Scribd
Upload
23 views

Optim_1_Final_Exam_June_2016

Uploaded by

phuongmaivu744
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views

Optim_1_Final_Exam_June_2016

Uploaded by

phuongmaivu744
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

INTERNATIONAL UNIVERSITY (IU) - VIETNAM NATIONAL UNIVERSITY - HCMC

FINAL EXAMINATION
June 2016
Duration: 120 minutes

SUBJECT: OPTIMIZATION 1
Deputy Head of Dept. of Mathematics: Lecturer:

Assoc. Prof. Pham Huu Anh Ngoc Assoc. Prof. Nguyen Ngoc Hai

INSTRUCTIONS: Each student is allowed a scientific calculator and a


maximum of two double-sided sheets of reference material (size A4 or sim-
ilar), stapled together and marked with their name and ID. All other docu-
ments and electronic devices are forbidden.

Question 1 (25 marks) Consider the function


1 1 1
f (x1 , x2 ) = x31 + x21 + 2x1 x2 + x22 − x2 + 9, (x1 , x2 ) ∈ R2 .
3 2 2
Find all stationary points of this function and determine whether they are
local minimizers and maximizers.
Question 2 (25 marks) Let C ⊂ Rn be a nonempty closed convex set and
x0 ∈ Rn . Consider the problem
minimize f (x) = |x − x0 |2
subject to x ∈ C.
(a) Show that f is strictly convex and the problem has a unique solution.
(b) Show that x∗ ∈ C is the solution of the problem if and only if
(x0 − x∗ )T (x − x∗ ) ≤ 0 for any x ∈ C.

Question 3 (25 marks) Suppose the cone z 2 = x2 + y 2 is sliced by the


plane z = x + y + 2 so that a conic section C is created. Use Lagrange
multipliers to find the points on C that are nearest to the origin in R3 .

------continued on next page------

1
Question 4 (25 marks) Consider the problem

minimize f (x1 , x2 ) = x21 + x22 + 2x1


subject to x1 + x2 = 0.

What is the dual to this problem? Solve the dual problem.

***END OF QUESTION PAPER***

2
SOLUTIONS

Question 1 Stationary points satisfy the system


fx1 (x1 , x2 ) = x21 + x1 + 2x2 = 0
fx2 (x1 , x2 ) = 2x1 + x2 − 1 = 0.
The second equation gives x2 = 1 − 2x1 , and if this is substituted into the
first equation, we obtain x21 − 3x1 + 2 = 0 so that x1 = 1 or x1 = 2. Hence
there are two stationary points:
   
1 2
xa = and xb = .
−1 −3
The Hessian matrix for f is
 
2x1 + 1 2
F(x) = ,
2 1
so    
3 2 5 2
F(xa ) = and F(xb ) = .
2 1 2 1
F(xb ) is positive definite, so xb is a local minimizer. However F(xa ) is indef-
inite, and xa is neither a local minimizer nor a local maximizer of f .

Question 2 We have ∇f (x) = 2(x − x0 )T and F(x) = 2I for all x ∈ Rn .


Since F(x) is positive definite for all x, f is strictly convex. Further more,
since lim|x|→∞ f (x) = ∞, the given problem has a unique solution. As the
given problem is convex, x∗ ∈ C is an optimal solution if and only if it
satisfies the condition
∇f (x)(x − x∗ ) ≥ 0 for all x ∈ C,
that is, if and only if
2(x − x0 )T (x − x∗ ) ≥ 0 for all x ∈ C,
The latter condition is equivalent to
(x0 − x∗ )T (x − x∗ ) ≤ 0 for all x ∈ C.

Question 3 The problem is


minimize f (x, y, z) = x2 + y 2 + z 2
subject to g1 (x, y, z) = x2 + y 2 − z 2 = 0
g2 (x, y, z) = x + y − z + 2 = 0.

3
Since the constraint set is closed and lim|(x,y,z)|→∞ f (x, y, z) = ∞, the given
problem has an optimal solution. Further more,
   
2x 1
∇g1 (x, y, z) =  2y  , ∇g2 (x, y, z) =  1 
−2z −1

These vectors are linearly dependent only when x = y = z. However, no


feasible point satisfies these conditions. Hence every feasible point is regular.
The KKT conditions and constraints are
2x + 2λ1 x + λ2 = 0
2y + 2λ1 y + λ2 = 0
2z − 2λ1 z − λ2 = 0
x2 + y 2 − z 2 = 0
x+y−z+2=0

Subtracting the first equation from the second gives

2(x − y)(1 + λ1 ) = 0,

whence x = y or λ1 = −1. The condition λ1 = −1 implies λ2 = 0, and the


third equation of the system becomes 4z = 0, so z = 0. If z = 0, then x and
y must be zero by the fourth equation. However, (0, 0, 0) is not a point on
the plane z = x + y + 2. Thus, λ1 6= −1 and so x = y. The the constraint
equations become
2x2 − z 2 = 0
2x − z + 2 = 0.
√
Substituting z = 2x + 2 yields 2x2 − (2x + 2)2 = 0. Hence x = −2 ± 2 . We
therefore get two points
 √ √ √ 
a1 = −2 + 2 , −2 + 2 , −2 + 2 2
 √ √ √ 
a2 = −2 − 2 , −2 − 2 , −2 − 2 2 .

Then √ √
f (a1 ) = 24 − 16 2 , f (a2 ) = 24 + 16 2 .
Therefore a1 is the nearest to the origin.

Question 4 The Lagrangian function is

L(x1 , x2 , λ) = x21 + x22 + 2x1 + λ(x1 + x2 ), (x1 , x2 ) ∈ R2 , λ ∈ R.

4
The stationary points satisfy the system

Lx1 (x1 , x2 , λ) = 2x1 + 2 + λ = 0


Lx2 (x1 , x2 , λ) = 2x2 + λ = 0.

The unique stationary point is (−1 − λ/2, −λ/2). Moreover,


 
2 2 0
∇x L(x1 , x2 , λ) = ,
0 2

which is positive for all (x1 , x2 ) ∈ R2 . Thus for each λ, L(x1 , x2 , λ) is convex
on R2 . Hence the dual function is
 
λ λ
φ(λ) = min 2 L(x1 , x2 , λ) = L − 1 − , − , λ
(x1 ,x2 )∈R 2 2
2
λ2
    
λ λ λ λ
= 1+ + −2 1+ +λ −1− −
2 4 2 2 2
2
λ
= − − λ − 1, λ ∈ R.
2
The dual problem is
λ2
max φ(λ) = − − λ − 1.
λ∈R 2
We have
φ0 (λ) = −λ − 1 = 0 ⇐⇒ λ = −1.
Since φ is concave, max φ = φ(−1) = − 12 − (−1) − 1 = − 12 .

You might also like