Presumably you are using a period for multiplication and you mean this to be
a pdf, but you are missing a scale factor to make the integral equal to 1.
It is also unclear what your data is. Do you intend to estimate the lambdas
from a single x and y (no unique solution), or a sample of n independent
bivariate observations?

Jerry

" wrote:

Hi,

I have a distribution
F(x,y)= exp{-lambda1.x-lambda2.y-lambda3.max(x,y)}, x,y0 and all the
lambdas are also positive. i want to find the maximum likelihood
estimates these lambdas.
Could you point to any sample program code that would help me to do
this work? i would like to do it in SAS or "R".

Best regards.

Eben.