Finding the CDF from a PDF with absolute value function

Click For Summary

Homework Help Overview

The discussion revolves around finding the cumulative distribution function (CDF) from a given probability density function (PDF) defined as f(x) = |x/4| for -2 < x < 2, and 0 otherwise. Participants are exploring the implications of the absolute value function on the CDF's properties.

Discussion Character

  • Exploratory, Assumption checking, Problem interpretation

Approaches and Questions Raised

  • Participants discuss the need to split the integration into cases due to the nature of the absolute value function. There are attempts to set up the CDF using integrals, with some questioning the correctness of limits and the monotonicity of the resulting function. Others suggest considering the areas under the PDF graph for different intervals.

Discussion Status

The discussion is ongoing, with various interpretations being explored. Some participants have offered guidance on how to approach the integration and the importance of ensuring the CDF is monotonic. There is recognition of the need to handle different cases separately, but no consensus has been reached on the final form of the CDF.

Contextual Notes

Participants note the requirement for the CDF to be non-decreasing and to satisfy specific values at certain points, such as F(0) = 0.5 and F(2) = 1. There are also mentions of the need to use the inversion method for generating observations from the PDF.

TeenieBopper
Messages
27
Reaction score
0

Homework Statement


Find the CDF of f(x) = <br /> |\frac{x}{4}| if -2&lt;x&lt;2 \\<br /> 0 otherwise<br />

Homework Equations


The Attempt at a Solution



I have to integrate the pdf and to do so, I have to split it into two parts

\int_{-x}^{0}\frac{-t}{4}dt + \int_{0}{x}\frac{t}{4}dt

integrating I get \frac{x^2}{8} + \frac{x^2}{8} = \frac{x^2}{4}

This isn't a strictly increasing function, which is a requirement to be a CDF. So I need to break it into cases. This is where I'm running into trouble (sorry, I don't know if/how to use cases environment here).
F(x)
<br /> 0 if x &lt; -2 \\<br /> 1-\frac{x^2}{4} if -2&lt;x&lt;0 \\<br /> some equation if 0&lt;x&lt;2 \\<br /> 1 if x&gt;2<br />

I know that the CDF must have a value of .5 if X=0, but I'm not sure how to set up the rest of the cases so that F(0)=.5 and F(2)=1

Am I allowed to keep the two integrals from above "split"? Because if I am, then the following should work:

F(x)=
<br /> 0 x&lt;-2 \\<br /> \frac{1}{2} -\frac{x^2}{8} if -2&lt;x&lt;0 \\<br /> \frac{1}{2} + \frac{x^2}{8} if 0&lt;x&lt;2 \\<br /> 1 if x &gt; 2<br />

Am I just overthinking it, or can I just use that for the CDF?
 
Last edited:
Physics news on Phys.org
Your limits are wrong. You want
$$F(x) = \int_{-\infty}^x f(x)\,dx = P(X \le x)$$
 
The CDF F(x) is by definition the integral of the PDF from -∞ to x. So I'm not sure what the question is asking for.

Edit: I see you've changed the problem statement. It makes more sense now.
 
Last edited:
I changed the pdf so it should be technically correct now

vela said:
Your limits are wrong. You want
$$F(x) = \int_{-\infty}^x f(x)\,dx = P(X \le x)$$

So, because it's an absolute value function, it should be:

$$\int_{-2}^{x}\frac{-x}{4} dt + \int_{0}^{x}\frac{x}{4}dt$$? Because then I get $$F(x) = \frac{x^2}{4} - \frac{1}{2}$$ which again, isn't monotone increasing.

Ultimately, my I need to use the inversion method to generate observations from f(x), but in order to do that, I need the CDF.
 
No, that's not correct. You need to look at different cases. It may help to sketch the PDF and the region of integration for the various cases.
 
TeenieBopper said:
I changed the pdf so it should be technically correct now



So, because it's an absolute value function, it should be:

$$\int_{-2}^{x}\frac{-\color{red}{x}}{4} dt + \int_{0}^{x}\frac{\color{red}{x}}{4}dt$$? Because then I get $$F(x) = \frac{x^2}{4} - \frac{1}{2}$$

Those ##x##'s should be ##t##'s, and those two integrals, separately, will be part of your work. By as vela points out, you aren't setting up the problem correctly. Consider the cases ##-2<x<0## and ##0<x<2## separately when considering the integral ##\int_{-\infty}^xf(t)~dt##.
 
It's an absolute value function, so if I graph it, it's essentially two triangles, each with a point at 0,0 and then vertical edges at x=-2 and x=2. So there's two cases, -2<x<0 and 0<x<2, that I look at separately.

Looking at the first case, the integral is

$$\int_{-2}^{x} \frac{-t}{4} dt = \frac{t^2}{8} |_{-2}^x = \frac{x^2}{8} - \frac{1}{2}$$

But this can't be part of a CDF because it's a decreasing function on (-2, 0), which is why I thought the function for the CDF in this case was $$\frac{1}{2} - \frac{x^2}{8}$$Looking at the case 0<x<2, the integral is

$$\int_{0}^{x} \frac{t}{4}dt = \frac{t^2}{8} |_{0}^{x} = \frac{x^2}{8}$$

Now, this is increasing on (0, 2), but the value at 2 is only 1/2, which is why I thought I could add 1/2 to this case of the CDF (which is the area under the curve from the first case), making it $$\frac{1}{2} + \frac{x^2}{8}$$

This would make the complete cdf

F(x)=
0 if x<-2
$$\frac{1}{2} -\frac{x^2}{8}$$ if -2<x<0
$$\frac{1}{2} + \frac{x^2}{8}$$ if 0<x<2
1 if x > 2

I don't understand why this isn't it.
 
What you wrote down is the CDF, you just made a couple mistakes. In calculating the CDF for the -2<x<0 part, you are integrating negative t/4, and the negative sign simply disappeared. When doing the 0<x<2 part, remember what you are calculating is
\int_{-2}^{x} f(t) dt = \int_{-2}^{0} f(t) dt + \int_{0}^{x} f(t) dt.

You know from your previous calculation that
\int_{-2}^{0} f(t) dt = 1/2,
and this is exactly the 1/2 that you are missing.
 
TeenieBopper said:
I changed the pdf so it should be technically correct now



So, because it's an absolute value function, it should be:

$$\int_{-2}^{x}\frac{-x}{4} dt + \int_{0}^{x}\frac{x}{4}dt$$? Because then I get $$F(x) = \frac{x^2}{4} - \frac{1}{2}$$ which again, isn't monotone increasing.

Ultimately, my I need to use the inversion method to generate observations from f(x), but in order to do that, I need the CDF.

If you plot f(t) and recognize that F(x) = area under the f(t)-curve to the left of t=x, you can see immediately that F(x) = 0 if x < -2 and F(x) = 1 if x > +2. For -2 < x < 0, the area to the left of x is the whole area from t=-2 to t=0, minus the area from t = x to t = 0. So, if you know how to compute areas of triangles you are home free. For 0 < x < 2 you have F(x) = F(0) + area between t=0 and t=x, and that is easy, too.
 

Similar threads

  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 3 ·
Replies
3
Views
1K
  • · Replies 105 ·
4
Replies
105
Views
11K
Replies
4
Views
2K
  • · Replies 3 ·
Replies
3
Views
5K
  • · Replies 11 ·
Replies
11
Views
3K
Replies
4
Views
3K
Replies
5
Views
3K
  • · Replies 14 ·
Replies
14
Views
2K
Replies
6
Views
3K