L\({}^ 2\) boundedness of oscillatory integral operators (Q808412)

From MaRDI portal
scientific article
Language Label Description Also known as
English
L\({}^ 2\) boundedness of oscillatory integral operators
scientific article

    Statements

    L\({}^ 2\) boundedness of oscillatory integral operators (English)
    0 references
    1991
    0 references
    The author investigates the problem when the operator of the form \[ (1)\quad (Tf)(x)=(p.v.)\int_{{\mathbb{R}}^ n}e^{i(Bx,y)}K(x-y)f(y)dy, \] ((Bx,y) is a real bilinear form, \(rank(B)=k\), K is a function which is smooth anyway from the origin, homogeneous of degree -(n-k)) is a bounded operator on \(L^ 2({\mathbb{R}}^ n).\) Let us remind a similar result by \textit{D. H. Phong} and \textit{E. M. Stein} [Acta Math. 157, 99-157 (1986; Zbl 0622.42011); Invent. Math. 86, 75-113 (1986; Zbl 0656.42009)]. They showed that if \(\mu >n-rank(B)\), and a kernel function K which is \(C^{\infty}\) away from the origin and which coincides with a homogeneous function of degree -\(\mu\) for large \(| x|\), and with a homogeneous function of degree -n for small \(| x|\), satisfies the ``cancellation condition'' \[ (2)\quad \int_{| x| =\epsilon}K(x)d\sigma (x)=0 \] for small \(\epsilon >0\), then the operator (1) is bounded on \(L^ 2({\mathbb{R}}^ n)\). The difference is that in the problem mentioned above, the kernel function K is homogeneous of critical degree \(-\mu =-(n-rank(B)).\) The result is known in extreme cases. When \(rank(B)=0 (\mu =n)\) the operator (1) is simply the classical singular integral operator, and it is bounded on \(L^ 2({\mathbb{R}}^ n)\) if, and only if, the kernel function K satisfies the cancellation condition (2) with \(\epsilon =1\). In the other extreme case \(rank(B)=n\), i.e. the bilinear operator has full rank and K is homogeneous of degree 0, Phong and Stein proved boundedness of operator (1) without any additional condition on a kernel function K. Note, that as a special case, when K is a constant, the operator (1) becomes the Fourier transform which is known to be bounded on \(L^ 2({\mathbb{R}}^ n).\) The author presents the following results (for \(0<k<n):\) (i) If the bilinear form is ``regular'' (i.e. \(Range(B)=Range(B^ t)\), where \(B^ t\) is the transpose of B), the necessary and sufficient condition for \(L^ 2\) boundedness of the operator (1) is that K has some cancellation property of type (2). (ii) If the bilinear form B is ``irregular'' (i.e. \(Range(B)\neq Range(B^ t))\) the operator (1) is always bounded on \(L^ 2({\mathbb{R}}^ n)\).
    0 references
    0 references
    0 references
    0 references
    0 references
    kernel function
    0 references
    cancellation condition
    0 references
    singular integral operator
    0 references
    Fourier transform
    0 references
    0 references
    0 references