Stats/prob: finding cumulative distribution function

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
3 replies · 2K views
Phox
Messages
36
Reaction score
0

Homework Statement



given pdf:

f(x) = 2/3x for 0<=x<=1
f(x) = 2/3 for 1<x<=2
f(x) = 0 elsewhere

Find the CDF.


Homework Equations





The Attempt at a Solution



I've found:

F(x) = 0 for x<= 0
F(x) = 1 for x>=2
F(x) = (1/3)x2 for 0<=x<=1

and I found:

F(x) = (2/3)x for 1<x<=2

However, the last bit is incorrect. It should be F(x) = (2/3)x -(1/3)

I'm unclear as to why. I think it has something to do with solving for the constant of integration, but I'm not sure exactly.
 
Physics news on Phys.org
Yes, it has to do with the integration constant. Your CDF has to be continuous, so you need to fix that constant value so that

$$\lim_{\epsilon \rightarrow 0} F(1 - \epsilon) = \lim_{\epsilon \rightarrow 0} F(1 + \epsilon).$$

That is, the limit of the CDF on either side of x = 1 have to be the same, which wouldn't be true if you didn't fix the constant of integration to be -1/3 just above x=1.
 
Mute said:
Yes, it has to do with the integration constant. Your CDF has to be continuous, so you need to fix that constant value so that

$$\lim_{\epsilon \rightarrow 0} F(1 - \epsilon) = \lim_{\epsilon \rightarrow 0} F(1 + \epsilon).$$

That is, the limit of the CDF on either side of x = 1 have to be the same, which wouldn't be true if you didn't fix the constant of integration to be -1/3 just above x=1.

Could you clarify.. what is epsilon?
 
Phox said:

Homework Statement



given pdf:

f(x) = 2/3x for 0<=x<=1
f(x) = 2/3 for 1<x<=2
f(x) = 0 elsewhere

Find the CDF.


Homework Equations





The Attempt at a Solution



I've found:

F(x) = 0 for x<= 0
F(x) = 1 for x>=2
F(x) = (1/3)x2 for 0<=x<=1

and I found:

F(x) = (2/3)x for 1<x<=2

However, the last bit is incorrect. It should be F(x) = (2/3)x -(1/3)

I'm unclear as to why. I think it has something to do with solving for the constant of integration, but I'm not sure exactly.

You have F correct for x ≤ 1 and for x ≥ 2. Since the random variable has a finite density function, its F(x) must be a continuous function, and since F'(x) = 2/3 on [1,2], F must increase linearly with slope 2/3, starting from F(1) = 1/3 and ending at F(2) = 1. You can figure out what the formula must be for F(x) in the region 1 ≤ x ≤ 2.

Basically, you need to use
[tex]F(x) = F(1) + \int_{1}^{x} f(t) \, dt, \: 1 \leq x \leq 2.[/tex]
Note that in this calculation there is NO constant of integration!