Professional Documents
Culture Documents
The JordanChevalley decomposition is particularly simple with respect to a basis for which the operator
takes its Jordan normal form. The diagonal form for
diagonalizable matrices, for instance normal matrices, is
a special case of the Jordan normal form.[5][6][7]
The Jordan normal form is named after Camille Jordan.
1 Overview
1.1 Notation
Some textbooks have the ones on the subdiagonal, i.e.,
immediately below the main diagonal instead of on the
superdiagonal. The eigenvalues are still on the main
diagonal.[8][9]
1.2 Motivation
An n n matrix A is diagonalizable if and only if the
sum of the dimensions of the eigenspaces is n. Or,
equivalently, if and only if A has n linearly independent
eigenvectors. Not all matrices are diagonalizable. Consider the following matrix:
5
4
2
1
0
1 1 1
A=
.
1 1
3
0
1
1 1
2
Including multiplicity, the eigenvalues of A are = 1, 2,
4, 4. The dimension of the kernel of (A 4I ) is 1 (and
not 2), so A is not diagonalizable. However, there is an
invertible matrix P such that A = PJP 1 , where
1
0
J =
0
0
0
2
0
0
0
0
4
0
0
0
.
1
4
2 COMPLEX MATRICES
Complex matrices
In general, a square complex matrix A is similar to a block Main article: Generalized eigenvectors
diagonal matrix
Consider the matrix A from the example in the previous
section. The Jordan normal form is obtained by some
J1
similarity transformation P 1 AP = J, i.e.
.
..
J =
Jp
AP = P J.
where each block J is a square matrix of the form
Ji =
1
i
..
..
1
i
[
A p1
p2
p3
] [
p4 = p1
p2
p3
1
]0
p4
0
0
0
2
0
0
0
0
4
0
I)
pr is an eigenj
j-1
1
size at least j is dim Ker(A - I) - dim Ker(A - I) .
vector
corresponding
to
.
In
general,
pi
is
a preimage of
Thus, the number of Jordan blocks of size exactly j
pi
under
A
I.
So
the
lead
vector
generates
the chain
is
via multiplication by (A I).
the statement that every square matrix A can
2 dim ker(Ai I)j dim ker(Ai I)j+1 dim ker(AiTherefore,
I)j1
be put in Jordan normal form is equivalent to the claim
Given an eigenvalue i, its multiplicity in the mini- that there exists a basis consisting only of eigenvectors
mal polynomial is the size of its largest Jordan block. and generalized eigenvectors of A.
0
[
0
= p1
1
4
2p2
2.3
Uniqueness
2.2
A proof
2.3 Uniqueness
It can be shown that the Jordan normal form of a given
matrix A is unique up to the order of the Jordan blocks.
(A I)k1 = 0
(A I)k1 1
is the number of Jordan blocks of size k1 . Similarly, the
rank of
(A I)k1 2
is twice the number of Jordan blocks of size k1 plus the
number of Jordan bl Jordan structure of A. The general
case is similar.
This can be used to show the uniqueness of the Jordan
form. Let J 1 and J 2 be two Jordan normal forms of A.
Then J 1 and J 2 are similar and have the same spectrum,
including algebraic multiplicities of the eigenvalues. The
procedure outlined in the previous paragraph can be used
to determine the structure of these matrices. Since the
rank of a matrix is preserved by similarity transformation,
there is a bijection between the Jordan blocks of J 1 and
J 2 . This proves the uniqueness part of the statement.
3 Real matrices
CONSEQUENCES
and describe multiplication by i in the complex plane. ideal domain of polynomials with complex coecients.
The superdiagonal blocks are 22 identity matrices. The The monic element that generates I is precisely P.
full real Jordan block is given by
Let 1 , ..., q be the distinct eigenvalues of A, and si be
the size of the largest Jordan block corresponding to i.
It is clear from the Jordan normal form that the minimal
Ci I
polynomial of A has degree si.
..
.
Ci
While the Jordan normal form determines the minimal
.
Ji =
.
polynomial, the converse is not true. This leads to the
.
. I
notion of elementary divisors. The elementary divisors
Ci
of a square matrix A are the characteristic polynomials
This real Jordan form is a consequence of the complex of its Jordan blocks. The factors of the minimal polynoJordan form. For a real matrix the nonreal eigenvec- mial m are the elementary divisors of the largest degree
tors and generalized eigenvectors can always be chosen corresponding to distinct eigenvalues.
to form complex conjugate pairs. Taking the real and The degree of an elementary divisor is the size of the corimaginary part (linear combination of the vector and its responding Jordan block, therefore the dimension of the
conjugate), the matrix has this form with respect to the corresponding invariant subspace. If all elementary divinew basis.
sors are linear, A is diagonalizable.
Consequences
4.1
Xi
Using the Jordan normal form, direct calculation gives a
i=1
spectral mapping theorem for the polynomial functional
calculus: Let A be an n n matrix with eigenvalues 1 , where each Xi is the span of the corresponding Jordan
..., n, then for any polynomial p, p(A) has eigenvalues chain, and k is the number of Jordan chains.
p(1 ), ..., p(n).
One can also obtain a slightly dierent decomposition via
the Jordan form. Given an eigenvalue i, the size of its
largest corresponding Jordan block s is called the index
4.2 CayleyHamilton theorem
of i and denoted by (i). (Therefore the degree of the
minimal polynomial is the sum of all indices.) Dene a
The CayleyHamilton theorem asserts that every ma- subspace Yi by
trix A satises its characteristic equation: if p is the
characteristic polynomial of A, then p(A) = 0. This can
be shown via direct calculation in the Jordan form, since Y = Ker( I A)(i ) .
i
i
any Jordan block for is annihilated by (X )m where m
is the multiplicity of the root of p, the sum of the sizes This gives the decomposition
of the Jordan blocks for , and therefore no less than the
size of the block in question. The Jordan form can be asl
sumed to exist over a eld extending the base eld of the
5.2
Compact operators
Comparing the two decompositions, notice that, in general, l k. When A is normal, the subspaces Xi's in the
rst decomposition are one-dimensional and mutually or- 5.2.1 Holomorphic functional calculus
thogonal. This is the spectral theorem for normal operators. The second decomposition generalizes more easily For more details on this topic, see holomorphic funcfor general compact operators on Banach spaces.
tional calculus.
It might be of interest here to note some properties of the
index, (). More generally, for a complex number , its Let X be a Banach space, L(X) be the bounded operators
index can be dened as the least non-negative integer () on X, and (T) denote the spectrum of T L(X). The
such that
holomorphic functional calculus is dened as follows:
1
f (T ) =
f (z)(z T )1 dz.
2i
Generalizations
The open set G could vary with f and need not be connected. The integral is dened as the limit of the Riemann
sums, as in the scalar case. Although the integral makes
5.1 Matrices with entries in a eld
sense for continuous f, we restrict to holomorphic functions to apply the machinery from classical function theJordan reduction can be extended to any square maory (e.g. the Cauchy integral formula). The assumption
trix M whose entries lie in a eld K. The result states
that (T) lie in the inside of ensures f(T) is well dened;
that any M can be written as a sum D + N where D
it does not depend on the choice of . The functional calis semisimple, N is nilpotent, and DN = ND. This is
culus is the mapping from Hol(T) to L(X) given by
called the JordanChevalley decomposition. Whenever
K contains the eigenvalues of M, in particular when K is
algebraically closed, the normal form can be expressed
(f ) = f (T ).
explicitly as the direct sum of Jordan blocks.
Similar to the case when K is the complex numbers, We will require the following properties of this functional
knowing the dimensions of the kernels of (M I)k for calculus:
1 k m, where m is the algebraic multiplicity of the
eigenvalue , allows one to determine the Jordan form of
1. extends the polynomial functional calculus.
M. We may view the underlying vector space V as a K[x]module by regarding the action of x on V as application
2. The spectral mapping theorem holds: (f(T)) =
of M and extending by K-linearity. Then the polynomials
f((T)).
(x )k are the elementary divisors of M, and the Jordan
3. is an algebra homomorphism.
normal form is concerned with representing M in terms
of blocks associated to the elementary divisors.
The proof of the Jordan normal form is usually carried 5.2.2 The nite-dimensional case
out as an application to the ring K[x] of the structure theorem for nitely generated modules over a principal ideal In the nite-dimensional case, (T) = {i} is a nite disdomain, of which it is a corollary.
crete set in the complex plane. Let ei be the function that
EXAMPLE
RT () = ( T )1
has a pole of order at .
f (T )ei (T ) = (T i )i ei (T )
We will show that, in the nite-dimensional case, the orhas spectrum {0}. By property 1, f(T) can be directly der of an eigenvalue coincides with its index. The result
computed in the Jordan form, and by inspection, we see also holds for compact operators.
that the operator f(T)ei(T) is the zero matrix.
Consider the annular region A centered at the eigenvalue
with suciently small radius such that the intersecBy property 3, f(T) ei(T) = ei(T) f(T). So ei(T) is pretion
of the open disc B() and (T) is {}. The resolcisely the projection onto the subspace
vent function RT is holomorphic on A. Extending a result
from classical function theory, RT has a Laurent series
representation on A:
Ran e (T ) = Ker(T )i .
i
The relation
RT (z) =
am ( z)m
ei = 1
where
implies
Cn =
Ran ei (T ) =
Ker(T i )i
Cn =
Yi
1
am = 2i
( z)m1 (z T )1 dz and
C
C is a small circle centered at .
By the previous discussion on the functional calculus,
am = ( T )m1 e (T ) where e is 1
on B () and 0 elsewhere.
But we have shown that the smallest positive integer m
such that
i 1
f (k)
(T i )k ei (T ).
k!
6 Example
i (T ) k=0
5
4
2
0
1
1
A=
1 1 3
1
1 1
0
P 1 AP = J =
0
0
1
1
0
2
0
2
0
0
0
0
4
0
0
0
.
1
4
If we had interchanged the order of which the chain vectors appeared, that is, changing the order of v, w and {x,
y} together, the Jordan blocks would be interchanged.
However, the Jordan forms are equivalent Jordan forms.
1
0
J = J1 (1) J1 (2) J2 (4) =
0
0
0
2
0
0
0
0
.
1
4
0
0
4
0
7 Numerical analysis
If the matrix A has multiple eigenvalues, or is close to
a matrix with multiple eigenvalues, then its Jordan normal form is very sensitive to perturbations. Consider for
instance the matrix
[
A=
]
1
.
1
]
1 1
.
0 1
1+
0
.
0
1
There are three chains. Two have length one: {v} and
{w}, corresponding to the eigenvalues 1 and 2, respec- This ill conditioning makes it very hard to develop a rotively. There is one chain of length two corresponding to bust numerical algorithm for the Jordan normal form, as
the result depends critically on whether two eigenvalues
the eigenvalue 4. To nd this chain, calculate
are deemed to be equal. For this reason, the Jordan normal form is usually avoided in numerical analysis; the sta
ble
Schur decomposition[11] or pseudospectra[12] are bet1
1
ter alternatives.
0 0
2
ker (A 4I) = span
0, 1 .
0
1
8 Powers
1 1
1 1
[ ] 1 1 0 0
.
P = vwxy =
0
0 1 0
0
1
1 0
For example,
2
0
1
2
0
0
0
0
1
2
0
0
0
0
0
5
0
4
0
16 32 24
0
0 16 32
0
0
0
0
= 0 0 16
0 0
1
0 625
5
0 0
0
0
0
0
0
.
500
625
12
11 References
Beauregard, Raymond A.; Fraleigh, John B. (1973),
A First Course In Linear Algebra: with Optional
Introduction to Groups, Rings, and Fields, Boston:
Houghton Miin Co., ISBN 0-395-14017-X
For example,
1
0
0
0
1
1
0
0
0
0
1
1
0
0
0
0
0
2
0
n n
1
0
0
0
0
=0
0
1
2
0
(n) n1
1 1
n1
0
0
0
(n) n2
1
(n2 )n1
1 1
n1
0
0
See also
Canonical form
Frobenius normal form
Jordan matrix
JordanChevalley decomposition
Matrix decomposition
Weyr canonical form
10
Notes
EXTERNAL LINKS
LCCN
0
066021267
0
0
N. Dunford
J.T. Schwartz, Linear Operators,
and
.
0 ( )0
Part
I:
General
Theory,
Interscience, 1958.
n n1
n2
1 2
n
0 DanielT.
2 Finkbeiner II, Introduction to Matrices and
Linear Transformations, Third Edition, Freeman,
1978.
Franklin, Joel N. (1968), Matrix Theory, Englewood
Clis: Prentice-Hall, LCCN 68016345
Gene H. Golub and Charles F. Van Loan, Matrix
Computations (4th ed.), Johns Hopkins University
Press, Baltimore, 2012.
Gene H. Golub and J. H. Wilkinson, Ill-conditioned
eigensystems and the computation of the Jordan normal form, SIAM Review, vol. 18, nr. 4, pp. 578
619, 1976.
Horn, Roger A.; Johnson, Charles R. (1985), Matrix
Analysis, Cambridge University Press, ISBN 978-0521-38632-6.
Glenn James and Robert C. James, Mathematics
Dictionary, Fourth Edition, Van Nostrand Reinhold,
1976.
Saunders MacLane and Garrett Birkho, Algebra,
MacMillan, 1967.
Anthony N. Michel and Charles J. Herget, Applied
Algebra and Functional Analysis, Dover, 1993.
Nering, Evar D. (1970), Linear Algebra and Matrix Theory (2nd ed.), New York: Wiley, LCCN
76091646
Georgi E. Shilov, Linear Algebra, Dover, 1977.
I. R. Shafarevich & A. O. Remizov (2012) Linear
Algebra and Geometry, Springer ISBN 978-3-64230993-9.
Jordan Canonical
world.wolfram.com
Form
article
at
math-
12 External links
On line tool on Jordan form and matrix diagonalizations by www.mathstools.com
13
13.1
13.2
Images
File:Jordan_blocks.svg Source: http://upload.wikimedia.org/wikipedia/commons/4/4f/Jordan_blocks.svg License: CC BY-SA 3.0 Contributors: Own work Original artist: Jakob.scholbach
13.3
Content license