anyone care to explain Importance Sampling?

[slavatrumpevitch]

I've got my finals coming up, and I still can't quite get my head wrapped around importance sampling. http://en.wikipedia.org/wiki/Importance_sampling

 

anyone familiar with this, that can help me out with this?

 

the problem is to do an importance sampling estimate of I=E(log X + 1)

 

given that X~rayleigh distribution, and using an exponetial distribution as our importance sampling approximator.

 

any thoughts?

 

thanks!

[Serpentine]

Saw stats... ran the hell away.

 

gl tho

[STiFU]

I never heard of it, so I was a little curious what this is about, but I don't quite get it and furthermore the german wikipedia article somehow differs quite a bit from the English one. Here is the google translate version of it, maybe that helps you a little.

[stumpy]

I think its on the lines of you have a random number that generates a known adverage number, but the number you need from another random number doesn't currently have a solution, you can then use the known adverage to get somewhere near the number you need from the unknown random, and the more known random adverages you use means you can get closer to a result for the unknown.

 

Although I could be wrong.