I know the policy of MO, probably it is my fault to not have explained the source, and it really comes from a research question. on the other hand, I just wanted to know if the tecniquenes to solve it were standard, at first sight I d say a method with Lagrangian multipliers might be enough, but I didnt know a better approach..

Thanks Josè, but I am not interested in a software solution, rather in methods how to solve it manually.. [Anyway, we expected that solution geometrically :P ]

I am sorry Anthony, I will take it into account next time.. I added the links to both threads, let me know if I forgot something..

Not really, just a curiosity which I would like to solve (to be precise, I asked myself why a variant of deFinetti representation theorem was meaningful); I posted this morning the same problem on MSE, but it didn't receive much attention..