如何通过MATLAB代码在R中创建协方差矩阵？

Question

I need some help with porting MATLAB code to R.

Below is MATLAB code for creating the covariance matrix, where Z is a matrix of 17 columns and 27 rows. ehat is a vector of 27 rows and 1 column.

omegahat=zeros(cols(Z),cols(Z));

for i=1:N    
    omegahat=omegahat+ehat(i)^2*Z(i,:)'*Z(i,:);   
end

How do I replicate a for loop in R that works similarly to this MATLAB code?

Answer 1

Try this code. This essentially mirrors what the MATLAB code is doing, but I do question its efficiency. Also cols is not a function that is defined in MATLAB except for SQL tables. I'm assuming you want to determine the total number of columns? Use size(Z,2) for that instead.

In any case, for R, t denotes transpose, and to perform matrix multiplication, you must encapsulate the * operator in % signs. As such:

omegahat <- matrix(0,ncol(Z),ncol(Z));
for (i in 1:N) { # N is 27 I suppose?
   omegahat <- omegahat + (ehat[i]^2 * Z[i,] %*% t(Z[i,]))
}

Here's a small test on a smaller matrix. Supposing Z is:

Z =

 1     2     3     4
 5     6     7     8
 9    10    11    12
13    14    15    16

Also, let ehat be:

ehat = 

1   2   3   4

I'm also going to assume N = 4 .

---

In MATLAB, this is the code to generate our output:

Z = vec2mat(1:16, 4);
ehat = 1:4;
N = 4;

omegahat = zeros(size(Z,2), size(Z,2)); %# Note the change
for i = 1 : N    
    omegahat = omegahat + ehat(i)^2*Z(i,:)'*Z(i,:);   
end

We thus get:

omegahat =

    3534        3844        4154        4464
    3844        4184        4524        4864
    4154        4524        4894        5264
    4464        4864        5264        5664

---

In R, this is the code to generate our output:

Z <- t(matrix(1:16,4,4))
ehat <- c(1,2,3,4)
N <- 4

omegahat <- matrix(0,ncol(Z),ncol(Z))
for (i in 1:N) {
   omegahat <- omegahat + (ehat[i]^2 * Z[i,] %*% t(Z[i,]))
}

We thus get:

> omegahat

     [,1] [,2] [,3] [,4]
[1,] 3534 3844 4154 4464
[2,] 3844 4184 4524 4864
[3,] 4154 4524 4894 5264
[4,] 4464 4864 5264 5664