On the Pauli problem (Q677688): Difference between revisions

The Fourier transform of a function \(f\in L_2 (\mathbb{R}^n)\) is given by \[ \widehat f(s)= \int_{\mathbb{R}^n} f(x) \exp \bigl(-2 \pi i(x,s) \bigr)dx. \] The Pauli problem (conjecture) is to describe a linear independent pair of functions \(f_1\), \(f_2 \in L_2 (\mathbb{R}^n)\) such that \[ \bigl|f_1(x) \bigr|=\bigl|f_2(x) \bigr|\;\forall x\in \mathbb{R}^n \bmod 0, \quad \bigl|\widehat f_1(s) \bigr|=\bigl|\widehat f_2 (s)\bigr|\quad \forall s\in \mathbb{R}^n \bmod 0. \tag{1} \] The paper contains two results: (i) For \(n=1\), a counterexample to the conjecture is given by constructing pairs of functions to satisfy the condition (1). That is by setting \[ f_1(x)= \rho(x) \exp \bigl(i\varphi (x)\bigr), \quad f_2 (x)= \rho(x) \exp\bigl(-i \varphi (u-x) \bigr), \tag{2} \] where \(\rho\in L_2(R)\), \(\rho\geq 0\), \(\rho(x)= \rho (a-x)\) \(\forall x\in R\bmod 0\). If \(\exp (\varphi (x)+ \varphi (a-x)) \neq\) constant, then \(f_1\) and \(f_2\) are linearly independent. (ii) For \(n\geq 1\), take measurable complex-valued functions \(\alpha (x)\), \(\beta (s)\), \(x,s \in \mathbb{R}^n\) such that \[ \bigl|\alpha(x) \bigr|=1 \text{ for }x \in\mathbb{R}^n \bmod 0,\quad \bigl|\beta (s)\bigr|=1\text{ for } s\in\mathbb{R}^n \bmod 0. \tag{3} \] Then consider \(f_1\), \(f_2 \in L_2 (\mathbb{R}^n)\) \[ f_2(x) =\alpha (x)f_1 (x)\quad \forall x\in \mathbb{R}^n \bmod 0,\quad \widehat f_2(s) =\lambda \beta(s) \widehat f_1(s) \quad \forall s\in \mathbb{R}^n \bmod 0 \tag{4} \] for a number \(\lambda\) with \(|\lambda |=1\). The paper solves this problem for the case \[ \alpha (x)= \exp \bigl(\pi i(ax,x) \bigr), \quad\beta (s)= \exp \bigl(\pi i(bs,s) \bigr), \] where \(a\) and \(b\) are symmetric real matrices of order \(n\) and \((\cdot, \cdot)\) is the usual inner product on \(\mathbb{R}^n\).