More generally, is anything available on quadratic regression, i.e., obtaining a least squares solution to a system of n equations quadratic in 3 unknowns? The sum of squares of residuals (SSE) can have either one or two local minima. With polar coordinates, I can tell if a solution is unique and find the other solution if it is not.
I can write my own, but I'd rather use more standard utilities if they exist.
Thanks,
Spencer Graves______________________________________________ [EMAIL PROTECTED] mailing list http://www.stat.math.ethz.ch/mailman/listinfo/r-help
