Weighted uncertainty principles in \(L^{\infty}\) (Q1841215)

The authors determine sufficient conditions on positive weight functions \(W,V \) such that there exist continuous, strictly increasing functions \(\Phi ,\Psi \) from \([0,\infty)\) onto \([0,\infty)\) such that \[ \Phi (\left\|Wf\right\|_{\infty }/|\widehat{f}(0)|)\Psi (\|V\widehat{f}\|_{\infty }/|f(0)|)\geq 1 \] whenever \(f\) is a continuous, integrable function on the real line. The conditions on \(V,W\) are that (i) neither \(1/V\) nor \(1/W\) are integrable in neighborhoods of zero while both \(1/V,1/W\) are integrable in neighborhoods of infinity that do not include zero and (ii) both \(V,W\) satisfy doubling conditions, e.g., \(W(2t)\leq CW(t)\). Typical examples include the case in which \(W(t)=t^{\alpha }\) for some \(\alpha >1\). The integrability condition guarantees that \(H(u)=\int_{1/u}^{\infty }ds/W(s)\) is a homeomorphism of \((0,\infty)\). Then \(\Phi (u)=cH(au)\) for suitable constants \(a,c\). \(\Psi \) is defined in a similar way. The norm inequality is an analogue of the Heisenberg uncertainty principle adapted the weighted \( L^{\infty }\). The authors cite a few of the numerous papers dealing with weighted-\(L^{p}\) Fourier uncertainty principles with \(1<p<\infty \). The Poisson summation formula is applied to get the desired lower bounds.