Jordan matrix

Jordan matrix

In the mathematical discipline of matrix theory, a Jordan block over a ring R (whose identities are the zero 0 and one 1) is a matrix which is composed of 0 elements everywhere except for the diagonal, which is filled with a fixed element lambdain R, and for the superdiagonal, which is composed of unities of the ring.

:egin{pmatrix}lambda & 1 & 0 & cdots & 0 \0 & lambda & 1 & cdots & 0 \vdots & vdots & ddots& vdots & vdots \0 & 0 & 0 & lambda & 1 \0 & 0 & 0 & 0 & lambda \end{pmatrix}

Any Jordan block is thus specified by its dimension "n" and its eigenvalue lambda and is indicated as J_{lambda,n}.Any block diagonal matrices whose blocks are Jordan blocks is called a Jordan matrix; using either the oplus or the “mbox{diag}” symbol, the (l+m+n) imes (l+m+n) block diagonal square matrix whose first diagonal block is J_{alpha,l}, whose second diagonal block is J_{eta,m} and whose third diagonal block is J_{gamma,n} is compactly indicated as J_{alpha,l}oplus J_{eta,m}oplus J_{gamma,n} or mbox{diag}left(J_{alpha,l}, J_{eta,m}, J_{gamma,n} ight), respectively.For example the matrix:J=left(egin{matrix}0 & 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \0 & 0 & 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \0 & 0 & 0 & i & 1 & 0 & 0 & 0 & 0 & 0 \0 & 0 & 0 & 0 & i & 0 & 0 & 0 & 0 & 0 \0 & 0 & 0 & 0 & 0 & i & 1 & 0 & 0 & 0 \0 & 0 & 0 & 0 & 0 & 0 & i & 0 & 0 & 0 \0 & 0 & 0 & 0 & 0 & 0 & 0 & 7 & 1 & 0 \0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 7 & 1 \0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 7 end{matrix} ight)is a 11 imes 11 Jordan matrix with a 3 imes 3 block with eigenvalue 0, two 2 imes 2 blocks with eigenvalue the imaginary unit and a 3 imes 3 block with eigenvalue 7. Its Jordan-block structure can also be written as either J_{0,3}oplus J_{i,2}oplus J_{i,2}oplus J_{7,3} or mbox{diag}left(J_{0,3},J_{i,2},J_{i,2},J_{7,3} ight).

Linear Algebra

Any n imes n square matrix A whose elements are in an algebraically closed field K is similar to a Jordan matrix J, also in mathbb{M}_n (K), which is unique up to a permutation of its diagonal blocks themselves. J is called the Jordan normal form of A and corresponds to a generalization of the diagonalization procedure. A diagonalizable matrix is similar, in fact, to a special case of Jordan matrix: the matrix whose blocks are all 1 imes 1.

More generally, given a Jordan matrix J=J_{lambda_1,m_1}oplus J_{lambda_2,m_2} oplusldotsoplus J_{lambda_N,m_N}, i.e. whose k^{mbox{th diagonal block, 1leq kleq N is the Jordan block J_{lambda_k,m_k} and whose diagonal elements lambda_k may not all be distinct, it can easily be seen that the geometric multiplicity of lambdain K for the matrix J, indicated as mbox{gmul}_J lambda,, corresponds to the number of Jordan blocks whose eigenvalue is lambda. Whereas the index of an eigenvalue lambda for J, indicated as mbox{idx}_J lambda,, is defined as the dimension of the largest Jordan block associated to that eigenvalue.

The same goes for all the matrices A similar to J, so mbox{idx}_A lambda, can be defined accordingly respect to the Jordan normal form of A for any of its eigenvalues lambda inmbox{spec}A. In this case one can check that the index of lambda for A is equal to its multiplicity as a root of the minimal polynomial of A (whereas, by definition, its algebraic multiplicity for A, mbox{mul}_A lambda,, is its multiplicity as a root of the characteristic polynomial of A, i.e. mbox{det}(A-xI)in K [x] ).An equivalent necessary and sufficient condition for A do be diagonalizable in K is that all of its eigenvalues have index equal to 1, i.e. its minimal polynomial has only simple roots.

Note that knowing a matrix's spectrum with all of its algebraic/geometric multiplicities and indexes does not always allow for the computation of its Jordan normal form (this may be a sufficient condition only for spectrally simple, usually low-dimensional matrices): the Jordan decomposition is, in general, a computationally challenging task.From the vector space point of view, the Jordan decomposition is equivalent to finding an orthogonal decomposition (i.e. via direct sums of eigenspaces represented by Jordan blocks) of the domain which the associated generalized eigenvectors make a basis for.

Functions of matrices

Let Ainmathbb{M}_n (mathbb{C}) (i.e. a n imes n complex matrix) and Cinmathrm{GL}_n (mathbb{C}) be the change of basis matrix to the Jordan normal form of A, i.e. A=C^{-1}JC.Now let f(z) be a holomorphic function on an open set mathit{Omega} such that mbox{spec}Asubset mathit{Omega}subseteqmathbb{C}, i.e. the spectrum of the matrix is contained inside the domain of holomorphy of f. Let

:f(z)=sum_{h=0}^{infty}a_h z^h

be the power series expansion of f around zero, then the matrix f(A), defined via the following formal power series

:f(A)=sum_{h=0}^{infty}a_h A^h

is absolutely convergent respect to the Euclidean norm of mathbb{M}_n (mathbb{C}). To put it in another way, f(A), converges absolutely for every square matrix whose spectral radius is less than the radius of convergence of f around 0 and is uniformly convergent on any compact subsets of mathbb{M}_n (mathbb{C}) satisfying this property in the matrix Lie group topology.

The Jordan normal form allows the computation of functions of matrices without explicitly computing an infinite series, which is one of the main achievements of Jordan matrices. Using the facts that the k^mathrm{th} power (kinmathbb{N}_0) of a diagonal block matrix is the diagonal block matrix whose blocks are the k^mathrm{th} powers of the respective blocks, i.e. left(A_1 oplus A_2 oplus A_3 oplusldots ight)^k=A^k_1 oplus A_2^k oplus A_3^k oplusldots, and that A^k=C^{-1}J^k C,, the above matrix power series becomes

:f(A)=C^{-1}f(J)C=C^{-1}left(igoplus_{k=1}^N fleft(J_{lambda_k ,m_k} ight) ight)C

where the last series must not be computed explicitly via power series of every Jordan block. In fact, if lambdainmathit{Omega}, any holomorphic function of a Jordan block f(J_{lambda,n}), is the following upper triangular matrix:

:f(J_{lambda,n})=left(egin{matrix}f(lambda) & f^prime (lambda) & frac{f^{primeprime}(lambda)}{2} & cdots & frac{f^{(n-2)}(lambda)}{(n-2)!} & frac{f^{(n-1)}(lambda)}{(n-1)!} \0 & f(lambda) & f^prime (lambda) & cdots & frac{f^{(n-3)}(lambda)}{(n-3)!} & frac{f^{(n-2)}(lambda)}{(n-2)!} \0 & 0 & f(lambda) & cdots & frac{f^{(n-4)}(lambda)}{(n-4)!} & frac{f^{(n-3)}(lambda)}{(n-3)!} \vdots & vdots & vdots & ddots & vdots & vdots \0 & 0 & 0 & cdots & f(lambda) & f^prime (lambda) \0 & 0 & 0 & cdots & 0 & f(lambda) \end{matrix} ight)=left(egin{matrix}a_0 & a_1 & a_2 & cdots & a_{n-1} \0 & a_0 & a_1 & cdots & a_{n-2} \0 & 0 & a_0 & cdots & a_{n-3} \vdots & vdots & vdots & ddots & vdots \0 & 0 & 0 & cdots & a_1 \0 & 0 & 0 & cdots & a_0 \end{matrix} ight).

As a consequence of this, the computation of any functions of a matrix is straightforward whenever its Jordan normal form and its change-of-basis matrix are known.Also, mbox{spec}f(A)=f(mbox{spec}A), i.e. every eigenvalue lambdainmbox{spec}A corresponds to the eigenvalue f(lambda)inmbox{spec}f(A) with the same algebraic multiplicity (i.e. mbox{mul}_{f(A)}f(lambda)=mbox{mul}_A lambda,) but it has, in general, different geometric multiplicity and index;

The function f(T) of a linear transformation T between vector spaces can be defined in a similar way according to the holomorphic functional calculus, where Banach space and Riemann surface theories play a fundamental role. Anyway, in the case of finite-dimensional spaces, both theories perfectly match.

Dynamical systems

Now suppose a (complex) dynamical system is simply defined by the equation:dot{mathbf{z(t)=A(mathbf{c})mathbf{z}(t),:mathbf{z}(0)=mathbf{z}_0 inmathbb{C}^n,where mathbf{z}:mathbb{R_+} ightarrow mathcal{R} is the (n-dimensional) curve parametrization of an orbit on the Riemann surface mathcal{R} of the dynamical system, whereas A(mathbf{c}) is an n imes n complex matrix whose elements are complex functions of a d-dimensional parameter mathbf{c}inmathbb{C}^d.Even if Ainmathbb{M}_n left(mathrm{C}^0(mathbb{C}^d) ight) (i.e. A continuously depends on the parameter mathbf{c}) the Jordan normal form of the matrix is continuously deformed almost everywhere on mathbb{C}^d but, in general, not everywhere: there is some critical submanifold of mathbb{C}^d which the Jordan form abruptly changes its structure whenever the parameter crosses or simply “travels” around it (monodromy). Such changes substantially mean that several Jordan blocks (either belonging to different eigenvalues or not) join together to a unique Jordan block, or vice versa (i.e. one Jordan block splits in two or more different ones).Many aspects of Bifurcation theory for both continuous and discrete dynamical systems can be interpreted with the analysis of functional Jordan matrices.From the tangent space dynamics this means that the orthogonal decomposition of the dynamical systems' phase space changes and, for example, different orbits gain periodicity, or lose it, or shift from a certain kind of periodicity to another (such as "period-doubling", cfr. Logistic map).In just one sentence, the qualitative behaviour of such a dynamical system may substantially change as the versal deformation of the Jordan normal form of A(mathbf{c}).

Linear ordinary differential equations

The most simple example of dynamical system is a system of linear, constant-coefficients ordinary differential equations, i.e. let Ainmathbb{M}_n (mathbb{C}) and mathbf{z}_0 inmathbb{C}^n::dot{mathbf{z(t)=Amathbf{z}(t),:mathbf{z}(0)=mathbf{z}_0,whose direct closed-form solution involves computation of the matrix exponential::mathbf{z}(t)=e^{tA}mathbf{z}_0.Another way, provided the solution is restricted to the local Lebesgue space of n-dimensional vector fields mathbf{z}inmathrm{L}_{mathrm{loc^1 (mathbb{R}_+)^n, is to use its Laplace transform mathbf{Z}(s)=mathcal{L} [mathbf{z}] (s). In this case:mathbf{Z}(s)=left(sI-A ight)^{-1}mathbf{z}_0.The matrix function left(A-sI ight)^{-1} is called the resolvent matrix of the differential operator frac{mathrm{d{mathrm{d}t}-A. It is meromorphic with respect to the complex parameter sinmathbb{C} since its matrix elements are rational functions whose denominator is equal for all to det (A-sI). Its polar singularities are the eigenvalues of A, whose order equals their index for it, i.e. mathrm{ord}_{(A-sI)^{-1lambda=mathrm{idx}_A lambda.

See also

* Jordan decomposition
* Jordan normal form
* Holomorphic functional calculus
* Matrix exponential
* Logarithm of a matrix
* Dynamical system
* Bifurcation theory
* State space (controls)

Further reading

Wikimedia Foundation. 2010.