0

Suppose i have a discrete random variable A such that:

  • $p(A=-1) = 3/4$
  • $p(A=0) = 1/8$
  • $p(A=1) = 1/8$

Now, i create a random variable $B = |A|$ and so

  • $p(B=0)= 1/8$
  • $p(B=1)= 7/8$

I want to compute $f_{A,B}(a,b)$ [generalized joint probability density function, using delta dirac function since it is a discrete case].

Are those variables independent so can I do $f_{A,B}(a,b)=f_A(a)\cdot f_B(B)$? Or, if they are dependent, when calculating $P(a_i,b_n)$ should i do $P(a_i,b_n) = P(A=a_i)\cdot P(B=b_n\mid A=-1)$ just like in Bayes Theorem ?

Thanks in advance.

Stefan Hansen
  • 25,582
  • 7
  • 59
  • 91
user1843665
  • 163
  • 7
  • This distribution is a sum of Dirac distributions at {-1,1}, {0,0} and {1,1}. Surely you can guess the weight at each point. – Did Dec 07 '12 at 06:48
  • Didn't get it .. Are the random variables A and B independent ? – user1843665 Dec 07 '12 at 06:51
  • Do you know the definition of some random variables being independent? 2. Can you compute P(A=1), P(B=0) and P(A=1,B=0)? 3. Ergo?
    • – Did Dec 07 '12 at 06:55
  • Yes, even tho what i'm interested in is the generalized joint probability density function.You said the sum of {-1,1}, {0,0} and {1,1} ( with proprer weigths and delta function ) equals fA,B(a,b) ? – user1843665 Dec 07 '12 at 06:56
  • the generalized joint probability density function... Meaning? // You said the sum of {-1,1}, {0,0} and {1,1}... Nope, I did not say that (and I do not know what it is that you call the sum of {-1,1}, {0,0} and {1,1}). – Did Dec 07 '12 at 14:52