Questions tagged [convolution]

Questions on the (continuous or discrete) convolution of two functions. It can also be used for questions about convolution of distributions (in the Schwartz's sense) or measures.

Convolution is a commutative, associative, distributive operation between two functions that produces a third function. It is defined in the continuous domain as

$$(x \ast y)(t)=\int_{-\infty}^\infty{x(\tau)\space{}y(t-\tau)}\space{}d\tau$$

And in the discrete domain as

$$(x \ast y)[n]=\sum_{k=-\infty}^\infty{x[k]\space{}y[n-k]}$$

Its identity is the Dirac delta function $\delta(t)$ in continuous domain, and the Kronecker delta function $\delta[n]$ in discrete domain.

Reference: Convolution.

2979 questions
1
vote
0 answers

Convolution of a product with focal kernel

Consider the following convolution of a product of two functions $f(x)$ and $g(x)$: $\int f(x')g(x')K_n(x-x') dx'$ where the kernel $K_n$ is a sequence of functions that approach a Dirac delta function as $n$ goes to $\infty$. In the limit as $n…
D_J_S
  • 151
  • 6
1
vote
3 answers

Proof of convolution inequality

I have to prove that if $f$, $g$ $\in L^1(\mathbb{R^n})$ then $\operatorname{dom}\left(f*g\right)$ is a set of full measure and: $\left\|f*g\right\|_{L^{1}} \le \left\|f\right\|_{L^1} \left\|g\right\|_{L^1}$ Any help would be appreciated.
luka5z
  • 6,359
1
vote
0 answers

LTI system and convolution

I'm reading a rather informal text on (continuous) Linear Time Invariant (LTI) systems. It is just said to be a "black box" that transforms an input signal $x(t)$ into an output signal $y(t)=(\mathcal{H}x)(t)$ which in addition satisfies $\alpha…
flavio
  • 734
  • 4
  • 12
1
vote
1 answer

$\lim_{\epsilon \rightarrow 0} || f \ast \phi_{\epsilon} - f||_{L^{\infty}(\mathbb{R}^n)}=0$ for Schwartz-function $f$

Let $\phi_{\epsilon}$ be an approximate identity, i.e. it has the following three properties: (a) $|| \phi_{\epsilon} ||_{L^1(\mathbb{R}^n)} \leq c$ for some constant $c>0$. (b) $\int_{\mathbb{R}^n} \phi_{\epsilon} dx =1$ (c) for $\delta >0$ it…
Andreas804
  • 139
  • 8
1
vote
0 answers

Proving that $\phi_{\epsilon}(x)=\frac{1}{\epsilon} \phi ( \frac{x}{\epsilon})$ is an approximate identity

An approximate identity is a function $\phi_{\epsilon} \in L^1(\mathbb{R}^n)$ with the following properties: (a) $\| \phi_{\epsilon} \|_{L^1(\mathbb{R}^n)} \leq c$ for some constant $c>0$. (b) $\int_{\mathbb{R}^n} \phi_{\epsilon} dx =1$. (c) for…
Andreas804
  • 139
  • 8
1
vote
0 answers

Closed-form expression of unit square pulse convolved with itself N times

If the unit square pulse $u_1(t)$ is defined as $$u_0(t) = \begin{cases} 1 & \mathrm{if}\, 0 \le t \le 1 \\ 0 & \mathrm{otherwise} \end{cases}$$ and the function $u_n(t)$ is defined as $$u_n(t) = (u_{n-1} \ast u_0)(t)$$ for integer $n > 0$, is there…
Jason S
  • 3,109
1
vote
1 answer

Is the following convolution property true?

Is the following convolution property true? $$\text{If} \ y(t)=x(t)*h(t) , \text{then} \ y(t)=\int_{-\infty}^{t} [x'(\tau)*h(\tau)] \,d\tau $$ My proof : Let's denote by $g$ the integrand $g(\tau)=x'(\tau)*h(\tau)$, then $ y(t)=\int_{-\infty}^{t}…
user1115701
1
vote
1 answer

convolution, $\frac{1}{p}+\frac{1}{q}\neq 1+\frac{1}{r}$

I need help/a hint for the following task. Let $p,q,r\in [1,\infty)$ with $\frac{1}{p}+\frac{1}{q}\neq 1+\frac{1}{r}$ and $C>0$. Show that there are $f\in L^p(\mathbb R^d), g\in L^q(\mathbb R^d)$ with $\|f*g\|_r>C\|f\|_p\|g\|_q$. ($f*g$ is the…
marc
  • 312
1
vote
1 answer

Calculating the convolution of $\theta(t)\cdot(e^t\theta(1-t))$

I try to calculate the convolution of $\theta(t)\cdot(e^t\theta(1-t))$. Using the formula \begin{equation} f*g(t)=\int_{-\infty}^\infty f(t-u)g(u)du \end{equation} I set $f(t-u)=\theta(t-u)$ and $g(u)=e^t\theta(1-t)$ So the integral would…
Luthier415Hz
  • 2,739
  • 6
  • 22
1
vote
0 answers

How can convolution be interpreted as a recognizer?

In many textbooks it is said, that convolution can be interpreted as a pattern recognizer and that if kernel is located in region similar to it, then it gives greater response, than when it locates in different region. But consider 1D case. Suppose…
1
vote
0 answers

How to convert function to kernel?

Imagine I have a signal function, several examples: PSF Wavelet. How can I transform its logic into a kernel? By kernel I mean matrix which is used in the convolution afterward.
Nourless
  • 111
1
vote
1 answer

Is it true that the convolution of two locally integrable functions is always defined?

As I understand it, the convolution $$(f \ast g)(x) = \int_{\mathbb{R}^n} f(y) g(x-y)\,\mathrm dy$$ is defined wherever the above integral is defined, i.e., finite. This is true whether $f$ and $g$ are functions, test functions and distributions,…
1
vote
1 answer

What would a 3-way convolution be?

Convolution is defined as $$ (f*g)(t) = \int^\infty_{-\infty} f(\tau) g(t-\tau)d\tau $$ If I want to do a 3-way convolution like $$ 3way(f,g,h)(t) = \int^\infty_{-\infty}\int^\infty_{-\infty} f(\tau) g(t-\tau)h(t-\upsilon)d\tau d\upsilon $$ What is…
azman
  • 67
1
vote
1 answer

Convolution with shifted arguments

Given two function $f(t)$ and $g(t)$ and denote the convultion of the functions by $h(t)$, that is $$f(t)*g(t)=h(t).$$ What is $f(t-1)*g(t+1)$ in terms of $h(t)$? Have no idea how to treat this question, everywhere I read, it is said, that this…
Ilya.K.
  • 1,288
1
vote
0 answers

Convolution of 2 signals, equivalences

If I have two signals $x(t)$ and $y(t)$ and define $z(t)=x(t)∗y(t)$. Can we say $x(t)∗y(-t)$ is equal to any of the following? $z(−t), −z(t), −z(−t), z(t)$