Convolution and Fourier transform over the spaces \({\mathcal K}_{p,k}'\), \(p>1\) (Q2566517)

To generalize the method used in defining the Fourier transform for tempered distributions, the authors introduce spaces of generalized functions denoted by \(K_{p,k}'(\mathbb R^n)\), \(p>1\). Let \(a\in \mathbb N^n\), \(k\in \mathbb Z\), \(k<0\), then \[ K_{p,k}=\left\{f;\partial^a f\in C(\mathbb R^n),\;|\partial^a f(x)|\exp(k|x|^p)<\varepsilon,\;|x|>C(f,a,\varepsilon),\;a\in \mathbb N^n\right\}. \] On \(K_{p,k}\), the family of semi-norms \(q_{k,a}(f)=\max|\partial^a f(x)|\times \exp(k|x|^p)\), \(x\in \mathbb{R}^n\), \(a\in \mathbb{N}^n\) is defined. The space of continuous linear functionals on \(K_{p,k}\) is denoted by \(K_{p.k}'\). The convolution \(f*k\) for \(f,g\) \(K_{p.2^pk}'\) is defined as \(\langle f*g,\varphi\rangle =\langle f(x),\langle g(y),\varphi(x+y)\rangle\rangle\) and it is proved that \(f*g\in K_{p,k}'\). If \(f\in K_{p,k}'\), then by definition \((Ff)(y) =\langle f(x),\exp(ixy)\rangle\), \(y\in \mathbb R^n\). The inversion formula for the Fourier transform is proved as well.