nLab pseudodifferential operator




Pseudodifferential operators generalize differential operators and are of similar importance to the theory of partial differential equations as Schwartz distributions, see also microlocal analysis.


Let X nX \subset \mathbb{R}^n be open. A pseudodifferential operator is a Fourier integral operator of the form

A:C 0 (X)𝒟(X) A: C^{\infty}_0 (X) \to \mathcal{D}'(X)
Au(x)=1(2π) ne i(xy)θa(x,y,θ)u(y)dydθ A u(x) = \frac{1}{(2 \pi)^n} \int \int e^{i (x-y) \theta} a(x, y, \theta) u(y) dy d\theta

where the function aa, called the symbol of the pseudodifferential operator AA, belongs to the space S ρ,δ m(X×X× n)S^m_{\rho, \delta}(X \times X \times \mathbb{R}^n) defined below.

If \mathcal{F} denotes the Fourier transform a short hand notation for this definition is Au= 1(au)A u = \mathcal{F}^{-1} (a \, \mathcal{F}u), put in words: Fourier transform uu, multiply with aa and transform back.

Definition of symbols

If one allows arbitrary functions as symbols there will be no control of the behaviour of the associated pseudodifferential operators, of course. In order to get a theory where these operators are, for example, continuous with respect to the standard topologies on the topological vector spaces that they are defined on, some assumptions have to be made. Different levels of generality of the theory correspond to different assumptions about the symbols. One standard symbol space is defined as follows:

Let X nX \subset \mathbb{R}^n be open, 0ρ1,0δ1,m,n,n00 \le \rho \le 1, 0 \le \delta \le 1, m \in \mathbb{R}, n \in \mathbb{N}, n \neq 0.

  • definition: S ρ,δ mS^m_{\rho, \delta} is the space of all aC (X× n)a \in C^{\infty}(X \times \mathbb{R}^n) such that for all compact KXK \Subset X and all α,β n\alpha, \beta \in \mathbb{N}^n there is a constant C K,α,βC_{K, \alpha, \beta} such that
    | x α θ βa(x,θ)|C K,α,β(1+|θ|) mρ|β|+δ|α| | \partial^{\alpha}_x \, \partial^{\beta}_{\theta} \, a(x, \theta)| \leq C_{K, \alpha, \beta} (1 + | \theta |)^{m - \rho | \beta | + \delta | \alpha |}

    The space S ρ,δ mS^m_{\rho, \delta} is called the space of symbols of order mm and of type (ρ,δ)(\rho, \delta).

It is easy to see that every space S ρ,δ mS^m_{\rho, \delta} is a Fréchet space: every X nX \subset \mathbb{R}^n open has a compact exhaustion, that is an increasing sequence (K i)(K_i) with each K iXK_i \Subset X compact such that i=1 K i=X\bigcup_{i = 1}^{\infty} K_i = X, and one can define a countable family of seminorms via

P K i,α,β=sup (x,θ)K i× n| x α θ βa(x,θ)|(1+|θ|) mρ|β|+δ|α| P_{K_i, \alpha, \beta} = \operatorname{sup}_{(x, \theta) \in K_i \times \mathbb{R}^n} \frac{ | \partial^{\alpha}_x \, \partial^{\beta}_{\theta} \, a(x, \theta)|}{(1 + | \theta |)^{m - \rho | \beta | + \delta | \alpha |}}

The space of symbols of order - \infty is defined to be

S = mS ρ,δ m S^{- \infty} = \bigcap_{m \in \mathbb{R}} S^m_{\rho, \delta}

Conversly the symbols of order \infty are defined by

S = mS ρ,δ m S^{\infty} = \bigcup_{m \in \mathbb{R}} S^m_{\rho, \delta}

Symbols of order - \infty are often called smoothing and their operators smoothing operators. The reason for this is that their pseudodifferential operators map distribution spaces into spaces of smooth functions, for example:


Every differential operator is a pseudodifferential operator

We restrict ourselfes to one dimension for simplicity, let D:=iddxD := -i \frac{d}{dx} and write a differential operator PP as

P(x,D):= k=0 nf n(x)D k P(x, D) := \sum_{k = 0}^n f_n(x) D^k

with given functions f nf_n. Then PP is a pseudodifferential operator with symbol a(x,y,θ)=P(x,θ)a(x, y, \theta) = P(x, \theta). The symbol of a differential operator therefore is a polynom in θ\theta, which motivates a part of the definition of symbol classes below: We expect that the growth of the symbol in θ\theta is polynomial at most, and the degree of the bounding polynomial decreases by 11 if we apply differentiation in θ\theta to the symbol.


Smoothing theorem

The pseudodifferential operator of a smoothing symbol maps \mathcal{E}', the dual space of C C^{\infty}, into 𝒮\mathcal{S}, the Schwartz space of rapidly decreasing smooth functions.

Propagation of singularities theorem


An elementary and short introduction can be found here:

  • Xavier Saint Raymond: Elementary Introduction to the Theory of Pseudodifferential Operators (ZMATH entry)
  • Sylvain Carpentier, Alberto De Sole, Victor G. Kac, Some algebraic properties of differential operators, arxiv/1201.1992
  • Lars Hörmander, The analysis of linear partial differential operators, in 4 vols.: I. Distribution theory and Fourier analysis, II. Differential operators with constant coefficients, III. Pseudo-differential operators, IV. Fourier integral operators.

Last revised on November 23, 2017 at 10:38:22. See the history of this page for a list of all contributions to it.