Abstract:The stretched Gaußian function $f(\mathbf{x})=\exp \left(-\|\mathbf{x}\|^s\right)$, as a real function defined on $\mathbb{R}^d$, has found numerous applications in mathematics and physics. For instance, to describe results from spectroscopy or inelastic scattering, the Fourier transform of the stretched Gaußian function is needed. For $s \in(0,2]$, we prove that the Fourier transform of $f(\mathbf{x})=\exp \left(-\|\mathbf{x}\|^s\right)$ is everywhere positive on $\mathbb{R}^d$.
What problem does this paper attempt to address?
This paper aims to solve the problem of whether the Fourier transform of the stretched Gaussian function is positive everywhere on \( \mathbb{R}^d \). Specifically, for the parameter \( s\in(0, 2] \), the author proves that the Fourier transform \( \hat{f}(\xi) \) of the stretched Gaussian function \( f(x)=\exp(-\|x\|^s) \) is positive throughout the \( \mathbb{R}^d \) space.
### Main Conclusions
- **Theorem**: For \( s\in(0, 2] \), the Fourier transform \( \hat{f}(\xi) \) of the stretched Gaussian function \( f(x)=\exp(-\|x\|^s) \) satisfies:
\[
\hat{f}(\xi)=\int_{\mathbb{R}^d} f(x)e^{2\pi i\langle x,\xi\rangle}\,dx > 0
\]
for all \( \xi\in\mathbb{R}^d \).
### Proof Ideas
1. **Lebesgue Integrability**: First, prove that \( f(x) \) is Lebesgue integrable, that is, \( f\in L^1(\mathbb{R}^d) \). Through spherical coordinate transformation and integration of area elements, we can obtain:
\[
\int_{\mathbb{R}^d} f(x)\,dx=\frac{2\pi^{d/2}}{\Gamma(d/2)}\int_0^{\infty}x^{d - 1}\exp(-x^s)\,dx=\frac{2\pi^{d/2}}{s\cdot\Gamma(d/s)\Gamma(d/2)}
\]
2. **Completely Monotonic Function**: Utilize the properties of completely monotonic functions, especially Bernstein's theorem, to prove that \( \phi(x)=\exp(-\sqrt{x^s}) \) is completely monotonic. Therefore, there exists a bounded increasing function \( g \) such that:
\[
\phi(x)=\int_0^{\infty}e^{-xt}\,dg(t)
\]
3. **Non - negativity of Fourier Transform**: Through Fubini's theorem and exchange of integration order, prove that \( \hat{f}(\xi)\geq0 \):
\[
\hat{f}(\xi)=\int_{\mathbb{R}^d} f(x)e^{-2\pi i\langle x,\xi\rangle}\,dx=\int_0^{\infty}\left(\int_{\mathbb{R}^d}\exp(-t\|x\|^2-2\pi i\langle x,\xi\rangle)\,dx\right)dg(t)
\]
where:
\[
\int_{\mathbb{R}^d}\exp(-t\|x\|^2-2\pi i\langle x,\xi\rangle)\,dx=\left(\frac{\pi}{t}\right)^{d/2}\exp\left(-\frac{\pi^2}{t}\|\xi\|^2\right)\geq0
\]
4. **Non - zero**: Assume that \( \hat{f}(\xi) = 0 \), and through the mean value theorem of Riemann - Stieltjes integral, derive a contradiction, thereby proving that \( \hat{f}(\xi)>0 \).
### Conclusion
Through the above steps, the author successfully proves that the Fourier transform of the stretched Gaussian function is everywhere on \( \mathbb{R}^d \).