How do you make a random vector of only real values, using two other random vectors?

The title may be a bit confusing but I've attached the image of the problems i'm currently working on.

My Code

>> x1 = randn(L,1)*sigma + mu;
>> x2 = randn(L,1)*sigma + mu; 
>> w1 = (sqrt(-2*log(x1))).*(cos(2*pi*x2));
>> w2 = (sqrt(-2*log(x1))).*(sin(2*pi*x2));
>> histogram(w1, 'normalization', 'pdf');
% Error using histogram
% Expected input number 1, x, to be real.

Is there anyway to only take the real portion of w1 and w2? Also am I going about the problem in the right way?

[Bruno Luong]

The problem states x1 and x2 are uniform distribution, why you are using Gaussian distribution and get complex from negative numbers?

[Kaleem Graham]

I see what you're saying now, instead I should make x1 and x2 uniform distribution.

>> pd1 = makedist('Uniform');
>> x = 0:.01:1;
>> x1 = pdf(pd1,x);

Like this?

[Bruno Luong]

No. Use RAND instead of RANDN

[Kaleem Graham]

Thanks it worked! Was I on the right in regards to actually solving the problem? The mean of both w1 and w2 are about 0 and the std is about 1 for both of them.