In my analysis class, we are covering tempered distributions now. I was given this two-part question in Fourier transforms of distributions.

a. We are asked to compute the Fourier transform of $\log |x|$ as a tempered distribution in $\mathbb{R}^2$. Here, $|x|$ is the 2d Euclidean norm of the vector $x$.

b. We are asked to find the fundamental solution to the Laplace equation in $\mathbb{R}^2$, $\Delta u = \delta_0$, where $u$ is a distribution and understood in the sense of weak solutions.

Here, we take the Laplacian $\Delta u = -\frac{\partial^2 u}{\partial x^2}-\frac{\partial^2 u}{\partial y^2}$. We take the Schwartz space of functions $S(\mathbb{R}^2)$ and its continuous dual space, $S'(\mathbb{R}^2)$, the space of tempered distributions. For $T \in S'(\mathbb{R}^2)$ and $\phi \in S(\mathbb{R}^2)$, we use the notation $$\langle T,\phi \rangle = T(\phi)$$ and when $T(x)$ is a function, we define $$\langle T,\phi \rangle = \int_{\mathbb{R}^2} T(x)\phi(x)dx$$ For the Foruier transform on tempered distributions, we define $$ \langle \mathcal{F}T,\phi \rangle = \langle T,\mathcal{F}\phi \rangle. $$ To be honest, I have no idea how to compute the Fourier transform of $\log|x|$ in $\mathbb{R}^2$ and how to use it to find the fundamental solution in part b, I do think I need to move to the Fourier domain in b but other than that I am lost. I thank anyone who can help with parts A and B.


The most coherent/memorable approach I know is not the most direct: first, compute that $\Delta |x|^s=s(s+n-1)\cdot |x|^{s-2}$ on $\mathbb R^n$. One can also check that the residue of the meromorphic family of tempered distributions $s\to |x|^s$ at $s=-n$ is an explicit constant multiple of $\delta$. Thus, for $n\not=2$, the limit as $s\to -n$ shows that $\Delta |x|^{2-n}$ is a constant multiple of $\delta$.

The squared factor $s^2$ disrupts this, moving the residue to the degree-one term in the (distribution-valued) Laurent expansion, but differentiating in $s$ gives reduces the degree. Thus, $$ \Delta(\log|x|\cdot |x|^s) \;=\; \Delta {\partial\over \partial s}|x|^s \;=\; {\partial\over \partial s}\Delta |x|^s \;=\; {\partial\over \partial s} s^2|x|^{s-2} $$ Taking the limit as $s\to 0$ gives $\Delta \log|x|$ in the first term, and captures the residue of $|x|^s$ in the last. Namely, $\Delta \log|x|=2\pi\cdot \delta$.

Taking Fourier transform of this shows that $|x|^2\cdot \widehat{\log|x|}$ is a constant multiple of $1$. Since $1/|x|^2$ is not locally integrable in $\mathbb R^2$, we cannot say that the Fourier transform of $\log|x|$ is (a constant multiple of) $1/|x|^2$...


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.