A remark on an endpoint Kato-Ponce inequality. (Q2343845): Difference between revisions

The results of the paper may be considered as a step towards an \(L^\infty \)-endpoint Kato-Ponce inequality \[ \|D^s(fg)\|_{L^\infty}\leq C\big (\,\|D^sf\|_{L^\infty}\|g\|_{L^\infty} +\|D^sg\|_{L^\infty}\|f\|_{L^\infty}\,\big) \] (where \(f, g\in \mathcal {S}(\mathbb R^n)\), \(s>0\), \(D^s\) is the \(s\)-derivative operator, \(C>0\) is a constant depending only on \(s\) and \(n\)) whose validity is not known. In particular, one of the results can be regarded as a combination of the Leibniz rule and bilinear version of the classical Gagliardo-Nirenberg interpolation inequalities (see e.g. Theorem~2.44 of the book [\textit{H. Bahouri, J.-Y. Chemin} and \textit{R. Danchin}, Fourier analysis and nonlinear partial differential equations. Berlin: Heidelberg (2011; Zbl 1227.35004)]): \[ \|D^s(fg)\|_{L^\infty}\leq C\big (\, \|D^rf\|_{L^\infty}^\alpha \|D^tf\|_{L^\infty}^\beta \|g\|_{L^\infty} + \|f\|_{L^\infty}\|D^rg\|_{L^\infty}^\alpha \|D^tg\|_{L^\infty}^\beta \,\big), \] where \(f, g\in \mathcal {S}(\mathbb R^n)\), \(0\leq r<s<t\), \(\alpha =\frac {t-s}{t-r}\), \(\beta =\frac {s-r}{t-r}\), and the constant \(C>0\) may depend only on \(r\), \(s\), \(t\) and \(n\).