a,b,c >0 and a+b+c=1 ;
   P=(1/a)+(2/b)+(3/c);

I think I have some idea of what you're talking about. However, it doesn't seem like a good general-purpose extremum finder. It seems like it only works if the lower bound is equal to the actual minimum, which doesn't seem like a safe general assumption.

Posted by friedlinguini on 2002-06-21 15:17:48