Search code examples
matlabvectorizationcomputation-theory

Fastest Way to Generate x Number of Random Variables

index = [10 100 1000 10000 100000]

Let's say I wanted to generate 10, 100,...,100000 (corresponding to index above) exponential random variables with parameter 1/10. I of course did this with a for loop and with the command for example 

x = exprnd(0.1,100,1)

One can also generate the exponential random variables from a uniform random variable with 

x = -10*log(u)

In this method, I first generate the uniform random variables then plugged the vector, u, into the previous equation. This method is faster in terms of computation time.

How might I speed this process up? Could I do this with vectorization, that is, without for loops?

Solution

  • Try this:

    expFun = @(size)exprnd(0.1,size,1);

index = [10 20 30 40];
data = cellfun(expFun,mat2cell(index,ones(size(index,1),1),ones(1,size(index,2))),'UniformOutput',false);