FBI transform

From Wikipedia, the free encyclopedia

In mathematics, the FBI transform or Fourier-Bros-Iaglonitzer transform is a non-linear version of the Fourier transform developed by the French mathematical physicists Jacques Bros and Daniel Iagolnitzer in order to characterise the local analyticity of functions (or distributions) on Rn. The transform provides an alternative approach to analytic wave front sets of distributions, developed independently by the Japanese mathematicians Mikio Sato, Masaki Kashiwara and Takahiro Kawai in their approach to microlocal analysis. It can also be used to prove the analyticity of solutions of analytic elliptic partial differential equations as well as a version of the classical uniqueness theorem, strengthening the Cauchy-Kowalevski theorem, due to the Swedish mathematician Erik Holmgren (1873-1943).

Contents

[edit] Definitions

The Fourier transform of a Schwartz function f in S(Rn) is defined by

({\mathcal F}f)(t) = (2\pi)^{-n / 2} \int_{{\mathbf R}^n}f(x) e^{-ix \cdot t}\, dx.

The FBI transform of f is defined for a ≥ 0 by

({\mathcal F}_a f)(t,y) = (2\pi)^{-n / 2} \int_{{\mathbf R}^n}f(x)e^{-a |x-y|^2/2} e^{-ix \cdot t}\, dx.

Thus, when a = 0, it essentially coincides with the Fourier transform.

The same formulas can be used to define the Fourier and FBI transforms of tempered distributions in S'(Rn).

[edit] Inversion formula

The Fourier inversion formula

f(x)= {\mathcal F}^2 f(-x)

allows a function f to be recovered from its Fourier transform.

In particular

f(0) = (2\pi)^{-n/2} \int_{{\mathbf R}^n} {\mathcal F} f(t) \, dt.

Similarly. at a positive value of a, f(0) can be recovered from the FBI transform of f(x) and xkf(x) by the inversion formula

f(0)= (2\pi)^{-n/2}\int_{{\mathbf R}^n} {\mathcal F}_a (f)(t,0) \, dt + (2\pi)^{-n/2} \int_{{\mathbf R}^n} \sum_{k=1}^n {2ait_k\over |t|} \cdot {\mathcal F}_a (x_kf)(t,0)\, dt.

This formula can be proved by calculating the coefficients of am in the analytic function of a defined by the right hand side. These coefficients can be expressed in terms of the Fourier transform {\mathcal F} f, the Euler operator

E=\sum_{k=1}^n t_k \partial_{t_k}

and the Laplacian operator

L=-\sum_{k=1}^n \partial_{t_k}^2,

and easily computed using integration by parts.

[edit] Criterion for local analyticity

Bros and Iagolnitzer showed that a distribution f is locally equal to a real analytic function near x if and only if its FBI transform satisfies an inequality of the form

|({\mathcal F}_af)(at,y)| \le C e^{-\varepsilon a},

for a > 0, y near x and |t| sufficiently large, with C and ε positive constants.

[edit] Holmgren's uniqueness theorem

A simple consequence of the Bros and Iagolnitzer characterisation of local analyticity is the following regularity result of Lars Hörmander and Mikio Sato (Sjöstrand 1982).

Theorem. Let P be a elliptic partial differential operator with analytic coefficients defined on an open subset X of Rn. If Pf is analytic in X, then so too is f.

When "analytic" is replaced by "smooth" in this theorem, the result is just Hermann Weyl's classical lemma on elliptic regularity, usually proved using Sobolev spaces (Warner 1983). It is a special case of more general results involing the analytic wave front set (see below), which imply Holmgren's classical strengthening of the Cauchy-Kowalevski theorem on linear partial differential equations with real analytic coefficients. In modern language, Holmgren's uniquess theorem states that any distributional solution of such a system of equations must be analytic and therefore unique, by the Cauchy-Kowalevski theorem.

[edit] The analytic wave front set

The analytic wave front set or singular spectrum WFA(f) of a distribution f (or more generally of a hyperfunction) can be defined in terms of the FBI transform (Hörmander 1983) as the the complement of the conical set of points (x, λ·ξ) (λ > 0) such that the FBI transform satisfies the Bros-Iagolnitzer inequality

|({\mathcal F}_af)(at,y)| \le C e^{-\varepsilon a},

for all a > 0, y near x and t = λ·ξ, with |t| sufficiently large. J.M. Bony (Sjöstrand 1982, Hörmander 1983) proved that this definition coincided with other definitions introduced independently by Sato, Kashiwara and Kawai and by Hörmander. If P is an mth order linear differential operator having analytic coefficients

P =\sum_{|\alpha|\le m} a_\alpha(x) D^\alpha,

with principal symbol

σP(x,ξ) = ∑ aα(x)ξα,
| α | = m

and characteristic variety

{\rm char}\, P =\{(x,\xi): \xi\ne0, \, \sigma_P(x,\xi) =0\},

then

  • WF_A(Pf) \subseteq WF_A(f)
  • WF_A(f) \subseteq WF_A(Pf) \cup {\rm char}\, P.

In particular, when P is elliptic, char P = ø, so that

WFA(Pf) = WFA(f).

This is a strengthening of the analytic version of elliptic regularity mentioned above.

[edit] References

  • Folland, Gerald B. (1989), Harmonic Analysis in Phase Space, vol. 122, Annals of Mathematics Studies, Princeton University Press, ISBN 0-691-08528-5 
  • Gårding, Lars (1998), Mathematics and Mathematicians: Mathematics in Sweden Before 1950, American Mathematical Society, ISBN 0821806122 
  • Hörmander, Lars (1983), Analysis of Partial Differential Operators I, Springer-Verlag, ISBN 3-540-12104-8  (Chapter 9.6, The Analytic Wavefront Set.)
  • Iagolnitzer, Daniel (1975), Microlocal essential support of a distribution and local decompositions - an introduction. In Hyperfunctions and theoretical physics, vol. 449, Lecture Notes in Mathematics, Springer-Verlag, pp. 121-132 
  • Krantz, Steven & Harold R., Parks (1982), A Primer of Real Analytic Functions, Birkhäuser, ISBN 0817642641 
  • Sjöstrand, Johannes (1982), “Singularités analytiques microlocales. [Microlocal analytic singularities]”, Astérisque 95: 1-166 
  • Trèves, François (1992), Hypo-analytic structures: Local theory, vol. 40, Princeton Mathematical Series, Princeton University Press, ISBN 0-691-08744-X  (Chapter 9, FBI Transform in a Hypo-Analytic Manifold.)
  • Warner, Frank (1983), Foundations of differential geometry and Lie groups, vol. 94, Graduate texts in mathematics, Springer-Verlag, ISBN 0387908943