Exercices 1

Sorin-Mihai GRAD ENSTA Paris
Differentiable Optimization 1 - APM 4OPT1 TA / 2024

Exercise 1 [05/11/2024]
1 Which of the following sets are convex or cones?
(i) ∅;
(ii) {(0, 0)⊤ } ∪ int R2+ ;
(iii) the set of symmetric positive semidefinite matrices
n o
S+n
= A ∈ Rn×n : A = A⊤ , x⊤ Ax ≥ 0 ∀x ∈ Rn ;
(iv) R2+ \ {(x, x)⊤ : x ∈ R};

(v) the set of double stochastic matrices
 
 n
X n
X 
Bn = (xij )i∈I n
j∈In ∈ R n×n
: x ij = x ij = 1, 0 ≤ x ij ≤ 1, i, j ∈ In .
 
i=1 j=1
2 Give an example of a nonconvex cone.
3 A two-stroke engine is operated with an oil-gasoline mixture, that contains at least 4% oil,
and at least 85% gasoline. The mixture should be as cost-effective as possible. Formulate
the corresponding (linear) optimization problem.
4 Three types of tables are produced in a joinery. The delivery of a minimum number of
tables has already been contractually agreed with the customers. Write an optimization
problem that models the profit maximization by taking into consideration the time and
material capacities specified below.
table 1 table 2 table 3 capacity

gain / unit 3 1 2
necessary time / unit 4 2 1 58
necessary material / unit 2 1 3 63
agreed quantity 4 2 3
5 On a chessboard (8 × 8) place as many queens as possible without threatening each other.

Formulate this as an (integer linear) optimization problem. (Recall that queens move
horizontally, vertically, or diagonally any distance on the chessboard.)
6 Rewrite the problem

n o
⊤
(P6 ) infn max cj x ,
x∈R+ j∈Im
where cj ∈ Rn ∀j ∈ Im , as a linear optimization problem.
7 (from Euclid’s Elements, one of the oldest mathematical treatises known to mankind)
1
For a given triangle ∆ABC, find an inscribed parallelogram ADEF of largest
possible area such that EF ||AB as well as DE||AC hold.
Let H and b denote the height of the triangle ABC and the length of the straight line
segment AC, respectively. Any inscribed parallelogram with the required properties can
be uniquely determined by choosing a point F on AC (see the figure). Let x < b be the
distance between the points A and F .
Show that the above problem is equivalent to the following quadratic optimization problem
Hx(b − x)
sup .
0<x<b b
8 Around 1615, the famous astronomer Johannes Kepler formulated the following problem
in his work Stereometria Doliorum Vinariorum (= Stereometry of Wine Barrels).
For a given sphere, find the inscribed cylinder of largest possible volume.
Formulate this Keplerian problem as a nonlinear optimization problem.
9 (backpack problem) A student is considering which of the books he is interested in to take

home from the library. Each book i ∈ In has a utility pi ≥ 0 for him and a weight wi > 0.
He can carry a maximum weight W . Formulate as an optimization problem the following
question
Which books should he take home so that their total utility (=sum of their
utilities) is maximal?
10 (location optimization) The location of a rescue station with helicopter aimed to serve m
villages is to be planned. The measure for the quality of a location is the weighted sum of
the distances between the helipad and the villages (the smaller, the better). The weights
of the localities are proportional to their respective populations. The aviation authority
allows a helipad only in a certain sector of a circle K := {x ∈ R2+ : ∥x∥2 ≤ 1}. Set up a
suitable mathematical model.
11 Formulate the following optimization problem as a problem with inequality constraints
(P11 ) inf f (x),

h(x)=0
x∈Rn+
where f : Rn → R and h : Rn → Rm .
12 Solve the following optimization problems by means of the graphical method
2
(x1 − 2)2 + (x2 − 1)2 ;

(i) (P12 ) inf
x1 ,x2 ∈R,
x21 ≤x2 ≤2−x1

(ii) (P12 ) sup 3x1 + 2x2 .
x1 ,x2 ≥0,x1 +x2 ≤8,
−x1 +3x2 ≤12,2x1 −x2 ≤10
Draw the level lines of the objective functions and the feasible sets and mark the optimal
solutions.
13 The (orthogonal) projection of a point p ∈ Rn onto a nonempty closed convex set U ⊆ Rn

is the point x̄ ∈ U which solves the optimization problem
(P13 ) inf ∥x − p∥2

x∈U
Show that the projection is well-defined (and thus defines a function PU : Rn → U ).

Exercices 1

Uploaded by

Copyright:

Available Formats

Exercices 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Exercices 1

Uploaded by

Copyright:

Available Formats

Sorin-Mihai GRAD ENSTA Paris

Differentiable Optimization 1 - APM 4OPT1 TA / 2024

1 Which of the following sets are convex or cones?

(iv) R2+ \ {(x, x)⊤ : x ∈ R};

2 Give an example of a nonconvex cone.

table 1 table 2 table 3 capacity

5 On a chessboard (8 × 8) place as many queens as possible without threatening each other.

6 Rewrite the problem

where cj ∈ Rn ∀j ∈ Im , as a linear optimization problem.

Formulate this Keplerian problem as a nonlinear optimization problem.

9 (backpack problem) A student is considering which of the books he is interested in to take

11 Formulate the following optimization problem as a problem with inequality constraints

(P11 ) inf f (x),

12 Solve the following optimization problems by means of the graphical method

13 The (orthogonal) projection of a point p ∈ Rn onto a nonempty closed convex set U ⊆ Rn

(P13 ) inf ∥x − p∥2

Show that the projection is well-defined (and thus defines a function PU : Rn → U ).

You might also like