What is the expected value for Y in the given limiter?

  • Thread starter magnifik
  • Start date
3 replies · 2K views · last reply
magnifik
Messages
350
Reaction score
0
For the limiter shown below, find the expected value for Y = g(X)
Graph of a piecewise linear limiter g(x): horizontal at -a for x below -a, rising linearly through the origin between -a and a, and horizontal at a for x above a, with axis labels -a, 0, a and x.


attempt at solution:
E[Y] = ∫g(x)f(x)dx, where f(x) is the probability density function with respect to x
so...
E[Y] = E[Y1] + E[Y2] + E[Y3]
E[Y1] = ∫-af(x)dx where the limits of integration are from -∞ to -a
so E[Y1] = -aFx(-a), where Fx is the cumulative distribution function
E[Y2] = ∫xf(x)dx where the limits of integration are -a to a
this is the part I'm having trouble with
E[Y3] = ∫af(x)dx where limits of integration are a to ∞
so E[Y3] = a(1-Fx(a))

i'm having trouble simplifying E[Y2] into one expression because of the limits of integration. i know it would be just E[Y2] if the limits were -∞ to ∞, but that is not the case here
 
I think that when the domain of integration is limited over a certain set A, the expected value becomes the conditional expected value of X given that X lies on A, times the probability that X lies on A.
 
magnifik said:
For the limiter shown below, find the expected value for Y = g(X)
Graph of g(x) against x: a piecewise-linear limiter that is flat at -a for x below -a, rises linearly through the origin between -a and a, and is flat at a for x above a.


attempt at solution:
E[Y] = ∫g(x)f(x)dx, where f(x) is the probability density function with respect to x
so...
E[Y] = E[Y1] + E[Y2] + E[Y3]
E[Y1] = ∫-af(x)dx where the limits of integration are from -∞ to -a
so E[Y1] = -aFx(-a), where Fx is the cumulative distribution function
E[Y2] = ∫xf(x)dx where the limits of integration are -a to a
this is the part I'm having trouble with
E[Y3] = ∫af(x)dx where limits of integration are a to ∞
so E[Y3] = a(1-Fx(a))

i'm having trouble simplifying E[Y2] into one expression because of the limits of integration. i know it would be just E[Y2] if the limits were -∞ to ∞, but that is not the case here

Unless you have a specific form for f(x) you have gone as far as you can: E(Y2) does not simplify further in any useful way. By the way: I would argue against your notation E(Y1), E(Y2), etc. You do not have three separate random variables Y1, Y2 and Y3; you just have three separate "pieces" of the same random variable Y.

RGV