Lower bounds for the truncated Hilbert transform (Q268236)

From MaRDI portal
Revision as of 11:52, 18 April 2024 by Importer (talk | contribs) (‎Changed an Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
In more languages
Configure
Language Label Description Also known as
English
Lower bounds for the truncated Hilbert transform
scientific article

    Statements

    Lower bounds for the truncated Hilbert transform (English)
    The paper investigates the following problem: Given two intervals \(I,J\subset \mathbb{R}\) and a real-valued function \(f\in L^2(I)\), when can \(f\) be reconstructed from knowing its Hilbert transform \(Hf\) on \(J\)? The solution depends on the relationship between \(I\) and \(J.\) There are four cases to consider: {\parindent=6mm \begin{itemize}\item[1.] The interval \(J\) covers the support \(I\) of \(f,\) that is \(I\subset J.\) \item[2.] The interval \(J\) is a subset of \(I\), that is \(J\subset I\). This case is known as the interior problem. \item[3.] The interval \(J\) does not intersect \(I\), that is \(I \bigcap J =\emptyset\). This case is known as the truncated Hilbert transform with a gap \item[4.] The interval \(J\) overlaps with \(I\). This case is known as the truncated Hilbert transform with overlap. \end{itemize}} The paper deals with Cases 3 and 4, and consequently throughout this review we assume that \(I,J\) are as described in Cases 3 and 4. Now let \(f\in BV(I)\) be a function of bounded variation supported on \(I.\) It was shown before that there exists a positive function \(h:[0, \infty)\rightarrow \mathbb{R}^+ \) such that \[ \left\| Hf\right\|_{L^2(J)}\geq h\left( \frac{|f|_{TV}}{\left\| f\right\|_{L^2(I)}} \right) \left\| f\right\|_{L^2(I)} , \] where \(|. |_{TV}\) denotes the total variation of \(f.\) The authors conjecture that \[ h(k)\geq c_1 e^{-c_2 k} , \] where \(c_1, c_2 >0\) are constants depending only on \(I\) and \(J.\) Because for weakly differentiable functions, we have \[ \left| f\right|_{TV}=\int_I \left| f_x(x) \right| dx, \] we can identify the total variation of \(f\) with \(\left\| f_x \right\|_{L^1 (I)}.\) It is conjectured that \[ \left\| Hf\right\|_{L^2(J)}\geq c_1 \exp\left( -c_2 \frac{\left\| f_x \right\|_{L^1(I)}}{\left\| f\right\|_{L^2(I)}} \right) \left\| f\right\|_{L^2(I)} . \] However, the authors' first result concerning that conjecture may be stated as follows: if \(f\) is weakly differentiable and \(f\in H^1(I),\) then \[ \left\| Hf\right\|_{L^2(J)}\geq c_1 \exp\left( -c_2 \frac{\left\| f_x \right\|_{L^2(I)}}{\left\| f\right\|_{L^2(I)}} \right) \left\| f\right\|_{L^2(I)} , \] where \(H^1\) is the Sobolev space \(W^{1,2}.\) Other results of similar types are derived, such as \[ \left\| Hf\right\|_{L^2(J)}\geq c_1 \exp\left( -c_2 \frac{| f|^2_{TV}}{\left\| f\right\|^2_{L^2(I)}} \right) \left\| f\right\|_{L^2(I)} , \] where \(f\in W^{1,1}.\) The authors also discuss the following more general problem. Let \(I,J\subset \mathbb{R}\) be disjoint intervals and let \(T:L^2 (I)\rightarrow L^2(J)\) be an integral operator of convolution type, \[ \left(Tf\right) (x)=\int_I K(x-y)f(y)dy , \] for some kernel \(K,\) under what conditions does one expect an inequality of the type \[ \left\|Tf\right\|_{L^2(J)}\geq h\left( \frac{| f|_{TV}}{\left\| f\right\|_{L^2(I)}}\right)\left\| f\right\|_{L^2(I)} \] to hold true for some positive function \(h:[0, \infty)\rightarrow \mathbb{R}^+ \) ?
    Hilbert transform
    truncated Hilbert transform
    truncated data
    total variation
    integral operator of convolution type

    Identifiers