Chapter 2

INTRODUCTION TO CALCULUS OF VARIATION

(Lectures 4-5)

2.1 INTRODUCTION
In the previous chapter, we introduced finite element method as a method to solve
differential equations. More often, the behavior of a physical system is described by governing
differential equations. However, sometimes, it is convenient to derive an integral expression
(called variational form), minimization or maximization of which leads to same solution as
obtained by solving the governing differential equations. Given a variational form, one can
obtain the governing differential equations and solve differential equations by suitable numerical
method including FEM. Differential equations can also be transformed into a variational form.
The branch of mathematics, which deals with transforming a variational form to differential
equation form and vice versa is called calculus of variation. We will study the necessary
techniques of calculus of variation in this chapter. Similar to calculus, where we are often
interested in finding out a point at which a function attains minimum/maximum value, in
calculus of variation, we find out the function that provides minimum/maximum value of the
variational form. This chapter will provide a brief introduction to calculus of variation.

2.2 FUNCTIONAL
In calculus, we come across functions. A function provides a dependent variable, whose
value depends on one or many independent variables. For example, y = f (x) = x3 is a function, in
which for each value of independent variable x, there is a scalar value of dependent variable y.
Similarly, in function z = x2 + y2, for each value of x and y, there is a value of z.
Now consider the definite integral

I   ydx   f  x  dx
5 5
(2.1)
0 0

Here, for each particular function, there is a scalar value of I. For example, when f(x) = 1, the
value of I is 5. When f(x) = x, the value of I becomes 2.5. In literature, I is called a functional
(function of function), whose value depends on independent function f (x). Following integral
expression is also a functional:

I  y    F  x, y, y  dx
b
(2.2)
a

19
 dy
where F depends on x, y  y(x) and y  . Mathematically, a functional is an operator I,
dx
mapping y into a scalar value.
A functional l(y) is said to be linear in y, if and only if, it satisfies the following relation:
l  y   z    l  y    l  z  (2.3)

for any scalars,  and  and independent functions y and z. A functional B(y, z) is said to be
bilinear, if it is linear in each of its argument y and z, i.e.,
B  y1   y 2 , z    B  y1 , z    B  y 2 , z  (2.4)

B  y,  z1   z 2    B  y, z1    B  y, z 2  (2.5)

The functional is symmetric if

B  y, z   B  z , y  (2.6)

An example of a linear functional is

I  u    u f dx
L
(2.7)
0

where f is a constant function and u is the independent variable function. An example of a

symmetric bilinear functional is

I  u, v    E A
du dv
dx
L
(2.8)
0 dx dx
where E and A are constant functions and u and v are independent variable function.

2.3 EXTREMIZATION OF A FUNCTIONAL

 dy 
Let I  ab F  x, y,  dx be some functional. The relation between y and x is not known and
 dx 

the problem consists of finding this relation so that I is a maximum or a minimum. Assume that
y0(x) is some known relation between y and x, which extremizes I. Let another function in the
neighborhood of y0(x) is denoted by y0(x) + (x), where (x) is an arbitrary (but sufficiently
differentiable) function of x and  is an arbitrary small quantity. The function (x) does not
violate the geometric boundary conditions of the problem. Hence, wherever y is prescribed  is
zero. The function (x) is called y, the variation of y at a given x,  being a variational
operator. Variational operator  is in many ways similar to differential operator d and has similar
type of mathematical properties. However, they are conceptually different. Differential of a
function dy is a first order approximation to the change in function along a particular curve,

20
while the variational of a function y is a first order approximation to the change from curve to
curve.

Figure 2.1: Variation of a function

Figure 2.1 shows the plot of function y0 with solid line. Assume that the value of
function at one end point is prescribed, then a general function y0(x)+(x) is shown by a dotted
curve. If we put the general function in the functional, I will be obtained as a function of . The
condition for extremum of this function is,
dI
0 (2.9)
d
However, if y0 itself is extremum, then above condition should hold good at  = 0, i.e.,
dI
0 (2.10)
d  0

Replacing y by y0(x) + (x) in functional I,


I   F x, y 0   , y '0    dx 
b
(2.11)
a

where a dash in the superscript indicates differentiation with respect to x. At a fixed value of x,
one can write using Taylor’s theorem,
 F F   2 F    2F  2 F    (2.12)
F  x, y, y  F ( x, y 0 , y 0' )        2
2 2

      2   ...
 y y  y 2! yy y' 2!

where all the derivatives are evaluated at y0 and y'0 . Note that while expanding F by Taylor

series, we treat x as fixed and y and y' as two independent variables. Once x is fixed, y and

y' become variables instead of function and expression (2.12) is possible. Integrating the above
expression between a and b, and taking the derivative with respect to ,

21
 b F    2 F 2 2F  
  0    
dI   F  2F (2.13)
      2   2     2   ... dx
d a
  y y   y yy y  

Applying the condition for extremum, we get

0 
dI b  F F 
     dx (2.14)
d  0
a
 y y  

The dI is also called the first variation of I, I. Thus, the condition for extremizing a
d  0

functional is I=0. is Similarly, d I is called the second variation of I, 2I, which can tell if
2

d 2  0

the function is minimized, maximized or neither minimized nor maximized. Integrating the right
hand side of the above equation by parts, so as to reduce the order of derivative of , Eq. (2.14)
gets transformed to

 F d  F  
a  y  dx  y  dx  y 
F
b
b
(2.15)
a

Thus,
 F d  F  
I   
F F
     dx        0
b
(2.16)
 y dx  y   y b y a
a

If the value of y is prescribed at a boundary,  at that boundary will be zero as there is no

variation at that point. At other places  is arbitrary. We can also put in Eq. (2.16), an arbitrary
, which is 0 at both the boundaries. In that case,
x2  F d  F  
x 1
      dx  0
 y dx  y   
(2.17)

In view of the  being arbitrary, Eq. (2.17) implies that

F d  F 
  0 (2.18)
y dx  y 
Thus, extremization of the functional I requires the satisfaction of the above differential
equation. Substituting Eq. (2.18), in Eq. (2.16), we have

F F
   0 (2.19)
y b y a
At a particular boundary, say at point a, either  is zero or arbitrary and can be made zero. Thus,

F
 0 (2.20)
y b

22
In the same way, it can be shown that

F
 0 (2.21)
y a
Eqs. (2.20-2.21) are the two boundary conditions, which must be satisfied along with the
differential equation given by Eq. (2.18) for the extremization of the functional. Boundary
conditions imply that at the boundaries either  should be 0 i.e. the value of y should be

prescribed or F should be zero. The first type of boundary condition is called geometric or
y 

essential boundary condition, whilst the second type of boundary condition is called natural
boundary condition. The Eq. (2.18) is called Euler-Lagrangian equation.
Example 2.1: Find out the Euler-Lagrangian equation, which extremizes the following
functional:


I   y 2  y 2  2 xy dx 
1
(2.22)
0

Solution: Using Eq. (2.18), we have

 2 y  0 i.e., y  x  y  0

d
2 y  2x  (2.21)
dx
As
F
 2 y (2.22)
y
The boundary conditions are:
Either y' = 0 or y is prescribed at x = 0 and 1 (2.23)
Note that the variational form of the differential equation (2.21) does not depend on the
prescribed value of y at the boundaries.

If F depends on several dependent variable, i.e. F  F ( y1, y1' , y2 , y'2 ,......., yn , y'n , x )

where each yi = yi (x), the analysis proceeds as before, leading to n separate but simultaneous
equations for the yi(x),

F d  F 
    0, i  1,......, n. (2.24)
yi dx  y'i 
with corresponding boundary conditions. With n independent variables, we need to extremise
multiple integrals of the form
y y
I   ..... F ( y , ,...... , , x1 ,......., xn ) dx1.....dxn (2.25)
x1 xn

23
Using the same kind of analysis as before, we find that the extremising function y = y(x1,
………..,xn) must satisfy

F n   F 
  0 (2.26)
y i 1 xi  y x 
 i 
y
where y x stands for .
i
xi
We will now derive the necessary differential equation for extremizing
b
I   F ( x, y, y' , y'' ) dx . We can apply the procedure adopted for extremizing
a

b  dy 
I  a F  x, y,  dx , however, we will now follow a short but not so rigorous approach. We will
 dx 

now make use of the variational operator , and the fact that this operator behaves in the same
way as a differential operator. We also make use of the following two properties:
Property 1: Differentiation and variation commute i.e.,  y ' =  y' .

Proof: We can write,

d
  y ' =  y   ( )  
d d
(2.27)
dx dx dx
and

 dy  d dy dy d d y d
 ( y' )      ( y   )      (2.28)
 dx  dx dx dx dx dx dx
Both expressions are equal, hence proved.

Property 2: Integration and variation commute

i.e.  a y  x dx  a  y dx

b b

Proof:

 ab y  x  dx  ab  y   dx  ab  y  dx

 ab ydx  ab   dx  ab ydx

 ab   dx  ab  ydx

24
Now, using second property, we can write

 I  ab  F  x, y, y, y  dx (2.29)

The first variation of F is given by

 F  x, y, y, y   F  x, y   , y   , y   n   F  x, y, y, y  (2.30)

where  is a small quantity. The expansion using Taylor series leads to

F 
F
y
y
F
y
 y 
F
y 
 y   O  2   (2.31)

Noting that  y   etc. As   0 , we can write

F F F
F  y  y   y  (2.32)
y y y 
Thus,
b  F F F 
I    y  y '  y  dx (2.33)
a  y y ' y  
Making use of the property 1, we can write

b  F F d F d 2 
 I     y  ( y )  ( y )  dx
a  y y ' dx y '' dx 2 
(2.34)

Integrating the second and third terms by parts,

b  F d F d F d  F F d
b b
 I     y  ( ) y  ( ) ( y )  dx  y  ( y ) (2.35)
a  y dx y ' dx y '' dx  y ' a
y'' dx a

Integrating the third term by parts,

 F
b d  F  d 2  F   F b F b d F b
I   y    y     y dx +  y   y'   y (2.36)
 y
a dx  y'  dx 2  y ''   y ' a y'' a dx y '' a

At this point, note the point that if you had adopted the procedure of putting
y  y0 ( x)   ( x ) in the functional, then the condition for extremum would have been

0 
dI b  F F F 
      dx (2.37)
d  0
a
 y y  y  
Multiplying the integral in equation (2.37) by , we can see that this is same as equation (2.33).
Thus it is seen that the necessary condition for extremization is  I  0 . We can argue it in a

25
different way also. Assume that  I  0 . In that case it is possible to increase and decrease the
functional by giving a small variation. The extremum is reached when it is not possible to
increase or decrease the functional by giving a small variation. Note also that if a function can be
increased by giving a small variation, it can be decreased also by giving a variation in opposite
direction. Thus, the first variation must vanish for extremization.
Applying the necessary condition for extremization, we get

b  F d F d 2 F   F d  F   b F b
I   )   y dx   (2.38)
 ( ) 2 (     ( y ) a   y'  0

a y dx y ' dx y ''   y ' dx  y''   y'' a

Thus, the differential equation is

F d  F  d 2  F 
    0 (2.39)
y dx  y '  dx 2  y '' 
The boundary conditions are

 F d F 
  ( ) y  0 at x = a and at x = b (2.40)
 y ' dx y '' 
and
F
 y' 0 at x = a and at x = b (2.41)
y ''
The equation (2.40) suggests that at a boundary point, we can have either  y = 0 or the quantity
in curly bracket equal to 0. The former is called the essential (or geometric) boundary, whilst the
latter is called the natural boundary condition. Similarly in equation (2.41), the boundary
condition  y '  0 is called essential boundary condition and F / y = 0 is natural boundary
condition. By convention, the boundary conditions associated with the variational operator are
always called essential boundary condition and other boundary conditions are called natural
boundary conditions. Note that  y '  0 does not mean that the slope is zero. It means that the
slope has been prescribed.

Example 2.2: Total potential energy of a beam of length l is loaded by a load of intensity q(x) is
given by:


l 1

I    EIw ''2  qw  dx (2.42)
0 2 

26
where E and I are the Young’s modulus of elasticity and second moment of area about the
perpendicular to the plane of bending respectively and are a known function of x, and w is the
unknown beam deflection function. Find out the governing differential equation and boundary
condition.
Solution: Minimization of the total potential energy will lead to finding out the beam deflection
under load intensity q.
Eq. (2.39 ) gives

d2
q  0  ( EIw '')  0
dx 2
or

d2
( EIw '')  q (2.43)
dx 2
which is the well known Euler-Bernoulli beam equation.
The boundary conditions are
d
[0 ( EIw '')] w  0 at x=0, l (2.44)
dx

and
EIw ''  w '  0 at x = 0, l (2.45)

d
The term ( EIw '') represents the shear force and EIw'' the bending moment. We will not
dx
talk about the sign convention at this stage. Thus, at both the ends of the beam, two sets of the
boundary conditions need to be satisfied:
Set1: either the shear force is zero or the differential is prescribed.

Set2: either the bending moment is zero or the slope is prescribed.

The boundary conditions concerning the slope and deflection are called ‘geometric’ or
‘essential’ boundary conditions. Boundary conditions involving shear and bending moment are
called ‘natural’ or ‘force’ boundary conditions. Boundary conditions for three differently
supported beams are shown in Fig. 2.2.

27
 EIw''  0 EIw''  0
d
dx

EIw''  0  d
dx

EIw''  0
Free-free beam

w0 w0
EIw  0
''
EIw''  0

Simply-supported beam

d
w0 ( EIw'' )  0
dx
w'  0
EIw''  0
Cantilever beam

Figure 2.2: Boundary conditions for 3 different types of beams

2.4 OBTAINING THE VARIATIONAL FORM FROM A DIFFERENTIAL

EQUATION
Let us learn the procedure of converting a differential equation into a variational form.
Consider the differential equation

L ( )  f  0 (2.46)

where L is linear or non linear differential operator,  is a scalar function defined over the
domain D and f is a known scalar function. Multiplying the equation by a variation of  and
integrating it over the domain

 [ L ( ) 
D
f ]  d D  0 (2.47)

28
We keep on manipulating equation (2.47) using integration by parts till we are able to put it in
the variational form

  [ L * ( )  f  ]d D  0 (2.48)
D

Then, we say that the variational form of the given differential equation is

I ( )   [ L * ( ) 
D
f  ]d D (2.49)

In the process, the order of derivatives gets reduced. Generally, in a particular term, we attempt
to keep the order of derivative on  and on the assocaited expression same. The procedure will
be clear after seeing the Examples (2.3-2.5).

Example 2.3: The governing differential equation for a rod loaded with axial force is
d  du 
 EA q  0 (2.50)
dx  dx 
where E is Young’s modulus of elasticity, A is the cross-sectional area, q is the load intensity
(load per unit length in axial direction) and u is the axial displacement as a function of axial
coordinate x. Obtain the variational form of this equation. Assume that the boundary conditions
are
du
At x =0, u  0; at x =l , EA P (2.51)
dx
where l is the length of the rod.
Solution: As a first step, multiply the above governing differential equation by u and integrate
between 0 to l. Thus,


 d  du  
 dx  EA dx   q   u dx  0 (2.52)
l

0
   
Integrating equation (2.51) by parts,

  u qdx   d x  u  E A d x d x  0
du d du
l l l
 u EA  (2.53)
dx 0 0 0

As the variational operator behaves like a differential operator and u at x=0 is 0, we can write

1  du  
0  2  d x   0  ( q u )d x    E A d x u  l  0
 du 
l 2 l
  E A  d x  (2.54)
 

29
Therefore,
 l1  
   EA 
 du 
2

  q u  d x  P u (l )   0 (2.55)
  2
0
 dx   
Hence, the variational form is given by
1 

 du 
l 2

I   EA    q u  d x  P u (l ) (2.56)
0  2  dx  
The reader may observe that this is the total potential energy of the rod. Thus, the displacement
function of the rod will be one which extremizes the total potential energy amongst the functions
that satisfy the essential boundary condition i.e., u=0 at x=0.

Example 2.4: The heat conduction in a rod without heat generation is governed by
d 2T
k 0 (2.57)
dx 2
where k is the thermal conductivity and T is the temperature. Assume that the temperatures at the
two ends of the rod are prescribed. Obtaining the variational form of the problem.
Solution: Multiplying the differential equation by T and integrating between 0 and l (length of
the rod),


d 2T
k  T dx = 0 (2.58)
l

0 dx2
Integrating by parts

  T  k
dT d dT
l

k T  dx = 0 (2.59)
l

dx 0
0 dx dx
Making use of the fact that variation of the temperature at the ends is 0 and the properties of the
variational operator, we can write

   2 k
1 dT dT 
l

 dx  0 (2.60)
0
dx dx 
or

  k
 1 dT dT 
l

 dx  0 (2.61)
0  2 d x d x 

30
Hence, the variational form is given by

I   k
1  dT 
l 2

 dx (2.62)
0
2  dx 
Thus, for finding out the temperature distribution, extremize equation (2.62) while satisfying the
essential boundary condition at the ends. Unlike in the case of rod subjected to axial load, we
cannot assign any name to the functional I in equation (2.62).

Example 2.5: Steady-state heat conduction in a isotropic and homogeneous plate, in which the
temperature across thickness direction remains constant is governed by the following differential
equation:

 2T  2T
 0 (2.63)
x 2 y 2
Assume that the temperatures at the edges have been prescribed. Obtain the variational form of
the problem.
Solution: We first multiply the differential equation by T and integrate it over the domain.
Thus,

  2T  2T 
A  x 2  y 2   T dA  0

(2.64)

Now, we can integrate this equation by parts to reduce the order of derivative; however, it is
better to make use of divergence theorem. Thus, the equation (2.64) can be written as

 I    2T  T dA   .  T  T  dA   T .  T  dA  0 (2.65)
A A A

Applying divergence theorem,

 I    T .nˆ   Td   T .  T  dA  0 (2.66)

 A

where  is the boundary of the domain. As the temperature is prescribed in the boundary, T
becomes 0 there. Thus, equation (2.66) becomes
 T  T T  T
 I    T .  T  dA    

  dA  0 (2.67)
A A  x x y y 
Using the properties of the variational operator, we can write

1  T   T  
 I           dA  0
2 2

(2.68)
2  x   y  
A  

31
Thus, the variational form is

1  T   T  
I    
2 2

    dA (2.69)
A
2  x   y  

2.5 PRINCIPLE OF VIRTUAL WORK

We obtained the variational form of the rod subjected to axial loading by converting the
differential equation into the variational form. We could have obtained it using the principle of
virtual work. According to it, when we subject a loaded body in equilibrium to small compatible
virtual displacements (which do not violate the essential boundary conditions), the total internal
virtual work is equal to the total external virtual work. The internal virtual work per unit volume
will be equal to the product of real stresses and virtual strain, which can be integrated to yield
the total internal virtual work. Note that virtual displacement are imaginary small displacements,
which when applied do not cause any change in the forces are stresses. During the application of
virtual displacements, the stresses can be considered constant. That is why we are able to take
internal virtual work per unit volume equal to the product of real stresses and virtual strain. The
virtual strain are found by taking the derivatives of the virtual displacement function. Therefore,
the virtual displacement function must be differentiable. The external work will be the
summation of works done by all forces acting on the body when subjected to virtual
displacements.
If the rod of Example (2.3) is subjected to virtual displacement u, the internal work will be

  u  EA dx
d du
Wint 
l
(2.63)
0 dx dx
and the external work will be

  u q d x  P  u (l )
l
W ex t  (2.64)
0

Applying the principle of virtual work

Wint  Wext    u  EA dx    u qdx  P  u (l )  0

d du
l
l
(2.64)
0 dx dx 0

The equation (2.64) is called integral form of the problem. If we treat  as the variational
operator, we can easily obtain equation (2.56) i.e., the variational form of the problem.

32
2.6 PRINCIPLE OF MINIMUM POTENTIAL ENERGY
In the process of obtaining the variational form of the rod problem, we have obtained a
functional whose extremization alongwith the satisfaction of essential boundary condition will
provide us the solution. We indicated that this functional is actually the total potential energy of
the rod composed of (a) the strain energy of elastic distortion, and (b) the potential possessed by
applied loads. We could have written this functional using the principle of minimum potential
energy for the stable conservative mechanical system. For a conservative mechanical system,
one can express the energy content of the system in terms of its configuration, without reference
to whatever deformation history or path may have led to the configuration.
The statement of principle of minimum potential energy is as follows. Among all
possible configurations of a conservative system satisfying internal compatibility and essential
boundary conditions, those that keep the body in stable equilibrium make the potential energy
minimum with respect to small admissible variations of displacement. This principle is
applicable evenif the material behavior is non-linear.
The reader may have a doubt how one can know that the functional in equation (2.56) is
to be minimized and not maximized, because the functional was obtained by putting the
necessary condition for extremization. For ascertaining that the functional has to be minimized,
one needs to calculate second variation 2I , which we have not done in this chapter. However,
often the form of the functional and the physics of the problem can provide the answer. For
example, we can see that equation (2.56) is unbounded for maximization problem. We can
choose the function u=-M, where M is a large positive number and can make the functional as
large as we wish. Thus, there is no physically realistic solution for a maximization problem.
Therefore, physically realistic solution should correspond to minimization problem.

2.7 CONCLUSIONS
In this chapter, we have briefly introduced calculus of variation. We can model a physical
problem in the form of differential equations or in the form of integral. The differential form is
called strong form, because it contains the higher order derivatives, whilst the integral form
containing lower order derivatives is called the weak form. In most of the physical problems, it
is possible to convert one form into the other. The integral form can also be obtained using
principle of virtual work or principle of minimum potential energy. These principles have been
developed for mechanical systems also. Thus, knowledge of calculus of variation will enable
you to obtain the variational form for other physical problems too.

33
REFERENCES (for further reading, not cited in the text)

1. J.N. Reddy, An Introduction to the Finite Element Method, McGraw-Hill, New York,
1993.
2. R.D. Cook, D.S. Malkus and M.E. Plesha, Concepts and Applications of Finite Element
Analysis, 3rd ed., John Wiley, New York 1989.
3. K.J. Bathe, Finite Element Procedures in Engineering Analysis, Prentice-Hall, Englewood
Cliffs, NJ 1982.
4. K.F. Riley, M.P. Hobson and S.J. Bence, Mathematical Methods for Physics and
Engineering, Cambridge University Press, Cambridge, 1998.

EXERCISE 2 (Submit it by August 18, 2025)

Q.1: The functional governing static buckling of the column in the figure shown below is

 d 2w 
 0  d x  d x + 2 kw L
1 P  dw  1
L 2 L 2

  EI  2 
dx  2
(a)
2 0  dx  2

dw
where w  0,  0 (b)
x0
dx x0

Invoke the stationary conditions    0 to derive the problem-governing differential

equation and the natural boundary conditions.

Figure: Q1
Q.2: Obtain the variational form of heat transfer problem in a rod.

Assume A, k, r are constants.

34
 Figure: Q2
Differential equation:

d 2T r
  0
dx 2
kA
Boundary conditions:

1. Essential boundary conditions: T = 0 at x = 0

dT T
2. Natural boundary conditions  1 at x = 1
dx 2

Q.3: A certain physical problem has the functional

  2 
1 
L
  2
 5 0  d x

,x
0

Essential boundary conditions are   0 at x = 0 and x = L. Express  as a function of x, which

extremizes the functional.

Q.4: The potential energy of an isotropic plate that carries lateral pressure q is

 
D  q
 ( w, xx  w, yy )  2(1   )( w, xx w, yy  w, xy )  2  dx dy
2 2
p
2  D
where D is a constant called flexural rigidity. Show that the governing differential equation
is  4  q / D , where  4 is the bi-harmonic operator.

Q.5: Steady state heat conduction without heat generation is governed by the following
differential equation (for domain with constant thermal conductivity):

 2T  2T
 0
x 2 y 2

35
If the temperatures at the boundary are specified, obtain the variational form of the above
equation.

Q.6: Consider the following boundary value problem

d2u
u 1 0  x 1
dx 2
du
with u(0) = 0 and = 0 at x = 1. Convert this problem into variational form.
dx

Q.7: The bending of a beam is governed by the following differential equation

d2 d 2 v dm z
( EI )  py  0
dx 2 dx 2 dx
zz

dv
with essential boundary conditions at x = 0 are v = v* and   z* and natural boundary
dx
d2 v d d2 v
conditions at x = l are EI zz 2  M z and
*
( EI zz 2 )  mz  Vy* . Where, mz is the
dx dx dx
distributed moment per unit length about z-axis and p y is the distributed force per unit length in

y-direction. Obtain the variational functional.

Q.8: Using calculus of variations, show that the shortest curve joining two points is a straight
line.

Q.9: frictionless wire in a vertical plane connects two points A and B, A being higher than B.
Let the position of A be fixed at the origin of an xy-coordinate system, but allow the B to lie
anywhere on the vertical line x = x0. Find the shape of the wire such that a bead of mass m placed
on it at A will slide under gravity to B in the shortest possible time.

Q.10: Consider a mechanical system whose configuration can be uniquely defined by a number
of coordinates qi (usually distances and angles) together with time t, and which only experiences

forces derivable from a potential. Hamilton’s principle states that in moving from one
configuration at time t1 the motion of such a system is such as to make stationary.

L   L(q1 ,..., qn , q1 ,..., q n , t ) dt

t1

t0

36
The Lagrangian L is defined in terms of the kinetic energy T and the potential energy V (with
respect to some reference situation) by L = T-V. Here V is a function of the qi only, not of the

qi . Applying the EL equation to L, we obtain Lagrange’s equations,

L d  L 
  , i  1,..., n.
qi dt  q i 
Using Hamilton’s principle, derive the wave equation for small transverse oscillations of a taut
string.

37