Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In short, I am trying to find sufficient conditions for an approximation to the identity function $K_h$ so that, for $h$ small enough and fixed, the asymptotics at the origin of an $L^1 \cap L^2$ function $f$ with a singularity at the origin remain unchanged when convolved with $K_h$.

More precisely, let $f \in L^1 \cap L^2$ such that for some constant $0<c<1/2$ and some $\delta>0$

\begin{align} f(x) &= x^{-c} &\forall x \in (0,\delta] \, \, \, \, \, \\ f(x) &= 0 &\forall x \in [-\delta,0], \end{align}

and $K_h(x):=h^{-1}K(x/h)$ where $K$ is a band-limited kernel function (i.e. whose Fourier transform is compactly supported without loss of generality on $[-1,1]$, so $K$ cannot be compactly supported!), \begin{equation*} %\label{EqKProp} \int_{\mathbb{R}} K(x)dx=1 \qquad and \qquad \left|K(x)\right|+\left|K'(x)\right|\leq C (1+\left|x\right|)^{-\beta} \end{equation*} for some real constants $C>0$ and $\beta>2$.

The question is, is there an $h$ small enough such that (possibly under stronger conditions on $K$, respecting those above) for some constants $\varepsilon, \delta'>0$ and $0<c_1\leq c_2$

\begin{align} c_1 x^{-c} \leq &(f \ast K_h) (x) \leq c_2 x^{-c} &\forall x \in(0,\delta'] \, \, \, \, \, \, \\ -\varepsilon \leq &(f \ast K_h) (x) \leq \varepsilon &\forall x \in [-\delta', 0] ? \end{align}

Obviously the constants may depend on $h$, but if they do we would like $\delta'$ and $c_1$ to stay bounded below by some positive constant independent of $h$, and $\varepsilon $ and $c_2$ to stay bounded above by some positive constant independent of $h$.

The standard results on convolutions tell us that $\| (f\ast K_h)(x) - x^{-c} 1_{(0,\delta]} \| \to 0$ as $h\to 0$ on $[-\delta, \delta ]$ in the $L^1$ and $L^2$ norm and also uniformly on compact subsets of $(0,\delta)$ and $(-\delta,0)$, but I cannot seem to prove the above result.

In fact, I am trying to prove it for more general asymptotics, but proving it for the case above would be a good starting point. Any help or references will be greatly appreciated.

share|cite|improve this question

As I argue in the next paragraph, the result I was looking for is too strong to be true. However, I do have a proof for a weaker result which essentially allows us to treat $(f\ast K_h)(x)$ as $x^{-c}1_{(0,\delta]}$ around the origin, and this is what I was looking for. I will include the weaker result and its proof when I have it written up in LaTex.

To see why the result above cannot be true note that, under its assumptions, $f\in L^1$ and $K_h\in L^{\infty}$ for any $h>0$ fixed, so $(f\ast K_h)$ is bounded and uniformly continuous (see Proposition 8.8 in Folland, Real analysis, if not familiarised with the area). Hence for any $\delta'>0$ it is impossible to find an $h>0$ and some constants $0<c_1\leq c_2$ such that $$ c_1 x^{-c} \leq (f\ast K_h)(x) \leq c_2 x^{-c} \qquad \forall x\in (0,\delta'].$$


share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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