Calculus of Variation and its Application

Tien-Tsan Shieh
Institute of Mathematics
Academic Sinica
July 14, 2011
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 1 / 1
Outline
1
Basic ideas in the Calculus of Variations

Some review on calculus

Classical variational problems

The indirect method:

The rst variation and the Euler-Lagrange equation

The second variation

The direct method

2
Applications of the Calculus of Variations in Physics and Chemistry

The Fermats Principles in Optics

The principle of least action

The law of maximal entropy

Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 2 / 1
Calculus
Calculus was developed in the late 17th century by Newton and
Leibniz individually.
It is a mathematical discipline focus on limit, derivative, integral and
innite series.
Derivative of a function at the point x
0
:
f

(x
0
) = lim
xx
0
f (x) f (x
0
)
x x
0
.
We also denote by
df
dx
(x
0
) as well.
Integral of a function:
_
b
a
f (x) dx = lim
n, x
i
0,
n1

i =0
f (x
i
)x
i
where x
i
= x
i +1
x
i
.
Fundamental theorem of calculus
_
b
a
f

(x) dx = f (b) f (a).

Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 3 / 1
Finding the Maximum and Minimum
Finding maximums and minimums for a giving a smooth function f (x):
Considerer the Taylor expansion of the smooth function f (x):
f (x) = f (x
0
) + f

(x
0
)(x x
0
) +
f

(x
0
)
2
(x x
0
)
2
+ higher oreder terms
f

(x
0
) > 0 implies the function f (x) is increasing around x
0
.
f

(x
0
) < 0 implies the function f (x) is decreasing around x
0
.
f

(x
0
) = 0 and f

(x
0
) > 0 implies x
0
is the local minimum.
f

(x
0
) = 0 and f

(x
0
) < 0 implies x
0
is the local maximum.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 4 / 1
Review on Multi-dimensional Calculus
Suppose f is a C
2
-function on R
2
.
The directional derivative of the function f at

x along the
direction

v is dened as
D

v
f (

x ) lim
t0
f (

x + t

v ) f (

x )
t
= f (

x )

v
Direction derivative can also be denoted by

v
f (x),
f (x)
v
, f

v
(x), D
v
f (x).
The derivative of f at x R
2
along the direction v:
D
2
v
f (x) lim
t0
f (x + tv) f (x) tv f (x)
t
2
= f
x
1
x
1
(x)v
2
1
+ 2f
x
1
x
2
(x)v
1
v
2
+ f
x
2
x
2
(x)v
2
2
Here x = (x
1
, x
2
) and v = (v
1
, v
2
).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 5 / 1
Taylor expansion of a multi-dimensional function
f (x + tv) = f (x) + t D
v
f (x) +
t
2
2
D
2
v
f (x) + higher order terms
= f (x) + t (f
x
1
(x)v
1
+ f
x
2
(x)v
2
)
+
t
2
2
_
f
x
1
x
1
(x)v
2
1
+ 2f
x
1
x
2
(x)v
1
v
2
+ f
x
2
x
2
(x)v
2
2
_
+ higher order terms
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 6 / 1
Johann Bernoullis Challenge
The calculus of variations as a recognizable as a part of mathematics had
its origins in Johann Bernoullis challenge in 1696 to the mathematicians
of Europe to nd the curve of quickest descent, or brachistochrone. The
brachistochrone means the shortest time in Greek.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 7 / 1
Brachistochrone Curve
Newton was challenged to solve the problem, and did so the very next
day.
In fact, the solution, which is a segment of a cycloid, was found by
Leibniz, LHospital, Newton, and the Johann Bernoullis and Jakob
Bernoullis.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 8 / 1
Formulation of the brachistochrone problem
The time of travel from a point P
1
to a point P
2
is given by
Time =
_
P
2
P
1
ds
v
where s is the arclength and v is the speed.
1
2
mv
2
= mgy v =
_
2gy
ds =
_
dx
2
+ dy
2
=
_
1 + (
dy
dx
)
2
dx
Therefore we obtain the following functional
F(y) :=
_
P
2
P
1
_
1 + y
2

2gy
dx
You may think functional F as a function dened on a class of paths
(x, y(x)) which connect two points P
1
and P
2
.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 9 / 1
Geodesic Problems
Problem: Find a curve, joining points A and B, with a shortest distance.
Geodesic problem on a plane.
Answer: a straight line joining the points A and B.
Geodesic problem on a sphere.
Answer: a great circle joining the points A and B.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 10 / 1
Geodesic problem on a sphere
Using spherical coordinate,
Y = (R cos sin , R sin sin , R cos )
A given the curve joining A to B could be
represented by a pair of function ((t), (t))
for t [0, 1] with (0) = (1) = 0; (1) =
1
The resulting curve dened by
Y(t) = (R cos (t) sin (t), R sin (t) sin (t), R cos (t))
and its derivative is
Y

(t) = R

sin (sin )

+ cos (cos )

, cos (sin )

+ sin (cos )

, (sin )

(t).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 11 / 1
Geodesic problem on a sphere
The length of the curve is
L(Y) :=
_
1
0
|Y

(t)| dt = R
_
1
0
_
sin
2
(t)

(t)
2
+

(t)
2
dt
The Problem
Find a minimum of the functional L among a curve Y(t), represented by
((t), (t)), satisfying (0) = (1) = 0; (1) =
1
min L(Y) = min R
_
1
0
_
sin
2
(t)

(t)
2
+

(t)
2
dt
L(Y) R
_
1
0

(t) dt = R(t)

1
0
= R
1
Let Y
1
(t) is a curve represented by (t) = 0, (t) = t
1
. Then
L(Y
1
) = R
_
1
0
_
sin
2
(t
1
) 0
2
+
2
1
dt = R
1
.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 12 / 1
Minimal Area Problem
Find a least-area surface among all
smooth surface which can be
represented as a graph of a smooth
function u = u(x, y) on a planar
domain D and satises the boundary
condition u|
D
= for some given
function dened on the boundary
D.
Let S(u) be the associated surface area. The problem is formulated in the
following form:
min
uA
S(u) = min
uA
_
D
_
1 + u
2
x
+ u
2
y
dxdy
where A is a collection of continuous dierentiable functions
A = {u C
1
(D) : u = g on D}.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 13 / 1
Plateaus Problem
Suppose C is a closed curve in space. What is the surface S of smallest
area having boundary C?
This problem is called Plateaus problem in honor of the physicist,
J. Plateau (1801-1883), who experimentally determined a number of
the geometric properties of soap lms and soap bubbles through
interesting experiments with soap lms.
In mathematics, Plateaus problem is to show the existence of a
minimal surface with a given boundary.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 14 / 1
The simplest One-dimensional Example:
The variational problem is
min
yA
E(y) = min
yA
_
1
0
|
dy
dx
|
2
dx
where the collection A of functions is given by
A = {y C
1
(0, 1) : y(0) = 0, y(1) = 2}.
Suppose y A is a minimizer of the functional E(y). Let (x) be any
continuous dierentiable function with (0) = (1) = 0.
E(y + t) =
_
1
0
|
dy
dx
+ t
d
dx
|
2
dx
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 15 / 1
The rst variation
E(y, )
dE(y + t)
dt

t=0
Since y is a minimizer of the functional E, the following derivative is zero.
E(y, ) = 2
_
1
0
_
dy
dx
+ t
d
dx
_
d
dx

t=0
dx = 2
_
1
0
dy
dx
d
dx
dx = 0
For any C
1
[0, 1] with (0) = (1) = 0, the following it true
_
1
0
dy
dx
d
dx
dx = 0.
By using the integration by part,
0 =
_
1
0
dy
dx
d
dx
dx =
dy
dx

1
0

_
1
0
d
2
y
dx
2
dx =
_
1
0
d
2
y
dx
2
dx.
The Euler-Lagrange equation:
_
d
2
y
dx
2
= 0 in (0, 1)
y(0) = 0, y(1) = 2
=y(x) = 2x.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 16 / 1
The second variation
The above discussion only guarantee the solution y = 2x is a critical
point for the functional E among all possible variation .
Consider the second variation

2
E(y, )
d
2
E(y + t)
dt
2

t=0
= 2
_
1
0
_
d
dx
_
2
dx > 0
Suppose we set (t) = E(y +t). According to the Taylor expansion
of the function (t) at t = 0, we nd
(t) = (0) +

(0) t +
1
2

(0) t
2
+ high order terms,
that is
E(y + t) = E(y) + E(y, ) t +
1
2

2
E(y, ) t
2
+ high order terms
We could conclude that
y = 2x
is a minimizer.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 17 / 1
Typical One-dimensional Variational Problems
Let L(x, y, z) a function with continuous rst and second (partial)
derivatives in x, y, z. Let E(y) be a functional dened by
E(y) =
_
1
0
L(x, y(x), y

(x)) dx
on the class A of functions
A = {y C
1
[0, 1] : y(0) = a , y(1) = b}.
Then the extreme y(x) of the functional E(y) satises the Euler-Lagrange
equation:
L
y

d
dx
L
y

= 0 in on (0, 1).
with the boundary condition y(0) = a, y(1) = b.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 18 / 1
The simplest 2-dimensional problem
Let D be some smooth bounded domain and g is a function dened on
the boundary D. Set
A = {u C
1
(D) : u(x, y) = g(x, y) for (x, y) D}.
The following is one of the simplest 2-d minimization problem
min
uA
E(u) = min
uA
_
D
u
2
x
+ u
2
y
dxdy
The Euler-Lagrange equation:

2
u
x
2
+

2
u
x
2
= 0 in D
with the boundary condition
u(x, y) = g(x, y) on (x, y) D.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 19 / 1
The typical 2-dimensional variational problem
Let D be some smooth bounded domain and g is a function dened on
the boundary D. Set
A = {u C
1
(D) : u(x, y) = g(x, y) for (x, y) D}.
Let L(x, y, z, p
1
, p
2
) be a C
2
function in all arguments. Let E(y) be a
functional dened by
E(u) =
_
D
L(x, y, u, u
x
, u
y
) dxdy
Then the extreme u(x, y) A of the functional E(u) satises the
Euler-Lagrange equation:
L
y

d
dx
_
L
u
x
_

d
dy
_
L
u
y
_
= 0 in D
with the boundary condition u(x, y) = g(x, y) on (x, y) D.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 20 / 1
The Area Minimizing Problem
Let S(u) be the associated surface
area.
min
uA
S(u) = min
uA
_
D
_
1 + u
2
x
+ u
2
y
dxdy
where A is a collection of continuous
dierentiable functions
A = {u C
1
(D) : u = g on D}.
The Euler-Lagrange equation

(u
x
, u
y
)
_
1 + u
2
x
+ u
2
y
= 0
with the boundary condition
u(x, y) = g(x, y) for (x, y) D.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 21 / 1
Ideas from Calculus
In calculus, suppose f (x) is a C
2
-continuous function.
f

(x
0
) = 0 x
0
is a critical point.
In order to check whether x
0
is a maximal point or a minimal point
we need to check the sign of the second derivative.
f

(x
0
) > 0 f (x
0
) is a local minimal.
f

(x
0
) < 0 f (x
0
) is a local maximum.
Suppose f (x, y) is a C
2
-continuous function. The Taylor expansion of
the function f (x, y) at (x
0
, y
0
) is
f (x
0
+ tu, y
0
+ tv) = f (x
0
, y
0
) + [f
x
(x
0
, y
0
)u + f
y
(x
0
, y
0
)v]t
+
1
2
[f
xx
(x
0
, y
0
)u
2
+ 2f
xy
(x
0
, y
0
)uv + f
yy
(x
0
, y
0
)v
2
]t
2
+higher order terms
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 22 / 1
Analogy to Calculus of Variation
Suppose E(y)
_
1
0
L(x, y(x), y

(x)) dx and
A = {y C
1
[0, 1] : y(0) = a , y(1) = b}.
The same issue happen in the calculus of variation.
E(u, ) = 0 for all possible variation
i.e.
L
y

d
dx
L
y

= 0 in on (0, 1).
only implies
y is a local extreme point in A.
In order to tell whether the extreme y of the functional E(y) is a
local minimum or a local maximum, we need check the second
variation of E(y).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 23 / 1
The Second Variation of E(y)

2
E(y, ) =
d
2
E(y + t)
dt
2

t=0
=
_
1
0
d
2
dt
2
L(x, y + t, y

+ t

t=0
dx
=
_
1
0
(L
yy

2
+ 2L
yy

+ L
y

y

2
) dx
Suppose y(x) is a local extreme, i.e. E(y, ) = 0 for all C
1
0
[0, 1]. We
have the following conclusion:
If L
y

y
(x, y(x), y

(x)) > 0 for all C

1
0
[0, 1] and
2
E(y, ) > 0 for
all C
1
0
[0, 1], then y(x) is a weak minimum.
If L
y

y
(x, y(x), y

(x)) < 0 for all C

1
0
[0, 1] and
2
E(y, ) < 0 for
all C
1
0
[0, 1], then y(x) is a weak maximum.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 24 / 1
Indirect Methods
The classical indirect method of variational problems is based on the
optimistic idea that every minimum problem has a solution. In order
to determine this solution, one looks for conditions which have to
satises by a minimizer. An analysis of the necessary conditions often
permits one to eliminate many candidates and eventually identies a
unique solution. For example, we only have one solution of the
Euler-Lagrange equation satisfying all prescribed conditions. We
tempted to infer that this solution is also be a solution to the original
minimum problem.
The above approach is false. We have to prove the existence of
minimizer before we could conclude the uniqueness of solutions to the
Euler-Lagrange equation.
Usually, in higher dimension solving the Euler-Lagrange equation is
more dicult than nding a minimizer of a functional .
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 25 / 1
The Direct Method for area minimizing problem
A surface u
1
with area + 1
A surface u
2
with area +
1
4
A surface u
3
with area +
1
16
Let F be the a functional dened on an
unit disk.
F(u) =
_
D
_
1 + u
2
x
+ u
2
y
dxdy
We can nd an area minimizing sequence
{u
n
} such that
lim
n
F(u
n
) = min
u
F(u) = .
If the area minimizing sequence is nice
enough, we could see the sequence {u
n
}
is getting closer and closer to the unit disk.
Guess: The minimal surface is the unit
disk.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 26 / 1
The Main Idea of Direct Method
Consider a minimization problem on some class A of functions:
min
uA
E(u)
Suppose there exist a minimizing sequence {u
n
} in A. i.e.
lim
n
E(u
n
) = min
uA
E(u) < +.
Suppose we could nd an element v
0
in A such that
u
n
v
0
as n .
Suppose the functional E has some kind of continuity. Therefore
lim
n
E(u
n
) = E(v
0
).
We conclude that v
0
is a minimizer of the functional E because
E(v
0
) = lim
n
E(u
n
) = min
uA
E(u).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 27 / 1
The Diculty of Direct Method
A surface with area + 1
A surface with area +
1
4
A surface with area +
1
16
A surface with area +
1
64
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 28 / 1
The Setting of the Direct Method
Consider a minimization problem on some class A of functions:
min
uA
E(u)
Suppose we have following properties on E and A.
1
(Lower-semi-continuity) For every sequence u
n
u
0
, we have
E(u
0
) liminf
n
E(u
n
).
2
(Compactness) For a bounded sequence {u
n
} in A, there exist a
convergent subsequence {u
n
k
} to some u
0
A.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 29 / 1
Existence of a Minimizer by the Direct method
Suppose {v
n
} is an minimizing sequence.
lim
n
E(v
n
) = inf
vA
E(v) < +
If we are possible to show {v
n
} is bounded in A, there there exist a
subsequence {v
n
k
} and v
0
A such that
v
n
k
v
0
as k .
By the lower-semi-continuity of the functional E, we have
E(v
0
) liminf
k
E(E
n
k
) = lim
n
E(v
n
) = inf
vA
E(v)
This implies v
0
is a minimizer.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 30 / 1
A minimum problem without a minimizer
min
uA
_
1
0
(u
2
x
1)
2
+ u
2
dx
where
A = {u C
1
[0, 1] : u(0) = u(1) = 0}
Choose a minimizing sequence {u
n
} which are zigzag functions with slope
1 and u
n
0 as n .
We have
lim
n
E(u
n
) = 0
but
E(0) = 1.
Therefore the minimum is not attained.
The direct method does not always work.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 31 / 1
Variational principles in physics
There are many laws of physics which are written as variational principles.
The Fermats principle in optics
The principle of least action This principle is equivalent to the
Newton second law of motion in a mechanical system.
The law of maximum entropy: Thermal equilibrium will reach the
maximum entropy of the system.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 32 / 1
Fermats Principle
In optics, the Fermats principle, also called the principle of the
least time, states that a path taken by a ray of light between two
points is the least-time path among all possible paths.
The law of reection and the law of refraction could be derived from
the Fermats principle.
Reection Refraction
The Fermats principle motivate the principle of least action in
mechanic systems.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 33 / 1
The Principle of Least Action
In Physics, the principle of least action states that the motion of a
mechanical system will follow the trajectory which minimize the
action of the system.
In classical mechanics, the action is dened as an integral along an
actual or virtual space-time trajectory q(t) connecting two space-time
events, initial event A (q
A
, t
A
= 0) and nal event
B (q
B
, t
B
= T). The action is dened as
S(q) =
_
T
0
L(q(t), q(t)) dt
where L(q, q) is the Lagrangian, and q =
dq
dt
. The Lagrangian is given
as the dierence of kinetic energy K and potential energy V along
the trajectory q(t):
L(q, q) = K V.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 34 / 1
Classical Mechanics
Lets consider the one-particle system in a conservative eld. The relation
between the potential energy of the particle and the force acting on the
particle is given by F = (F
1
, F
2
, F
3
) = V = (V
x
, V
y
, V
z
) or
Work = V(A) V(B) =
_
B
A
F ds.
The kinetic energy of the particle is
K =
1
2
m( x
2
+ y
2
+ z
2
).
The associated Lagrangian is
L(x, y, z, x, y, z) = K V =
1
2
m( x
2
+ y
2
+ z
2
) V(x, y, z).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 35 / 1
Consider the variational problem
min S(x(t), y(t), z(t)) = min
_
T
0
1
2
m( x
2
+ y
2
+ z
2
) V(x, y, z) dt
S =
_
T
0
m( x x + y y + z z) (V
x
x + V
y
y + V
z
z) dt
S =
_
T
0
m( xx + yy + zz) + (F
1
x + F
2
y + F
3
z) dt
S =
_
T
0
(F
1
m x)x + (F
2
m y)y + (F
3
m z)z dt
According to the principle of least action S = 0, we obtain
_
_
_
m x = F
1
m y = F
2
m z = F
3
.
This is the Newton second law of motion,
F = ma.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 36 / 1
Lagrangian Mechanics
The Lagrangian is given by
L(q
1
, . . . , q
n
, q
1
, . . . , q
n
) = K( q
1
, . . . , q
n
) V(q
1
, . . . , q
n
, q
1
, . . . , q
n
)
where K is the kinetic energy and V is the potential energy of the system
and (q
1
, . . . , q
n
) is a generalized coordinate. The action S is dened by
S(q
1
, . . . , q
n
) =
_
T
0
L(q
1
, . . . , q
n
, q
1
, . . . , q
n
) dt.
According to the principle of least action, the Euler-Lagrange equations are
_

_
L
q
1

d
dt
L
q
1
= 0
.
.
. =
.
.
.
L
q
n

d
dt
L
q
n
= 0
.
We usually called F (
L
q
1
, . . . ,
L
q
n
) a generalized force and
P (
L
q
1
, . . . ,
L
q
n
) a generalized momentum. Therefore, we derive a
Newton second law,

P = F, in the general coordinate.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 37 / 1
The principle of least action
min S(q
1
, . . . , q
n
) = min
_
T
0
L(q
1
, . . . , q
n
, q
1
, . . . , q
n
) dt
provide us a way to nd a suitable coordinate system for the
mechanical system.
The principle of least action could also be applied in the theory of
relativity, quantum mechanics and quantum eld theory.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 38 / 1
Legendre Transformation
A Legendre transformation
of a convex function f is
dened by
f

(p) = max
x
(px f (x)).
If f is dierentiable, then f

(p) can be interpreted as the negative of

the y-intercept of the tangent line to the graph of f that has slope p.
The value of x attains the maximum has the property
f

(x) = p.
f

(f

(x)) = x f

(x) f (x).
Another denition: Set p = f

(x). The inverse function theorem

implies x = x(p). Thus we dene
f

(p) = p x(p) f (x(p)).

Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 39 / 1
Properties of Legendre Transformation
Set p = f

(x). The inverse function theorem implies x = x(p). Thus we

dene
f

(p) = p x(p) f (x(p)).

We have following relations:
p =
df
dx
(x),
x =
df

dp
(p).
Also we have
f (x) + f

(p) = xp.
And also
f = f

Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 40 / 1
Hamiltonian Mechanics
The Lagrangian of a system is given by L(q, q). We introduce the
Hamiltonian function
H(q, p) = max
q
(p q L(q, q)).
This is equivalent to
H(q, p) = p q L(q, q) and p =
L
q
(q, q),
where the second equation gives us the relation q = q(q, p).
Since the double Legendre transformation is itself, we have
L(q, q) = p q H(q, p) and q =
H
p
(q, p).
Example: A system of a single particle
L(q, q) =
m
2
q
2
V(q). p =
L
q
= m q q =
p
m
H(q, p) = p qL(q, q) =
p
2
m
(
p
2
2m
V(q)) =
p
2
2m
+V(q) = Total energy
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 41 / 1
Hamiltons equations
Applying the principle of least action to L(q, q) = p q H(q, p), we have

_
t
2
t
1
L dt =
_
t
2
t
1
[p q H(q, p)] dt.
We will nd
_
t
2
t
1
_
q p + p q
H
p
p
H
q
q
_
dt = 0,
_
t
2
t
1
_
q
H
p
_
p +
_
p
H
q
_
q dt = 0.
This leads us to Hamiltons equations
_
q =
H
p
,
p =
H
q
.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 42 / 1
Thermodynamics
Thermodynamics is a branch of physics which study physical
systems through macroscopic quantities, such as temperature,
volume, pressure, etc.
Thermodynamics concerns phenomena that are in thermal
equilibrium. Studies of non-equilibrium physical processes are out of
the scope of the classical thermodynamics.
The results of thermodynamics are essential for other elds of physics
and chemistry, chemical engineering, aerospace engineering,
mechanical engineering, cell biology, biomedical engineering, materials
science, and economics.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 43 / 1
Laws of Thermodynamics
Zeroth law of thermodynamics: states that if two systems are in
thermal equilibrium with a third system, they are also in thermal
equilibrium with each other.
First law of thermodynamics: states that heat is a form of energy.
The law is no more than a statement of the principle of conservation
of energy. Energy can be transformed from one to another.
U = Q W
Q = Q
H
Q
C
Second law of thermodynamics: is also called the law of increase
of entropy. It states that if a closed system is in a conguration that
is not the equilibrium conguration, the most probable consequence
will be that the entropy of the system will increase monotonically in
successive of time.
Third law of thermodynamics: states that the entropy of a system
approaches to a constant value as the temperature approaches zero.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 44 / 1
Entropy
There are two denitions of entropy. One is the thermodynamic
denition and the other is the statistic mechanics denition.
Thermodynamic entropy is a non-conserved state function of
physical systems. Thermodynamic entropy is more generally dened
from the statistical viewpoint, in which the molecular nature of
matter is explicitly considered.
In statistical mechanics, statistical entropy is a measure of ways
which a system could be arranged. The entropy S is dened as
S = k
B
ln
where is a number of ways and k
B
is the Boltzmann constant.
Boltzmann showed the statistical entropy is equivalent to the
thermodynamic one.
The rst law of thermodynamics tells us that energy is conserved in a
thermodynamic system. The second law tells us natural processes
have a preferred direction of progresses.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 45 / 1
Thermal Equilibrium
A system that is in equilibrium experience no change when is
isolated from its surroundings.
In thermodynamics, a system is in thermal equilibrium when it is in
thermal equilibrium, mechanical equilibrium, radiative equilibrium,
and chemical equilibrium and chemical equilibrium.
One of important questions in
thermodynamics is what is the
equilibrium state after we
remove the boundary between
a isolated thermal system and
its surrounding.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 46 / 1
Callens Postulates of Thermodynamics
I. There exist particular states (called equilibrium states) that,
macroscopically, are characterized completely by the specication of
the internal energy U and a set of extensive parameters X
1
, X
2
, . . . ,
X
t
later to be specically enumerated.
II. There exists a function (called the entropy)of the extensive
parameters, dened for all equilibrium states, and having the following
property. The values assumed by the extensive parameters in the
absence of a constraint are those that maximize the entropy over the
manifold of constrained equilibrium states.
III. The entropy of a composite system is additive over the constituent
subsystems (whence the entropy of each constituent system is a
homogeneous rst-order function of the extensive parameters). The
entropy is continuous and dierentiable and is a monotonically
increasing function of the energy.
IV. The entropy of any system vanishes in the state for which
T
_
U
S
_
X
1
,X
2
,...
= 0.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 47 / 1
The Internal Energy of a system
U = U(X
0
, X
1
, X
2
, X
3
, . . . , X
t
) = U(S, V, X
2
, . . . , X
t
)
where X
0
denote the entropy S, X
1
the volume, and the remaining X
j
are
the mole numbers. For non-simple systems, the X
j
may represent
magnetic, electric, elastic extensive parameters to the system considered.
dU = TdS +
t

k=1
P
k
dX
k
=
t

k=0
P
k
dX
k
T =
U
S
P =
U
V
P
k
=
U
X
k
for k = 0, . . . t
TdS is the ux of heat and

t
k=1
P
k
dX
k
is the work.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 48 / 1
Entropy Maximum Principle and Energy Minimum Principle
S = S(U, X
1
, X
2
, . . . , X
t
) U = U(S, X
1
, X
2
, . . . , X
t
)
max
X
1
,...,X
t
S(U
0
, X
1
, . . . , X
t
)
Entropy Maximum Principle
min
X
1
,...,X
t
U(S
0
, X
1
, . . . , X
t
)
Energy Minimum Principle
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 49 / 1
Legendre Transformations of the Internal Energy
Experimenters frequently nd that the intensive parameters are the
more easily measured and controlled and therefore is likely to think of
the intensive parameters as operationally independent variables and of
the extensive parameters as operationally derived quantities.
A partial Legendre transformation can be made by replacing the
variables X
0
, X
1
, . . . , X
s
by P
0
, P
1
, . . . , P
s
.
U

(P
0
, P
1
, . . . , P
s
, X
s+1
, . . . , X
t
) = min
X
0
,...,X
s
_
U(X
0
, X
1
, . . . , X
t
)
s

k=0
P
k
X
k
_
.
The equilibrium values of any unconstrained extensive parameters
(X
s+1
, . . . , X
t
) in a system in contact with reservoirs of constant
P
0
, P
1
, . . . , P
s
minimize the thermodynamic free energy
(potential) U

.
min
X
0
,...,X
t

U(X
0
, X
1
, . . . , X
t
)
s

k=0
P
k
X
k

= min
X
s+1
,...,X
t
U

(P
0
, P
1
, . . . , P
s
, X
s+1
, . . . , X
t
).
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 50 / 1
Some Classical Thermodynamic Free Energies
Suppose the internal energy take the form
U = U(S, V, N).
Helmholtz Free Energy
A(T, V, N) = min
S
[U(S, V, N) TS]
min
V,N
A(T, V, N)
Gibbs Free Energy
G(T, p, N) = min
S,V
[U(S, V, N) + pV TS]
min
N
G(T, p, N)
Enthaply
H(S, p, N) = min
V
[U(S, V, N) + pV]
min
S,N
H(S, p, N)
Remark: Here pV related to the work done to the surrounding. It could be replaced by other kinds of work in other system.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 51 / 1
Nonuniform Binary Systems
Consider a binary system of A component and B component.
The Gibbs energy of the system take the form
G(c) =
_

|c|
2
+ c(1 c) + kT[c ln c + (1 c) ln(1 c)] dx
where c is the mole fraction of the A component and (1 c) is the mole
fraction of the B component.
T temperature
k Boltzmann constant
interaction constant between two
components
some constant
We will see that sucient cooling may lead
to phase separation. (Phase transition)
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 52 / 1
Equilibrium state
The equilibrium state c of the binary system will minimize the Gibbs energy
min
c
G(c) = min
c
_

|c|
2
+ f (c) dx.
Here we set
f (c) = (1 c
2
)
2
.
The Euler-Lagrange equation is
2c +
f
c
= 0 in
and the boundary condition
n c = 0 on .
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 53 / 1
One-dimensional Interface
Suppose interface is one-dimensional. We have to consider the problem
min
c
_

|
dc
dx
|
2
+ f (c) dx
and c(x) 1 as x and c(x) 1 as x .
2
d
2
c
dx
2
+
df
dc
= 0
_
2
d
2
c
dx
2
+
df
dc
_
dc
dx
= 0
d
dx
_

_
dc
dx
_
2
+ f (c)
_
= 0

_
dc
dx
_
2
+ f (c) = constant = 0
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 54 / 1

1/2
dc
dx
=
_
f (c) = 1 c
2

1/2
1 c
2
dc
dx
= 1
c(x) =
exp(
x

) exp(
x

)
exp(
x

) + exp(
x

)
= tanh
_
x

_
This tells us the thickness of the interface is

.
The energy concentrate near the interface.
We may guess
G(c) Area functional of the interface
as 0. It is true but the rigorous proof requires more mathematics.
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 55 / 1
Thank you for your attention!
Tien-Tsan Shieh (Institute of MathematicsAcademic Sinica) Calculus of Variation and its Application July 14, 2011 56 / 1