7.10.

2022

𝝅/𝟐
Ex[128]: Find the extremal for the functional 𝑱 𝒙 = ∫𝟎 𝒙̇ 𝟐 𝒕 − 𝒙𝟐 𝒕 𝒅𝒕 which satisfies the
boundary conditions as 𝒙 𝟎 = 𝟎 and 𝒙(𝝅/𝟐) = 𝟏.

Introduction to Optimal Control, Lecture Notes #3 27

Problem 2: Fixed-End Time and Free-End State System (The initial time(𝑡 ) and state(𝑥 ) and
the final time(𝑡 ) are fixed and state(𝑥 ) is free.)
The admissible curves which all begin at the same point and terminate on vertical line as seen
on the Figure below:

Formulation of the problem is similar to the Pr. 1. The variation and integration of the variation
is similar process with the formulation of the Pr.1, but, now 𝛿𝑥 𝑡 = 0 and 𝛿𝑥 𝑡 is arbitrary
for all admissible trajectories.

3.8

Introduction to Optimal Control, Lecture Notes #3 28

14
 7.10.2022

For an optimal point we know that 𝛿𝐽 𝑥 ∗ , 𝛿𝑥 must be zero. So that we need to show (3.8) must
be zero on an extremal.
Suppose that 𝑥 ∗ is an extremal for free end point problem as shown in the following figure
(𝑥 ∗ 𝑡 = 𝑥 ):

This case is similar with the Pr.1, so that this is also the solution of Euler Equ. Given in (3.7). In
other words, an optimal for free end problem is also an optimal for the fixed end point problem
with the same end points and same functional. Hence, regardless the boundary conditions, the
Euler Equ. must be satisfied:
𝛿𝐽 𝑥 ∗ , 𝛿𝑥 = 0 and (𝑥 ∗ 𝑡 , 𝑥̇ ∗ 𝑡 , t)− [ ̇
𝑥 ∗ 𝑡 , 𝑥̇ ∗ 𝑡 , t) = 0
for all 𝑡 ∈ [t ,t ] for the (3.8).

̇
(𝑥 ∗ 𝑡 , 𝑥̇ ∗ 𝑡 , 𝑡 ] 𝛿𝑥 𝑡 =0
Since, 𝑥 =𝑥(𝑡 ) is free, 𝛿𝑥 𝑡 is arbitrary. Then, it is required that ̇ (𝑥 ∗ 𝑡 , 𝑥̇ ∗ 𝑡 , 𝑡 ]=0.
Note that: The Euler equ. is second order we need two boundary cond.s as 𝑥(𝑡 ) = 𝑥 and

̇
(𝑥 ∗ 𝑡 , 𝑥̇ ∗ 𝑡 , 𝑡 ]=0.

Introduction to Optimal Control, Lecture Notes #3 29

Ex[HW]: Find the smooth trajectory of smallest length connecting the point x(0)=1 to the line
t=5.

Ex[133]: Determine an optimal for the funcional 𝐽 𝑥 = ∫ 𝑥̇ 𝑡 + 2𝑥 𝑡 𝑥̇ 𝑡 + 4𝑥 𝑡 𝑑𝑡 as

𝑥 0 = 0 and x(2) is free.

Introduction to Optimal Control, Lecture Notes #3 30

15
 7.10.2022

3.17

Also, this problem cover the extremal of for free end point problem(Pr.1), therefore the Euler Eq.
must be provided and as the 𝛿𝑡 is arbitrary, so its coefficients must be zero:

Introduction to Optimal Control, Lecture Notes #3 35

Ex(137): Find the extremal for the functional 𝐽 𝑥 = ∫ 2𝑥 𝑡 + (1/2)𝑥̇ 𝑡 𝑑𝑡 with 𝑥 1 = 4

and x 𝑡 = 4 and 𝑡 > 1 is free.

Introduction to Optimal Control, Lecture Notes #3 36

18
 7.10.2022

Ex.(141): Find the extremal trajectory for the functional 𝐽 𝑥 = ∫ 1 + 𝑥̇ 𝑡 𝑑𝑡 as the

boundary conditions of 𝑡 = 0, 𝑥 0 = 0 are fixed and t and 𝑥(𝑡 ) are free. Moreover, 𝑥(𝑡 ) is
required to lie on the line 𝜃 𝑡 = −5𝑡 + 15. The functional 𝐽(𝑥) is the length of the curve 𝑥,
thus the function that minimizes the 𝐽 to the shortest curve from the origin to the specified line.

Introduction to Optimal Control, Lecture Notes #3 41

Functionals Involving Several Independent Functions

Let us generalize the discussion to include functionals that may be contain several independent
functions and its first variables.
Extremal problem with fixed-end points: Consider the following functional(𝑡 and 𝑡 are spe.d):

where 𝑥 , 𝑥 , … , 𝑥 are independednt functions with continuous first derivatives, and 𝑔 has
continuous first and second part. deriv. w.r.t. all of it arguments. And, the boundary cond.s:
𝑥 𝑡 = 𝑥 ,…,𝑥 𝑡 = 𝑥
𝑥 𝑡 = 𝑥 ,…,𝑥 𝑡 = 𝑥 .
Now, our aim to apply fundemental theorem to determine the necessary condition for the
extremum function of 𝑥 ∗ , 𝑥 ∗ , … , 𝑥 ∗ . So that, let us to find the variation of 𝐽 by retaining only the
first-order terms:

Expanding in a Taylor series about 𝑥 𝑡 , … , 𝑥 𝑡 , 𝑥̇ 𝑡 , 𝑥̇ 𝑡 , … , 𝑥̇ (𝑡):

Introduction to Optimal Control, Lecture Notes #3 42

21
 7.10.2022

0 is nx1 matrix with zeros. The matrix representation of the Euler equations is as follows:

𝝏𝒈 ∗ 𝒅 𝝏𝒈 ∗
(𝒙 𝒕 , 𝒙̇ ∗ 𝒕 , t)− [ 𝒙 𝒕 , 𝒙̇ ∗ 𝒕 , t) = 0
𝝏𝒙 𝒅𝒕 𝝏𝒙̇

Note that, if one consider the x as scalar then we get the result for Pr.1.

Introduction to Optimal Control, Lecture Notes #3 45

Ex.1(146): Find the euler equ.s for the functional 𝐽 = ∫ [𝑥 (𝑡) 𝑥 𝑡 +𝑡 𝑥 ̇ 𝑡 −

𝑥 ̇ (𝑡)𝑥̇ (𝑡)]𝑑𝑡 with the end points 𝑡 , 𝑡 , 𝑥 𝑡 , 𝑥 𝑡 , 𝑥 𝑡 , 𝑥 𝑡 are specified.

Introduction to Optimal Control, Lecture Notes #3 46

23
 7.10.2022

̇ + 𝑥 ̇ (𝑡)]𝑑𝑡 where 𝑥 and 𝑥 is

Ex.:(149) Find an extremal for 𝐽 𝑥 = ∫ [𝑥 (𝑡) + 𝑥 ̇ 𝑡 𝑥 (𝑡)
independed and the boundary cond.s are 𝑥 0 = 1, 𝑥 𝑝𝑖/4 = 2, 𝑥 0 = , 𝑥 𝑝𝑖/4 is free.

Introduction to Optimal Control, Lecture Notes #3 49

The Second Variation

So far, we have only considered the the necessary conditions for functional and the condition
that the first variation vanish leading to the classical Euler-Lagrange equation.

On the other hand, to find the optimum, we need the examine the second variation to
characterize the extremum.

For this aim, let us define the functional for Pr.1. and deal the increment for this problem:

3.27

Note that this result is same with (3.3) if definiton of J includes the function g. We need to deal
the terms corresponding the second variation as follows:

3.28

Introduction to Optimal Control, Lecture Notes #3 50

25