# Floquet theory

A theory concerning the structure of the space of solutions, and the properties of solutions, of a linear system of differential equations with periodic coefficients

$$\tag{1 } x ^ \prime = \ A ( t) x,\ \ t \in \mathbf R ,\ \ x \in \mathbf R ^ {n} ;$$

the matrix $A ( t)$ is periodic in $t$ with period $\omega > 0$ and is summable on every compact interval in $\mathbf R$.

1) Every fundamental matrix $X$ of the system (1) has a representation

$$\tag{2 } X ( t) = F ( t) \ \mathop{\rm exp} ( tK),$$

called the Floquet representation (see ), where $F ( t)$ is some $\omega$- periodic matrix and $K$ is some constant matrix. There is a basis $x _ {1} \dots x _ {n}$ of the space of solutions of (1) such that $K$ has Jordan form in this basis; this basis can be represented in the form

$$x _ {i} = \ ( \psi _ {1i} \mathop{\rm exp} ( \alpha _ {i} t) \dots \psi _ {ni} \mathop{\rm exp} ( \alpha _ {i} t)),$$

where $\psi _ {ki}$ are polynomials in $t$ with $\omega$- periodic coefficients, and the $\alpha _ {i}$ are the characteristic exponents (cf. Characteristic exponent) of the system (1). Every component of a solution of (1) is a linear combination of functions of the form (of the Floquet solutions) $\psi _ {ki} \mathop{\rm exp} ( \alpha _ {i} t)$. In the case when all the characteristic exponents are distinct (or if there are multiple ones among them, but they correspond to simple elementary divisors), the $\psi _ {ki}$ are simply $\omega$- periodic functions. The matrices $F ( t)$ and $K$ in the representation (2) are, generally speaking, complex valued. If one restricts oneself just to the real case, then $F ( t)$ does not have to be $\omega$- periodic, but must be $2 \omega$- periodic.

2) The system (1) can be reduced to a differential equation with a constant matrix, $y ^ \prime = Ky$, by means of the Lyapunov transformation

$$\tag{3 } x = F ( t) y,$$

where $F ( t)$ and $K$ are the matrices from the Floquet representation (2) (see ). The combination of representation (2) together with the substitution (3) is often called the Floquet–Lyapunov theorem.

3) Let $\{ \alpha _ {1} \dots \alpha _ {l} \}$ be the spectrum of the matrix $K$. For every $\alpha \in \mathbf R$ such that $\alpha \neq \mathop{\rm Re} \alpha _ {j}$, $j = 1 \dots l$, in view of (2) the space $\mathbf R ^ {n}$ splits into the direct sum of two subspaces $S _ \alpha$ and $U _ \alpha$

$$( \mathbf R ^ {n} = \ S _ \alpha + U _ \alpha ,\ \ S _ \alpha \cap U _ \alpha = \emptyset )$$

such that

$$\lim\limits _ {t \rightarrow + \infty } \ \mathop{\rm exp} (- \alpha t) V ( t) x ( 0) = 0 \ \iff \ \ x ( 0) \in S _ \alpha ,$$

$$\lim\limits _ {t \rightarrow - \infty } \mathop{\rm exp} (- \alpha t) V ( t) x ( 0) = 0 \ \iff \ x ( 0) \in U _ \alpha ;$$

here $V ( t)$ is the fundamental matrix of (1) normalized at zero. This implies exponential dichotomy of (1) if $\mathop{\rm Re} \alpha _ {j} \neq 0$ for any $j = 1 \dots l$.

How to Cite This Entry:
Floquet theory. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Floquet_theory&oldid=46944
This article was adapted from an original article by Yu.V. Komlenko (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article