.
Hence that det Br1 = det Bk0 1 6= 0 and rankBk0 =
rankBk0 1 = k0 it follows that the set of equations
...
(r1)
. . . br+1,n+1
(r1)
(r1)
. . . br+2,n+1
(r1)
(r1)
36 84 56
A3 = 28 92 56 ,
28 84 48
1 3
10
36
0 3 18 84
0 2
12
56
0 1 6 28
rankB3 = rank 1 5
22
92
0 2 12 56
0 1 6 28
0 3
18
84
1 0
8 48
1 3
10
36
0 3 18 84
0 0
0
0
0 0
0
0
= rank 0 0
0
0 = 2,
0 0
0
0
0 0
0
0
0 0
0
0
0
=
B
"
1
0
0
0
0
# "
#
3
10
, b
,
3
18
k0 = 2,
= [0 1 ]T = [8 6]T .
Therefore, () = 2 6+8. is the minimal polynomial
of the matrix A.
REFERENCES
[1] S. Barnett, Matrices in Control Theory, Van Nostrand Reinhold
Company, London, 1960.
[2] T. Kaczorek, Vectors and Matrices in Automatics and Electrotechnics, WNT, Warszawa, 1998.
393