How to get and plot correlation coefficient matrix as a one dimentional function of the random vectors differences

Hi, I have a correlation coefficient matrix for two random vectors (size 94x94).

The matrix enterences are numbered as angels from -\pi to \pi. i.e. the enterences of the 94x1 vector of variables are angels.

Each random variable is identified by it's preffered direction. I will refer it as it's name.

I want, to creat a function, where the x axis is the angel differences of the variables, and the y axis is the (maximal) relevant correlation. i.e. the correlation between any two variables as a function of the difference between their names (preffered directions).

for example, if I had:

corr = [1,2;2,3]

where the name of v1 is \pi and the name of v2 is 0.5*\pi,

then I would like my function to be:

f(0) = max(corr(\pi,\pi),corr(0.5*\pi,0.5*\pi)) = 3
f(0.5*\pi) = corr(\pi,0.5*\pi) = 2

Is there a matlab function that does that?

Basically I need the linear regression of the function that I have described.

I thonk that what I am trying to do is a linear extrapolation of the correlation coefficient function but I am not sure..

Thanks!

Basically, I think that what I am trying to do is,

Thanks!