1
$\begingroup$

I have a problem understanding the meaning of the delta-function on the sense of distributions. E.g. I have the following equation:

$$\left(\frac{d}{dt} \theta(t) \right) f(t) = \delta(t) f(t)$$

What does this mean for the function $f(t)$? I cannot really understand what the delta function does here in the sense of a distribution.

$\endgroup$
  • $\begingroup$ A "distribution" is a functional that assigns a number to every function. The "delta function" assigns the number f(0) to the function f. $\delta(t)f(t)= f(0)$ $\endgroup$ – user247327 Dec 11 '15 at 13:57
  • $\begingroup$ Thanks! Actually I need to show that $\delta(t) f(t) = \delta(t)$. I know, that f(0) = 1, so I think that this is correct then? Or does the $\delta(t)$ disappear then after acting on f(t)? $\endgroup$ – Darius Dec 11 '15 at 14:05
2
$\begingroup$

I assume $f$ is $\mathcal{C}^{\infty}(\mathbb{R})$.

$\delta f$ is a nonsense, however $f\delta$ has a meaning. One is only allowed to multiply a distribution by a smooth function ($\mathcal{C}^{\infty}(\mathbb{R})\times\mathcal{D}'(\mathbb{R})$), the product of a smooth function by a distribution is not defined.

Definition. Let $T\in\mathcal{D}'(\mathbb{R})$ and $f\in\mathcal{C}^{\infty}(\mathbb{R})$, one define $fT$ by: $$\forall\varphi\in\mathcal{C}^{\infty}_0(\mathbb{R}),\langle fT,\varphi\rangle=\langle T,f\varphi\rangle.$$

Remark. $f\varphi\in\mathcal{C}^{\infty}_0(\mathbb{R})$, since $\textrm{supp}(f\varphi)\subseteq\textrm{supp}(\varphi)$.

One has the:

Proposition. Let $T\in\mathcal{D}'(\mathbb{R})$ and $f\in\mathcal{C}^{\infty}(\mathbb{R})$, $fT\in\mathcal{D}'(\mathbb{R})$.

Proof. Since $T$ is a distribution, $fT$ is cleary linear map from $\mathcal{C}^{\infty}_0(\mathbb{R})$ to $\mathcal{C}^{\infty}_0(\mathbb{R})$. Let $K$ be a compact of $\mathbb{R}$, since, $T$ is a distribution there exists $C\in\mathbb{R}_+^*$ and $k\in\mathbb{N}$ such that: $$\forall\varphi\in\mathcal{C}_0^{\infty}(\mathbb{R}),\textrm{supp}(\varphi)\subseteq K,|\langle T,\varphi\rangle|\leqslant C\sum_{i=1}^n\sup_{K}\left|\varphi^{(i)}\right| $$ Let $\varphi\in\mathcal{C}^{\infty}_0(\mathbb{R})$, $\textrm{supp}(f\varphi)\subset K$ and one has: $$\begin{align}|\langle fT,\varphi\rangle|&=|\langle T,f\varphi\rangle|\\&\leqslant C\sum_{i=1}^k\sup_{K}\left|(f\varphi)^{(i)}\right|\\&\leqslant C\sum_{i=1}^k\sup_{K}\left|\sum_{j=0}^i{i\choose j}f^{(j)}\varphi^{(i-j)}\right|\\&\leqslant C\sum_{i=1}^k\sum_{j=0}^i{i\choose j}\sup_{K}\left|f^{(i)}\right|\sup_{K}\left|\varphi^{(j-i)}\right|\\&\leqslant\widetilde{C}\sum_{i=1}^k\sup_{K}\left|\varphi^{(i)}\right|\end{align}.$$ Hence, $fT$ is a distribution. $\Box$

Remark. $\widetilde{C}$ can for example be : $$C\max_{i\in\{1,\ldots,k\}}\left(\max_{j\in\{1,\ldots,i\}}{i\choose j}\sup_{K}\left|f^{(i)}\right|\right).$$


In your case, one has: $$\forall\varphi\in\mathcal{C}_0^{\infty}(\mathbb{R}),\langle f\delta,\varphi\rangle=(f\varphi)(0)=f(0)\langle\delta,\varphi\rangle.$$ Therefore, one gets: $$f\delta=f(0)\delta.$$

$\endgroup$
  • $\begingroup$ This is the best answer I could ever think of. Thank you so much! And you are right, it was of course $f \delta$ and not $\delta f$. $\endgroup$ – Darius Dec 12 '15 at 8:41

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.