[R] R code for var-cov matrix given variances and correlatio

[(Ted Harding)]

On 21-Dec-04 Haynes, Maurice (NIH/NICHD) wrote:
> Dear list members,
> 
> Where can I find code for computing the p*p variance-covariance 
> matrix given a vector of p variances (ordered varA, varB, ..., 
> varp) and a vector of all possible correlations (ordered corAB, 
> corAC, ..., corp-1,p)?
> [...]

example:

> X<-cbind(rnorm(10),rnorm(10))
> cov(X)
           [,1]       [,2]
[1,] 0.90141614 0.09364018
[2,] 0.09364018 0.88476825

> cor(X)
          [,1]      [,2]
[1,] 1.0000000 0.1048540
[2,] 0.1048540 1.0000000

> vars<-apply(X,2,var)
> vars
[1] 0.9014161 0.8847682

> diag(sqrt(vars))%*%cor(X)%*%diag(sqrt(vars))
           [,1]       [,2]
[1,] 0.90141614 0.09364018
[2,] 0.09364018 0.88476825

Ted.


--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at nessie.mcc.ac.uk>
Fax-to-email: +44 (0)870 094 0861  [NB: New number!]
Date: 21-Dec-04                                       Time: 13:54:45
------------------------------ XFMail ------------------------------