"Kirk" <kwythers.nospam@umn.edu> wrote in message <gmv9tq$c4e$1@fred.mathworks.com>...
> I am using 'mvnrnd' to produce random monthly estimates for some climate data (TMAX TMIN PAR and PRECIP), these data are based on 100 years of measured means and a calculated covariance matrix.
>
> How would manipulating the means effect the correlations? I think if I were to reduce all means by 25%, then the correlations among the variables would remain consistent. However, what I would really like to do is to reduce only one of the 4 variables (PRECIP) by 25%. As an example, I could break out each column in the matrix of means individually. Columns are in the order of tmax, tmin, par prec.
>
> mu=cat(2,climate_mu(:,1),climate_mu(:,2),climate_mu(:,3),climate_mu(:,4));
>
> then pass use 'mu' as an argument in 'mvnrnd' along with the original covariance matrix 'sigma'.
>
> Therefore, I could multiply climate_mu(:,4) by 0.25, and let 'cat' reassemble the matrix. In this way all the precip means are being reduced, but temps and solar radiation is not. The question becomes... Have I just broken my correlation assumptions by manipulating only one of the variables? or, is it that since the correlations all defined in the covariance matrix, that what ever I do to the means is of no consequence to the correlations among TMAX TMIN PAR and PREC?
Changing one or any of them will have no
impact on the correlation, at least if all you
do is to add a constant to a variable, or
multiply it by any scalar variable.
John
