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

I want to calculate the first and second distributional derivate of the $2\pi$-periodic function $f(t) = \frac{\pi}{4} |t|$, it is $$ \langle f', \phi \rangle = - \langle f, \phi' \rangle = -\int_{-\infty}^{\infty} f \phi' \mathrm{d}x $$ and $$ \langle f'', \phi \rangle = \langle f, \phi'' \rangle = \int_{-\infty}^{\infty} f \phi'' \mathrm{d}x $$ so have to evaluate those integrals, but other than $\phi$ is smooth and has compact support i know almost nothing about $\phi$, so it is enough to integrate over finite bound, but how does i use the definition of $f$? So how could i calculate those integrals?

share|cite|improve this question
Integrate by parts, and use the fact that $\phi(x)\to 0$ when $x\to \infty$ – Norbert Jun 23 '12 at 12:37
for the first term i got $-\int_{-\infty}^{\infty} f \phi' \mathrm{d}x = -[ f \phi ]_{-\infty}^{\infty} + \int_{-\infty}^{\infty} f' \phi \mathrm{d}x$, the first summand vanishes because of the compact support ($\phi(x) \to 0$ for $x\to \infty$). the derivate of $f$ is $f''(t) = -\pi/4$ for $t \in [-\pi,0]$ and $f''(t) = \pi/4$ for $t\in [0,\pi]$, and $f(t) = f(t + 2\pi)$. so now i have difficulty to bring my knowledge of $f'$ in an derivation of an expression for the derivational derivate? – Stefan Jun 23 '12 at 12:54
Your conclusions is wrong. After you have integrated by patrs you get must get derivative of $f$, not a second derivative. – Norbert Jun 23 '12 at 12:58
oh yes, i am sorry that was a typo, i mean the first derivative $f'$. – Stefan Jun 23 '12 at 13:01
So have you heard of Heaviside function and its derivative? – Norbert Jun 23 '12 at 13:05

Let $G(t)=\max(t,0)$. The $G'=H$ where $H$ is the Heaviside function ($H(t)=1$ if $t>0$ and $0$ otherwise). Indeed, for any test function $\phi$ we have $$\langle G, \phi'\rangle = \int_0^{\infty} t\phi'(t)\,dt = - \int_0^{\infty} \phi(t)\,dt = - \langle H, \phi\rangle$$ in agreement with the definition of the distributional derivative. Also, $H'=\delta_0$ by a similar argument: $$\langle H, \phi'\rangle = \int_1^{\infty} \phi'(t)\,dt = -\phi(0) = -\langle \delta_0, \phi\rangle$$

Any piecewise linear function can be written in terms of the translates of $G$. To do this for your function, you can begin with $f_0(t)=\frac{\pi}{4}G(t)-\frac{\pi}{2}G(t-\pi)+\frac{\pi}{4}G(t-2\pi)$ which agrees with $f$ on $[0,2\pi]$ and is zero elsewhere. Note that $f(t)=\sum_{n\in\mathbb Z}f_0(t+2\pi n )$, where the sum converges in a very strong sense: for any finite interval $I$, there exists $N$ such that $f(t)=\sum_{|n|\le N}f_0(t+2\pi n )$ for $t\in I$. This kind of convergence implies convergence of all distributional derivatives, since test functions have compact support.

share|cite|improve this answer

ok, i tried to solve, for the first and second derivate i got $$ f'(t) = \frac{\pi}{4}(2H(t)-1) $$ (where $H(t)$ denotes the Heaviside-Function) on $[-\pi, \pi]$ and $f'(t) = f'(t+2\pi)$. For the second derivate i got $$ f''(t) = \frac{\pi}{4}\delta_0(t) $$ (where $\delta_0$ ist the Dirac-Delta at $0$) on $[-\pi, \pi]$ and $f''(t) = f''(t+2\pi)$. So for the distibutional derivate $$ \langle f', \phi \rangle = - \langle f, \phi' \rangle = -\int f \phi' = -\int f'\phi = \langle \frac{\pi}{4}(2H-1), \phi \rangle $$ and $$ \langle f'', \phi \rangle = \langle \frac{\pi}{4}\delta_0, \phi \rangle $$ is this right and is my notation acceptable?

share|cite|improve this answer
The second derivative should have a negative multiple of Dirac mass at $\pi+2\pi n$. – user31373 Jun 23 '12 at 20:38
so i should be $$f''(t) = \frac{\pi}{4}\delta_0(t) - \frac{\pi}{4}\delta_0(t + \pi)$$ – Stefan Jun 23 '12 at 21:06
Yes, and then by periodicity. The angle at the endpoints of the initially defined period is easy to miss. – user31373 Jun 23 '12 at 21:08

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.