Piddling Pertinent
  • Home
  • Categories
  • Tags
  • Archives

Convolution Demysitifed

Problem¶

Given \(f(x) = \frac{x}{2}\) for \(0 \leq x \leq 2\). Find the pdf of \(x_1+x_2\) for \(x_1,x_2\) which are i.i.d.

Wrong solution¶

$$ f*g(t) = \int_{-\infty}^{\infty} f(w)g(t-w)dw $$

Thus, blindly,

\(f_S(s) = \int_{0}^s \frac{x(s-x)}{4}dx = \frac{s^3}{24}\) and \(0 \leq s \leq 4\)

But is \(f_S(s)\) a PDF?

No.

\(\int_0^4 f_S(s)ds = \frac{4^4}{96} \neq 1\)

Respecting the bounds.¶

\(f_S(s) = \int_{0}^s \frac{x}{2} \frac{s-x}{2}dx\) for :

$$ 0 \leq x \leq 2 $$

and

$$ 0 \leq s-x \leq 2 $$

Which gives us, these bounds:

\(s-2 \leq x \leq s\) and \(0 \leq x \leq 2\)

In order to respect the above bounds,

\(f*g(s) = \int_{-\infty}^{\infty}f(x)f(s-x)dx\)

But, \(f(x) \neq 0\) for \(0 \leq x \leq 2\)

and \(f(s-x) \neq 0\) for \(s-2 \leq x \leq s\)

thus, \(f(x)f(s-x) \neq 0\) for \(max(0,s-2) \leq x \leq min(s,2)\)

Range of s: \([0,4]\)

\(f*g(s) = \int_{max(0,s-2)}^{min(s,2)}f(x)f(s-x)dx\)

For \(0 \leq s \leq 2\) : \(min(s,2)=s\) and \(max(0,s-2)=0\)

For \(2 \leq s \leq 4\) : \(min(s,2)=2\) and \(max(0,s-2)=s-2\)

Now,

$$ f_S(s) = \begin{cases} \int_{0}^{s}\frac{x}{2}\frac{s-x}{2}dx = \frac{s^3}{24} & 0 \leq s \leq 2\\ \int_{s-2}^{2}\frac{x}{2}\frac{s-x}{2}dx = \frac{1}{4}(\frac{s(2^2-(s-2)^2)}{2}-\frac{2^3-(s-2)^3}{3}) & 2 \leq s \leq 4\\ \end{cases} $$

Share on: Twitter ❄ Facebook ❄ Google+ ❄ Email

Comments
comments powered by Disqus

Related Posts

  • Proof for triangle inequality for case $x+y<0$
  • Gumbel distribution expectation
  • SVD v/s MDS v/s PCA
  • Distribution of one random variable less than other
  • A frequent inequality

  • « MATH 501 Project
  • Restriction Fragments Size Distribution »

Published

May 9, 2015

Category

misc

Tags

  • math 4
  • pdf 2
  • stats 4

Contact

  • Piddling Pertinent - trivially relevant
  • Powered by Pelican. Theme: Elegant