x 2 </msup> y &#x2192;<!-- → --> max, such that x 2 </

excluderho 2022-06-26 Answered
x 2 y max,
such that x 2 + 4 x y 1 , x 0 and y 0.

I think I need to use the KKT conditions here. I did however not yet succeed in solving it, so could someone please give me an example of how this should be done? And should I include the constraints x 0 and y 0 into the Lagrangian function?
You can still ask an expert for help

Expert Community at Your Service

  • Live experts 24/7
  • Questions are typically answered in as fast as 30 minutes
  • Personalized clear answers
Learn more

Solve your problem for the price of one coffee

  • Available 24/7
  • Math expert for every subject
  • Pay only if we can solve it
Ask Question

Answers (1)

sleuteleni7
Answered 2022-06-27 Author has 28 answers
The KKT conditions in this case for x , y that are feasible x 0 , y 0 , x 2 + 4 x y 1 is that α , β , γ 0 (feasible dual variables for the constraints in the last parens in respective order) such that

1. Stationary: 2 x y = α + γ ( 2 x + 4 y ), x 2 = β + γ x 2 , and
2. Complementary slackness: α x = β y = γ ( x 2 + 4 x y 1 ) = 0

However, the problem is not convex therefore KKT conditions are not sufficient. In particular, the eigenvalues of the hessian of your objective are y ± 4 x 2 + y 2 so that one of them is always positive (so not concave for a maximization problem) and also one is negative whenever x > 0 which is certainly feasible (so it's not even definite in either direction).

But they are necessary. So you can guess which constraints are active or not and go through all permutations of this to find all KKT-satisfying points. For example, guess that the nonnegativity constraints are slack ( α = β = 0) and that the other constraint is tight ( x 2 + 4 x y = 1). Then plugging this into the KKT conditions we get three equations:
2 x y = γ ( 2 x + 4 y )
( 1 γ ) x 2 = 0
x 2 + 4 x y = 1
Which solve for x = 1 / 3 and y = 1 / 12 and give objective value 1 / 108 . (Hint: this is optimal.)
Not exactly what you’re looking for?
Ask My Question

Expert Community at Your Service

  • Live experts 24/7
  • Questions are typically answered in as fast as 30 minutes
  • Personalized clear answers
Learn more

You might be interested in

asked 2022-06-24
Define the function
f ( a , b , c , α , β , γ , x ) = max ( 0 , max ( ( a + x ) α , ( b + x ) β ) ( c + x ) γ ) ,
where
a , b , c , α , β , γ , x [ 0 , M ] .
Is it true that for any
ξ = ( a , b , c , α , β , γ ) ,
the maximum of f ( ξ , ) occurs either when x = 0 or x = M?

I think the answer is yes, but I have trouble prooving it.
My argument is as follows:
Given any ξ, f ( ξ , c ) will be
f ( ξ , x ) = { 0 case A ( a + x ) α ( c + x ) γ case B ( b + x ) β ( c + x ) γ case C
Hence, f ( ξ , x ) is a linear function of x in all three cases, and the result follows.

I think this argument works only if each case is independent of x, but this is not the case.

As, when ξ is given, judiciously choosing x may put f in another case.

What would be a right way to prove this result?
asked 2022-06-26
I am given the following problem:
A = [ 4 1 1 1 2 3 1 3 2 ]
Find
max x | ( A x , x ) | ( x , x )
where ( . , . ) is a dot product of vectors and the maximization is performed over all x = [ x 1 x 2 x 3 ] T R 3 , such that i = 1 3 x i = 0
I have found the eigenvectors for A and they happen to match the sum criterion:
E ( λ 1 ) = span ( [ 2 1 1 ] T )
for λ 1 = 3 and
E ( λ 2 ) = span ( [ 0 1 1 ] T )
for λ 2 = 1.
(For λ 3 = 6 there are no eigenvectors).
Can the above eigenvectors and eigenvalues be used for solving this maximization problem?
asked 2022-05-08
Given a right circular cone with the line of symmetry along x = 0, and the base along y = 0, how can I find the maximum volume paraboloid (parabola revolved around the y-axis) inscribed within the cone? Maximising the volume of the paraboloid relative to the volume of the right circular cone. In 2-D, the parabola has 2 points of tangency to the triangle, one of each side of the line of symmetry. I have tried using the disk method to find the volume of the cone, and the parabola, both with arbitrary equations such as y = b a x, and y = c d x 2 , but I end up with a massive equation for several variables, instead of a simple percentage answer. Any help is appreciated! Thanks in advance.
asked 2022-05-08
I want to solve the following optimization problem
maximize  f ( X ) = log d e t ( X + Y ) a T ( X + Y ) 1 a subject to  X W ,
where the design variable X is symmetric positive semidefinite, W , Y are fixed symmetric positive semidefinite matrices, and a is a given vector. The inequality X W means that X W is symmetric positive semidefinite.

I was wondering if there's hope to find an analytical solution to either the constrained or unconstrained problem. And if there is none, could I use convex optimization techniques to solve this numerically?
asked 2022-06-13
I have an optimization problem of the form
max γ f γ , w f g γ , w g
where f γ is the composition of f and γ and the inner product is defined as
f γ , w f = ( f γ ) ( t ) w f ( t ) d t
In other words I am trying to find a γ that maximizes the product of the two inner products above. I know one could pursue a gradient approach, which would be slow. Is there a more direct optimization method?
asked 2022-05-07
Statement:

A manufacturer has been selling 1000 television sets a week at $ 480 each. A market survey indicates that for each $ 11 rebate offered to a buyer, the number of sets sold will increase by 110 per week.

Questions :

a) Find the function representing the revenue R(x), where x is the number of $ 11 rebates offered.

For this, I got ( 110 x + 1000 ) ( 480 11 x ). Which is marked correct

b) How large rebate should the company offer to a buyer, in order to maximize its revenue?

For this I got 17.27. Which was incorrect. I then tried 840999, the sum total revenue at optimized levels

c) If it costs the manufacturer $ 160 for each television set sold and there is a fixed cost of $ 80000, how should the manufacturer set the size of the rebate to maximize its profit?

For this, I received an answer of 10, which was incorrect.
asked 2022-06-02
Is it true that for all symmetric n × n matrix, ( a i j ), such that for x { ± 1 } n
i j a i j x i x j 1
there exists a universal constant such that
i j a i j x i y j C
for all x , y { ± 1 } n . I tried to use the polarization identity
A x , y = A u , u A v , v
where u = ( x + y ) / 2 and v = ( x y ) / 2. However, as x and y vary over ± 1 vectors, u and v can be vectors in { ± 1 , 0 }.

New questions