CLASSICAL MECHANICS

Unit-2

Lagrange’s Equations
And
Applications
Recall:
virtual displacement δri  change in the configuration of the system as
a result of an infinitesimal change in coordinates at instant t

If the virtual work of the forces of constraints in zero, then:

∑ i − p& i ) ⋅ δri = 0
a
( F (D’Alembert’s principle) (1)
i

So forces of constraint no longer appear; we can drop superscript a

To be in useful form  transform into expression involving virtual

displacements of generalised coordinates, independent of each other
(for holonomic constraints), so that coefficients of δqi can be set
separately equal to zero.

Principles of Lagrange’s Equations

For transformation of ri to qi, start with transformation equation:
ri = ri ( q1 , q2 , q3 ,....., qn , t )
(assuming n independent coordinates)
Then using chain rule of partial differentiation, v i ≡ r&i is expressed in
terms of q&i :
dri ∂r ∂r
vi = = ∑ i q& j + i (2)
dt j ∂q j ∂t
∂ri
Virtual displacement δri is connected with δqj: δri = δq j
∂q j
(here virtual displacement means, displacement of coordinates only)

Differentiating equation (2) w.r.t. q& j and q j :

∂v i ∂ri
= (3)
∂q& j ∂q j
∂v i ∂ 2ri ∂ 2ri d ∂ri
and =∑ q& k + = (4)
∂q j k ∂q j ∂qk ∂q j ∂t dt ∂q j
Now 1st term in the equation (1) stating D'Alembert's principle:
∂ri
∑ Fi ⋅ δri = ∑ Fi ⋅ δq j
i i ∂q j
= ∑ Q jδq j (5)
j
Qj  components of generalised force,
∂ri
Q j = ∑ Fi ⋅ (6)
i ∂q j
As q’s may not have dimensions of length, Q’s do not necessarily have
dimensions of force; but Q jδq j must have dimensions of work.
Now the 2nd term of equation (1):
dp i ∂ri
∑ p& i ⋅ δri = ∑ ⋅ δq j
i i , j dt ∂q j
Re-writing:
dp i ∂ri
∑ p& i ⋅ δri = ∑ ⋅ δq j (6-a)
i i , j dt ∂q j

d  ∂ri  dp i ∂ri d ∂ri

as
 ∑ pi .  = ∑ . + ∑ pi ⋅
dt  i , j ∂q j  i , j dt ∂q j i , j dt ∂q j

or rearranging the terms:

dp i ∂ri d  ∂ri  d ∂ri
∑ . =  ∑ pi .  − ∑ pi ⋅
i , j dt ∂q j dt  i , j ∂q j  i , j dt ∂q j
Substituting in Eq. (6-a):

d  ∂ri   d ∂ri 
so ∑ p& i ⋅ δri = ∑  ∑ p i ⋅ δq j − ∑ p i ⋅  δq j

i j dt  i ∂q j  i, j  dt ∂q j 
 Re-writing previous equation:

d  ∂ri   d ∂ri 
∑ p& i ⋅ δri = ∑  ∑ p i ⋅ δq j − ∑ p i ⋅  δq j

i j dt  i ∂q j  i, j  dt ∂q j 
Using eqns. (3) and (4):

d  ∂v i  ∂v i
∑ p& i ⋅ δri = ∑  ∑ p i ⋅ δq j − ∑ p i ⋅ δq j
i j dt  i ∂q& j  i, j ∂q j

d  ∂v i  ∂v i
= ∑  ∑ mi v i ⋅ δq j − ∑ mi v i ⋅ δq j
j dt  i ∂q& j  i, j ∂q j
d  ∂  1 2  ∂  1 2

= ∑   ∑ 2 mi vi  −  ∑ 2 mi vi δq j
j  dt  ∂q
&j  i  ∂q j  i 

As system kinetic energy: T = ∑ 12 mi vi2
i

 d  ∂T  ∂T 
therefore ∑ p& i ⋅ δri =∑    − δq j (7)
i  dt  ∂q& j  ∂q j 
j 

So using (5) and (7), D’Alemberts principle (1) takes the form:

 d  ∂T  ∂T  
∑   −
  − Q j δq j = 0 (8)
j   dt  ∂q
& j  ∂q j  

We assumed we had a holonomic system and the q's were all
independent, so every virtual displacement δqj is independent of δqk
every coefficient should vanish separately.
d  ∂T  ∂T
so
  − = Qj (n such equations) (9)
dt  ∂q& j  ∂q j

These equations are often referred to as Lagrange’s equations for

dynamical systems.
Lagrange’s Equations for Conservative Systems
Conservative forces are derivable from a scalar potential V: Fi = −∇ iV
Using equation (6), the generalised forces can be written as:
∂ri ∂r
Q j = ∑ Fi ⋅ = −∑ ∇ iV ⋅ i
i ∂q j i ∂q j
The RHS is the same as partial derivative of a function − V (r1 , r2 ,.....rn )
w.r.t. q j :
∂V ∂q j ∂V
Such that: ∇ iV = =
∂q j ∂ri ∂ri
∂V
Therefore: Qj = −
∂q j
Then Lagrange’s equations (9) can be written as:

d  ∂T  ∂T ∂V
  − =−
dt  ∂q& j  ∂q j ∂q j
 d  ∂T  ∂ (T − V )
or
  − =0 (10)
dt  ∂q& j  ∂q j

As V does not depend on q& j ,we can include V in partial derivative with
respect to q& j
d  ∂ (T − V )  ∂ (T − V )
  − =0
dt  ∂q& j  ∂q j
Defining a new function, Lagrangian L: L = T −V
d  ∂ ( L)  ∂ ( L)
  − =0 (11)
dt  ∂q& j  ∂q j

These are Lagrange's equations, which we have now derived in the

more general context of constrained systems.
Goal of Lagrangian formulation  to eliminate forces of constraints

Other benefits:
Newtonian mechanics  setting up equations of motion, like

∑ F ji +Fi( e) = p& i
j

 necessary to work with many vector forces

and accelerations
Lagrangian method  only two scalar functions, T and V
Simple Applications of Lagrangian Formulation
Lagrangian method  two scalar functions, T and V
 write T and V
 substitute in Lagrange’s equations (9) or (11)
Examples
1. Motion of one particle: using cartesian coordinates

Generalised forces are given by: Fx , Fy , Fz

Velocities are: v x = x& v y = y& v z = z&

K.E. T = 12 m( x& 2 + y& 2 + z& 2 ) (12)

By partial differentiation of (12) w.r.t. x, y and z :

∂T ∂T ∂T
= = =0
∂x ∂y ∂z
Now by partial differentiation of (12) w.r.t. x& , y& and z& :
∂T ∂T ∂T
= mx& = my& = mz&
∂x& ∂y& ∂z&
d  ∂T  ∂T
for this problem use: − = Qj
 
dt  ∂q& j  ∂q j

Making substitutions: d
(mx& ) = Fx
dt
d
(my& ) = Fy
dt
d
(mz& ) = Fz
dt
Which are simply Newton’s equations of motion.
2. Motion of one particle: using polar coordinates

To express T in terms of r& and θ& , use then transformation equations:

x = r cosθ
and y = r sin θ y

Differentiating w.r.t. t:
dx dr d P
= (cosθ ) + r (cosθ )
dt dt dt
r
dy dr d
= (sin θ ) + r (sin θ )
dt dt dt θ
x
or x& = r& cosθ − rθ& sin θ
(13)
y& = r& sin θ + rθ& cosθ
Now K.E: T = 12 m( x& 2 + y& 2 ) (14)
Substituting from (13) in (14):

T= 1m
2
[
(r& cosθ − rθ sin θ ) + (r& sin θ + rθ cosθ )
& 2
& 2
]
( )
= 12 m r& 2 cos 2 θ + r 2θ& 2 sin 2 θ − 2r&θ& sin θ cosθ +
(
+ r& 2 sin 2 θ + r 2θ& 2 cos 2 θ + 2r&θ& sin θ cosθ )
T = 12 m(r& 2 + r 2θ& 2 ) (15)

Plane polar coordinates of velocity are r& along direction of r and rθ&
along direction ⊥ r denoted by unit vector n

Using equation (6), ∂ri

Q j = ∑ Fi ⋅
i ∂q j
the two components of generalised force can be obtained:
∂r r
Qr = F. = F. = Fr
∂r r
 ∂r
and Qθ = F. = F.rn = rFθ
∂θ r∆θ n
Because, derivative of r w.r.t. θ is a vector in r(θ+∆θ )
direction of n.
∆θ
Two generalised coordinates, r and θ ⇒ two r(θ )

Lagrange’s equations
d  ∂T  ∂T
For r:  − = Fr (16)
dt  ∂r&  ∂r
So differentiating equation (15) T = 12 m( r& 2 + r 2θ& 2 )
respectively w.r.t. r and r& :
∂T ∂T
= mrθ & 2
= mr&
∂r ∂r&
d  ∂T 
then   = m&r&
dt  ∂r& 
Equation (16) becomes: m&r& − mrθ& 2 = Fr (17)
 Now the Lagrange’s equation for θ :
d  ∂T  ∂T
 &− = rFθ (18)
dt  ∂θ  ∂θ
∂T ∂T
We get from (15): = 0, = mr 2θ&,
∂θ ∂θ&
d  ∂T 
 &  = 2mrr&θ& + mr θ&&
2
and
dt  ∂θ 
So the second Lagrange’s equation (18) is:

2mrr&θ& + mr 2θ&& = rFθ (19)

Equations (17) and (19) are equations of motion of single particle in polar
coordinates.
3. Atwood’s machine:
A conservative system with holonomic,
scleronomous constraint.
Two masses m1 and m2, connected by a string x
of length l, passing over a frictionless and l-x
massless pulley.
M1

Only independent coordinate  x M2

Only constraint  l, length of rope between two masses

Potential energy of the system:

V = − M 1 gx − M 2 g (l − x) (20)

Two masses move with velocity of same magnitude: = x&

So K.E. of the system: T = 12 ( M 1 + M 2 ) x& 2 (21)

Lagrange’s equation for the system:

d  ∂ ( L)  ∂ ( L)
 − =0 (22)
dt  ∂x&  ∂x

From equations (20) and (21), the Lagrangian has the form:

L = T − V = 12 ( M 1 + M 2 ) x& 2 + M 1 gx + M 2 g (l − x)
∂L
∴ = M1g − M 2 g = (M1 − M 2 ) g
∂x
∂L
= ( M 1 + M 2 ) x&
∂x&
d  ∂L 
and   = ( M 1 + M 2 ) &x&
dt  ∂x& 
Making substitutions in (22):
( M 1 + M 2 ) &x& = ( M 1 − M 2 ) g
(M1 − M 2 )
&x& = g
(M1 + M 2 )
Tension in the rope is eliminated using Lagrangian method.

4. A bead sliding on a uniformly rotating wire in a force-free space:

Straight wire rotating uniformly about fixed axis ⊥ wire
Constraint is time dependent.
Using plane polar coordinates as generalised coordinates, the
transformation equations:
x = r cosθ
y = r sin θ
Angular frequency of uniform rotation = ω
then θ = ωt
 ∴ x = r cos ωt
y = r sin ωt
Differentiating w.r.t. t
x& = r& cos ωt − rω sin ωt
y& = r& sin ωt + rω cos ωt
The K.E. is then: T = 12 m( x& 2 + y& 2 )
Making substitutions: T = 12 m(r& 2 + r 2ω 2 ) (23)

Free space ⇒ potential energy, V = 0

Therefore, Lagrangian: L = T −V = T
Using Lagrange’s equation:

d  ∂ ( L)  ∂ ( L)
 − =0
dt  ∂r&  ∂r
So differentiating Eq. (23):

∂L ∂L
= mrω 2 = mr&
∂r ∂r&
d
Then Lagrange’s equation: (mr& ) − mrω 2 = 0
dt

or m&r& − mrω 2 = 0

&r& = rω 2
The bead moves outward because of centripetal acceleration.
5. Consider a case of projectile motion under gravity in two dimensions.
Find the equations of motion in (i) cartesian and (ii) polar coordinates
using Lagrange’s equation.
y

mg

θ
x

(i) Cartesian coordinates  generalised coordinate x, y:

1 1
Kinetic energy: T = mx& 2 + my& 2
2 2
Potential energy: V = mgy (V = 0 when y = 0)

1 2 1 2
The Lagrangian: L = T − V = mx& + my& − mgy
2 2
d  ∂ ( L)  ∂ ( L)
Lagrange’s equation for x - coordinate:  − =0
dt  ∂x&  ∂x
 d ∂L d
so = mx& = m&x&
dt ∂x& dt
∂L
and =0
∂x

Therefore, the equation of motion: m&x& = 0

&x& = 0
Similarly Lagrange’s equation for y - coordinate:
d  ∂ ( L)  ∂ ( L)
 − =0
dt  ∂y&  ∂y
d
so my& + mg = 0
dt
m&y& + mg = 0
&y& = − g
(ii) Polar coordinates  generalised coordinate r, θ :

Transformation equations:
x = r cosθ
and y = r sin θ
Differentiating w.r.t. t:
x& = r& cosθ − rθ& sin θ
y& = r& sin θ + rθ& cosθ
1 2 1
Kinetic energy: T = mr& + m(rθ&) 2
2 2
Potential energy: V = mgr sin θ
1 2 1 2 &2
so L = T − V = mr& + mr θ − mgr sin θ
2 2
Lagrange’s equation for r - coordinate:
d  ∂L  ∂L
 − =0
dt  ∂r&  ∂r
d
(mr&) − mrθ& 2 + mg sin θ = 0
dt
&r& − rθ& 2 + g sin θ = 0
for θ - coordinate:
d  ∂L  ∂L
 &− =0
dt  ∂θ  ∂θ
d
(mr 2θ&) + mgr cosθ = 0
dt
2mrr&θ& + mr 2θ&& + mgr cosθ = 0
r 2θ&& + 2rr&θ& + gr cosθ = 0
(Cartesian coordinates preferable)