# WolframAlpha: PDF(X+Y) where X and Y are iid Poisson(lambda)

Posted 9 years ago
6365 Views
|
2 Replies
|
3 Total Likes
|
 I'm trying to play around with two random variables and I can't get wolfram alpha to parse my input.I've tried something really simple that like PDF(X+Y) where X and Y are iid Poisson(lambda) trying to get wolfram alpha to tell me that this is Poisson(2*lambda) or even just give me the joint cdf or pdf.all I can get out is Using closest Wolfram|Alpha interpretation: Poisson(lambda) I really want the distribution of X/(X+Y) where X and Y are iid Poisson(lambda) but I figured I would start small.any ideas what I could try to help it interpret my input?
2 Replies
Sort By:
Posted 9 years ago
 thanks seth.I've tried a few other ways too.  Unfortunately this has also been timing out on me.  (maybe if i restrict the domain to be x > 0 and y > 0 it would work)P[ X/(X+Y) < f ]  --> exp(-2*lambda) * sum_{y=0}^{+inf} sum_{x=0}^{floor[ f/(1-f) * y ]} lambda^(x+y)/(x!*y!)
Posted 9 years ago
 So, I have bad news.  Even if you could develop a syntax that would persuade WolframAlpha to try to compute the PDF of distribution of X/(X+Y) where X and Y are iid Poisson(lambda) , WA is going to time out on the computation.  I just tried running it on actual Mathematica for a few minutes on a reasonably powerful machine and got nowhere.  Also, the domain of the Poisson distribution includes 0.  So, don't we have a division by zero problem in your transformation? (which might be why Mathematica is unhappy when it tries to compute the PDF).