Let $X_1$ and $X_2$ be two i.i.d. continuous random variables which have probability density function $$f(x) = \frac{1}{\pi}\frac{1}{1+x^2}.$$ Now I want to calculate the density of $X_1 + X_2$.
$$
f_{X_1+X_2}(z)
= \int_{-\infty}^{\infty} f_{X_1}(x_1)f_{X_2}(z-x_1)dx_1
= \int_{-\infty}^{\infty} \frac{1}{\pi} \frac{1}{1+x_1^2}
\frac{1}{\pi} \frac{1}{1+(z-x_1)^2} dx_1
$$
Is there any efficient way to calculate this integral?
Asked
Active
Viewed 330 times
0
Jonny
- 433
-
1Does this integral exist? What happens when $x_1 = z$? – Ryan Goulden Mar 29 '19 at 01:35
-
@RyanGoulden this one does not but the correct one does – gt6989b Mar 29 '19 at 01:56
2 Answers
1
You made a mistake, the correct integral is $$ \frac{1}{\pi^2} \int_\mathbb{R} \frac{1}{1+x^2} \frac{1}{1 + (z-x)^2} dx $$ and I would look around trig substitution or partial fractions...
gt6989b
- 54,930
0
You could try partial fractions. An integral of (1+(x1)^2)^-1 on its own is fine to work out, it is an arctan integral. Likewise, an integral of (1+(z-x1)^2)^-1 is also very doable.
fGDu94
- 4,072