Calculus of Variations - Answers To Exercises

Calculus of Variations — Answers to Exercises
13 February 2015, Niels Chr Overgaard
Answers to problems for Lecture 1 and Lecture 2

1.1 Consider minimization of the functional
Z 1
J [ y] = y( x)2 y0 ( x)2 dx,
0
subject to the boundary conditions y(0) = 0 and y(1) = 1.

a) Determine an upper bound to the minimum J ∗ of this problem by restricting
the investigation to trial-functions of the type y² ( x) = x² where ² > 1/4.
(M. Mesterton-Gibbons, Exercise 1.1.)
Answer. The function j (²) = J [ x² ] = ²2 /(4² − 1), ² > 1/4, attains its minimum
for ² = 1/2 with j (1/2) = 1/4. Since all the y² = x² are admissible for the pro-
blem, J ∗ = inf J ≤ 1/4. (Since the family of functions y² do not exhaust the
set of admissible functions, strict inequality J ∗ < 1/4 may hold. But see part
b). )
b) Guess the function which solves the variational problem. Can the guess
be verified?
p
Answer. A good guess is φ( x) = y1/2 ( x) = x. That φ solves the minimiza-
tion problem can be cheked using a direct verification. An alternative is to
introduce the new dependent variable z( x) = y( x)2 /2 in the functional J .
1.2. Show that the descent time down the cycloid curve generated by a rolling
circle of radius R ,
x(θ ) = a + R (θ − sin θ )
(
, 0 ≤ θ ≤ 2π ,
y(θ ) = ya − R (1 − cos θ )
is given by T = R / g(θb − θa ) where θb is the angle corresponding to the

p
terminal point ( b, yb ) and θa = 0 is the angle corresponding to the initial

point (a, ya ). What is the descent time to the lowest point on the cycloid?
(M. Kot, Exercise 1.6.1.)
Answer. Given in the problem text.
1.3. Try to guess the solution of the problem of minimizing the functional
Z 2
J [ y] = y( x)2 (1 − y0 ( x))2 dx
0
subject to y(0) = 0 and y(2) = 1. Which function class does your guess belong
to? How do you know that the function you guessed is a minimizer?
1
Answer. A natural guess is
0 for 0 ≤ x < 1,
½
φ( x ) =
x − 1 for 1 ≤ x ≤ 2.
because φ ∈ D 1 (the class of piecewise continuously differentiable functions)

is admissible and J [φ] = 0. Optimality of φ follows because the integrand in
J [ y] is non-negative for any admissible function y.
1.4. Show that the straight line φ( x) = α + (β − α)( x − a)/( b − a) minimizes the
curve length functional
Z b q
0
L[ y] = f ( y ( x)) dx, f ( p) = 1 + p2 ,
a
subject to y(a) = α and y( b) = β, by establishing the inequality L[ y] − L[φ] ≥ 0

for all admissible y. Do this by writing y( x) = φ( x) + h( x), with h(a) = h( b) =
0, and make a first-order Taylor expansion of f (φ + h) with remainder in
Lagrange’s form. (This is an example of a direct verification.)
Answer. The proof can be found on p. 22 in the classical book Calculus of
Variations by Gilbert Ames Bliss. The first edition was published in 1925.
(Accessible online for LU-students via
http://universitypublishingonline.org/maa/ebook.jsf?bid=
CBO9781614440017
Please, have a look.)
2.0. Consider the problem of minimizing the functional
Z 1
J [ y] = y( x)2 y0 ( x)2 dx
0
subject to the end point conditions y(0) = 0, y(1) = 1. Determine its admissib-
le extremals. (Cf. Problem 1 from the first lecture.)
Answer. Euler’s equation is y( yy0 )0 = 0 so the extremals are the parabolas
y2 = Ax + B (with y = 0 as a special case, of course.) The admissible extremal
p
is φ( x) = x, as we would expect from Exercise 1.1.
2.1. Find the admissible extremals for the problem of minimizing the func-
tional Z b
1 0
J [ y] = ( y ( x) − x)2 dx
0 2
subject to the end point conditions y(0) = 0 and y( b) = β, where b > 0 och
β ∈ R are given numbers.
2
Guess the minimum of the above problem. Write the answer as a function
S = S ( b, β) of the coordinates of the right end point ( b, β). Can the guess be
verified?
Answer. The extremals are of the form y( x) = 21 x2 + Ax + B where A, B ∈
R. The admissible extremal is φ( x) = 12 x2 − (β − 12 b2 ) x/ b, and S ( b, β) = (β −
1 2 2
2 b ) /2 b, b > 0. Verification is possible, but is not presented here.
2.2. Determine the admissible extremals of the functionals below subject to
the end point conditions y(0) = 0 and y(1) = 1.
R1
a) J [ y] = 0 yy0 dx.
R1
b) J [ y] = 0 y2 + 2 x yy0 dx.
R 1 x+ yy0
c) J [ y] = 0 p 2 2 dx.
1+ x + y
What are the extremals (i.e. solutions of Euler’s equation)? Can the above
examples be generalized?
Answer to a). J [ y] = 21 y(1)2 − 21 y(0)2 = 21 for all y satisfying y(0) = 0 and
d d
y(1) = 1. If F = yy0 then Euler’s equation F y − dx F y0 = 0 becomes y0 − dx y=
0 0
y − y = 0 identically for all admissible functions y.
2.3. Determine the minimum value of the functional
Z 1
J [ y] = − x yy0 dx
0
among the functions in C 1 which satisfies the end point conditions y(0) = 0
and y(1) = 1. Does the problem have admissible extremals?
R1
Answer. J [ y] = − 12 + 0 21 y( x)2 dx ≥ −1/2, so inf J ≥ −1/2. Since J [ x k ] → −1/2
as k → ∞ it follows that inf J = −1/2. Notice that Euler’s equation is y = 0 so
the zero function is the only extremal of J , and it is not admissible for the
problem.

Calculus of Variations - Answers To Exercises

Uploaded by

Copyright:

Available Formats

Calculus of Variations - Answers To Exercises

Uploaded by

Document Information

Original Description:

Original Title

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

Calculus of Variations - Answers To Exercises

Uploaded by

Copyright:

Available Formats

Calculus of Variations — Answers to Exercises

13 February 2015, Niels Chr Overgaard

Answers to problems for Lecture 1 and Lecture 2

subject to the boundary conditions y(0) = 0 and y(1) = 1.

is given by T = R / g(θb − θa ) where θb is the angle corresponding to the

terminal point ( b, yb ) and θa = 0 is the angle corresponding to the initial

because φ ∈ D 1 (the class of piecewise continuously differentiable functions)

subject to y(a) = α and y( b) = β, by establishing the inequality L[ y] − L[φ] ≥ 0

You might also like