Optimization of two functions

  • A
  • Thread starter elere
  • Start date
  • #1
5
0
hello, in the task of finding the optimal phase covariant cloning machine, i have to maximize two functions of six variables :f1=a.C+b.D and f2=a.B+c.D , they are many constraints, but I've already used them to get to those expressions in the first place, the variables are real scalars and vary between -1 and 1. of course i'm not searching for numerical values for the six variables, in fact i've already got a transformation (set of variables) which i know is optimal, all i'm looking for is a way to prove that it is actually optimal and i know there can be others sets of variables which provide optimality.
 

Answers and Replies

  • #2
35,269
11,542
The functions are clearly maximized if all variables become 1. You can also flip the signs of a, B and C together, or the signs of b, c, and D together, or both groups together, without changing the result. Proof: looked at the expression. But you can also use some inequalities to prove that those solutions maximize both f1 and f2 in a formal way.
 
  • #3
5
0
The functions are clearly maximized if all variables become 1..
my fault, the 6 variables along with two others that i've managed to simplify in the expressions verifies a orthonormalization conditition :
a^2+b^2+c^2+d^2=1 and aA+bB+cC+dD=0 . and here are the ramaining constraints : Ac+Bd=Ab+Cd=0 and ac+AC+bd+BD=ab+AB+cd+CD=0. the set that i want to proof it's optimality and which verify the above conditions is {a=1,b=c=d=0, A=D=0}.
 
  • #4
35,269
11,542
Okay, so a=B=C=1, and everything else zero? Then f1=f2=1.

From looking at it: a=-b=C=D=1/sqrt(2) and everything else zero lead to f1=1 as well, but then f2=0. How exactly do you want to maximize the functions? The sum of both? Both individually? Something else?

There are 6 constraints on 8 variables, unless you need a mathematical proof most algorithms should quickly find all local maxima.
 
  • #5
5
0
well, B and C aren't equal to one, they are parameters which satisfy A^2+B^2=1 (first constraint). and for the way to maximize the functions, well i guess in anyway, as the two variables B and C (which can be seen as Cos and Sin) are not specified, the symmetrical case correspond to maximizing the Sum and minimizing the difference. to be specific i need to show that this set of variables with the constraint A^2+B^2=1, verify the optimality condition (df1=0, df2=0), so i was thinking that maybe i could write this conditions in a more explicit form, as a function of the variables, i've tried the la grange multipliers, but i'v got confuse with putting the second optimality condition (df2=0) as a constraint for the first one.
 
  • #7
5
0
Where does that come from now?
i'm really sorry, i've forget one costraint, the constraint is A^2+B^2+C^2+D^2=1, and since A=D=0 ...
 
  • #8
35,269
11,542
Then you have 7 constraints on 8 parameters. That is a one-dimensional problem, fixing one parameter fixes all others (up to some discrete changes like flipping signs).
 
  • #9
5
0
yes it is, i managed (finally) to write the lagrange's equations, which are satisfied for the specific set of variables i've mentioned ! thanks for your time and you help :smile:
 

Related Threads on Optimization of two functions

  • Last Post
Replies
3
Views
4K
Replies
5
Views
6K
  • Last Post
Replies
3
Views
1K
Replies
12
Views
862
Replies
5
Views
887
  • Last Post
Replies
1
Views
2K
  • Last Post
Replies
4
Views
2K
  • Last Post
Replies
2
Views
2K
  • Last Post
Replies
1
Views
807
  • Last Post
Replies
1
Views
2K
Top