jeudi 21 avril 2016

Different simulation resuts in R and Matlab

Given the following decomposition

where , is constant, and are independent standard normal random variables,

the goal is to estimate the target function

where is scalar in (-4,-1).

I calculated the target function by simulating Z and Y for M=10,000, N = 10 and 100,000 simulations in R and Matlab. The results are very different: Matlab produces almost twice higher results, which makes me wonder if Matlab's random number generator is able to produce more extreme values comparing to R.

BTW, I use rnorm() in R and normrand() in Matlab to generate random numbers.




Aucun commentaire:

Enregistrer un commentaire