Classical Mechanics

Lecture 3
Syllabus
 Why do we need a new formulation?

• Newtonian formulation deals with forces (a vector quantity).

• A force field is given by three functions

• Fx , Fy , Fz .

• All fundamental forces are conservative.

• A conservative force field is given by a single function :

• So all information we need is contained in one function, instead of

three.
• Using the vector force field is an overkill.
• There are must be a simpler way to formulate the Newtonian
mechanics for the conservative forces.
 Why do we need a new formulation?

• We want to reformulate the Newtonian mechanics in such a way as to

avoid using forces and use only the potential energy

• Not invariant when we change the coordinate system:

• Cartesian: Cylindrical:

• Too complicated, too tedious. Consider two pendulums.

• Difficult to find conservation laws.

• There is more
Symmetries are hard to notice in vector formulation.
The idea of force cannot be used in quantum mechanics.
The idea of force cannot be used in relativistic physics.
 Why do we need a new formulation?

• The motion of a charged particle in an electromagnetic field experience

Lorentz force and the Equation motion is

• Or

• Where , ,
 Why do we need a new formulation?
• In cartesian coordinates we have eqn of motion

• In cylindrical coordinates , velocity vector is

and the eqn of motion is
 Why do we need a new formulation?

• We see that the form in which our space variables enter into the
equations depends on the coordinate system we choose.
• i.e. we cannot write down a generalized form of scalar equation which
applies to every component equation in any given coordinate system.
• This shortcoming of the Newtonian form of the equation of motion is
avoided in the Lagrangian–Hamiltonian formalism, where generalized
coordinates and generalized potentials are introduced
 D’Alembert’s Principle & Lagrange’s Equations
• Virtual (infinitesimal) displacement ≡ Change in the system configuration as
the result of an arbitrary infinitesimal change of coordinates δri, consistent
with the forces & constraints imposed on the system at the given instant of
time t.
• The key point in this formulation: imagining that a system of bodies in a
given configuration undergo small displacements away from their assumed
positions, calculating the resultant work done on the system and then
demanding that it sum to zero

• The displacements are imagined— bodies actually in equilibrium obviously

do not undergo any real displacements—and they are assumed to be of
infinitesimal extent. They are assumed to take place instantaneously (i.e., δt
= 0), and they are assumed to take place in a way that is consistent with any
imposed constraints.
 D’Alembert’s Principle & Lagrange’s Equations
• The key point in this formulation is if the work vanishes for a system
subjected to such virtual displacements, then the system is in equilibrium.
• Thus, we have the conditions necessary for equilibrium by dealing with the
scalar quantity of work, rather than the vector quantity of force.

• “Virtual” distinguishes it from an actual displacement dri, occurring in

small time interval dt (during which forces & constraints may change)
 D’Alembert’s Principle & Lagrange’s Equations
• Virtual work : δW =
• However, problems in dynamics deal with bodies that are not in
equilibrium. They must be solved using Newton's second law of motion

• D'Alembert's insight was to realize that problems in dynamics could be cast

in the same language as the principle of virtual work by including the
inertial term —p as a real force

• D'Alembert's principle, is equivalent to Newton's second law of motion, by

taking the inertial term as a real force

• In effect, D'Alembert reduced a problem in dynamics to one in statics

 D’Alembert’s Principle & Lagrange’s Equations contd…
• Let us consider the system at equilibrium:
• i.e. The total force on each particle is Fi = 0.

• Virtual work done by Fi in displacement δri: δWi = Fi∙δri = 0.

• Sum over i: ⇒ δW = …………. (1)

• Decompose Fi into applied force Fi(a) & constraint force fi:

…………. (2)

⇒ δW = ≡ δW(a) + δW(c) = 0
…………. (3)
 D’Alembert’s Principle & Lagrange’s Equations contd…
Special case (often true, see Goldstein discussion): we shall restrict to Systems
for which the net virtual work due to constraint forces is zero:
∑ifi∙δri ≡ δW(c) = 0 …………. (4)

Note - if a particle is constrained to move on a surface, the force of constraint is

perpendicular to the surface, while the virtual displacement must be tangent to
it, and hence the virtual work vanishes.(this is true for rolling friction cases but
not valid for sliding friction)

Using (4) in eqn (3)

⇒ Condition for system equilibrium: Virtual work due to APPLIED forces
vanishes:
(a)
δW = ≡ Principle of Virtual Work

…………. (5)
 D’Alembert’s Principle & Lagrange’s Equations contd…
• Note: In general coefficients of δri , Fi(a) ≠ 0 even though ∑iFi(a)∙δri
= 0 because δri are not independent, but connected by constraints.
– In order to have coefficients of δri = 0, we must transform
Principle of Virtual Work into a form involving virtual
displacements of generalized coordinates qi , which are
independent.
– Eqn (5) is good since it does not involve constraint forces fi . But so
far, only statics. Want to treat dynamics!
 D’Alembert’s Principle & Lagrange’s Equations contd…
• Dynamics: Start with Newton’s 2nd Law for particle i:
• Fi = = (dpi/dt)
• Or:
…………. (6)

⇒ the particles in the system will be in “equilibrium” under a force

= actual force + “reversed effective force” = - = - (dp/dt)
• Virtual work done is
δW = …………. (7)

• Again decompose Fi into applied forces and forces of constraint:

Fi = Fi(a) + fi
⇒ δW =

…………. (8)
 D’Alembert’s Principle & Lagrange’s Equations contd…
• Again we restrict consideration to special case: Systems for which the net
virtual work due to constraint forces is zero: ∑i fi∙δri ≡ δW(c) = 0

• ⇒ δW =
…………. (9)
• ⇒ δW = ∑i[Fi - (dpi/dt)]∙δri = 0
• ≡ D’Alembert’s Principle

– Dropped the superscript (a)!

• In the present form eqn (9) does not lead to eqns of motion of the system.
• We transform (9) to an expression involving virtual displacements of q
(which, for holonomic constraints, are independent of each other). Then,
by linear independence, the coefficients of the δq = 0
 D’Alembert’s Principle & Lagrange’s Equations contd…

• Translation from ri to qj involve several steps

• Much manipulation follows! Only highlights here!
• Transformation eqns:
(i = 1,2,3,…n)

• Differntiating with respect to t by Chain rule of differentiation, (velocities):

• or

…………. (10)

• Where = generalized velocities

 D’Alembert’s Principle & Lagrange’s Equations contd…

• Virtual displacements δri are connected to virtual displacements δq:

• Or

…………. (11)

• by definition virtual displacement is time independent

• In terms of the generalized coordinates, the virtual work of the Fi becomes

[using (11)]

…………. (12)

Where Qj = Generalized Force

 D’Alembert’s Principle & Lagrange’s Equations contd…

• We define Generalized Force (corresponding to Generalized Coordinate

qj):
…………. (13)

– Generalized Coordinates qj need not have units of length!

⇒ Corresponding Generalized Forces Qj need not have units of force!
But must always have the dimensions of work.

– For example: If qj is an angle (such that the differential dqj ≡

differential angle dθj), corresponding Qj will be a torque ( such that
Qj ≡ Nj) then the product Nj dθj = differential of work
 D’Alembert’s Principle & Lagrange’s Equations contd…
• 2nd term of De-Alembert’s principle eqn (9) (using eqn (11) again):

…………. (14)

• From eqn (14), we shall write the coefficient of δqj as :

…………. (15)
• Also we have

…………. (16)
 D’Alembert’s Principle & Lagrange’s Equations contd…

• Using eqn (10), eqn(16) can be written as

…………. (17)

• From eqn s(10), we also find

…………. (18)

• Substituting eqns (17) and (18), eqn (10) becomes

…………. (19)
 D’Alembert’s Principle & Lagrange’s Equations contd…
• Using eqn (12), (14) and (19) , D’Alembert’s Principle, eqn (9) becomes
∑i[Fi - (dpi/dt)]∙δri = 0

⇒ - =0

⇒ - =0

⇒ =0

…………. (20)

Where system kinetic energy T =

 D’Alembert’s Principle & Lagrange’s Equations contd…

• Thus D’Alembert’s Principle becomes

…………. (21)

• Note: If qj are Cartesian coords, (∂T/∂qj) = 0

• ⇒ In generalized coords, (∂T/∂qj) comes from the curvature of the qj.
(Example: Polar coords, (∂T/∂θ) becomes the centripetal acceleration).

• So far, no restriction on constraints except that they do no work under

virtual displacement. qj are any set.
•
• Special case: Holonomic Constraints ⇒ It’s possible to find sets of qj for
which each δqj is independent.
• ⇒ Each term in (21) is separately 0!
 D’Alembert’s Principle & Lagrange’s Equations contd…

• So Holonomic constraints ⇒ D’Alembert’s Principle:

(j = 1,2,3, … n)

• i.e There are n such equations in all.

…………. (22)
• Special case: When the forces are derivable from a scalar potential function V
i.e. A Potential Exists ⇒

• Needn’t be conservative! V could be a function of t!

⇒ Generalized forces have the form

…………. (23)
 D’Alembert’s Principle & Lagrange’s Equations contd…

• Putting (23) in (22):

• So far, V doesn’t depend on the velocities

⇒
 D’Alembert’s Principle & Lagrange’s Equations contd…

• We Define: The Lagrangian L of the system:

L≡T-V

⇒ We can write D’Alembert’s Principle as:

(j = 1,2,3, … n)

≡ Lagrange’s Equations
 D’Alembert’s Principle & Lagrange’s Equations contd…

• Note: L is not unique, but is arbitrary to within the addition of a derivative

(dF/dt). F = F(q,t) is any differentiable function of q’s & t.

• i.e. , For the Lagrangians of the type and functions of the

type , we define a new Lagrangian L´.
Let’s introduce a transformation (gauge transformation or Gauge invariance):

It is easy to show that L´ satisfies the same Lagrange’s Equations (above).

 Gauge invariance

• we have , ,

• Now

• Further
 Gauge invariance

• now

[ If we write

the corresponding
terms cancel out]
 Gauge invariance

• So
 Velocity-Dependent Potentials & the Dissipation Function

• Non-conservative forces i.e. if there is no potential function, V ?

• It’s still possible, in a Special Case, to use Lagrange’s Equations unchanged,
provided a Generalized or Velocity-Dependent Potential
exists, where the generalized forces Qj are obtained as:

• The Lagrangian is now: L ≡ T - U & Lagrange’s Equations are still:

(j = 1,2,3, … n)

• A very important application: Electromagnetic forces on moving charges.

 Velocity-Dependent Potentials & the Dissipation Function
• Let us consider a charged Particle of mass m, charge q moving with velocity
v in combined electric (E) & magnetic (B) fields.
• Lorentz Force (SI units!):
…………. (1)

• E&M results that we should know!

Both & are continuous functions of
time and position and derivable from a scalar potential
and a vector potential as:

…………. (2)

…………. (3)
 Velocity-Dependent Potentials & the Dissipation Function

• We have

• Using (2) and (3)

• We shall use the vector identity (Vector triple product)

• Hence
 Velocity-Dependent Potentials & the Dissipation Function

• So

• Now we can write

• Hence, we can write

 Velocity-Dependent Potentials & the Dissipation Function

• So

…………. (4)

• Thus we Can obtain the Lorentz Force (1) from the velocity dependent
U ≡ qφ - qA∙v
potential: …………. (5)
and F = - ∇U

• So the Lagrangian is:

L≡T-U
…………. (6)
 Velocity-Dependent Potentials & the Dissipation Function

• Let us use Cartesian coordinates.

• Lagrange Equation for coordinate x (noting
& ) is

• The x-component of Lagrange’s eqn

• The total time derivative of Ax …………. (7)

…………. (8)
 Velocity-Dependent Potentials & the Dissipation Function

• So

• From eqns (2) and (3) …………. (9)

• Ex= - -
…………. (10)
• and

…………. (11)
 Velocity-Dependent Potentials & the Dissipation Function

• Combining eqns (9), (10) and (11), we have

• = Fx

…………. (12)
• Eqns (1) and (12) are identical ⇒ Lorentz force equation is derivable from
scalar and vector potenials.

• If some forces in the problem are conservative & some are not (i.e not
derivable from a potential): ⇒ Have potential V for conservative ones &
thus have the Lagrangian L ≡ T - V for these.
• For non-conservative ones, still have generalized forces:
•
 Velocity-Dependent Potentials & the Dissipation Function

• For such cases, Lagrange’s equations can always be written as

(j = 1,2,3, ..n)

• Where L ≡ T - V for conservative forces and

Generalized forces: Qj ≡ ∑iFi∙(∂ri/∂qj) for non-conservative forces.

• Such kind of problem arises when frictional forces are present

• Friction: A common non-conservative force.
• Friction (or air resistance): A common model: frictional force components are
proportional to some power of velocity v of the particle (often the 1st power) :
•

• (kx = const)
 Velocity-Dependent Potentials & the Dissipation Function

• Frictional forces of this type may be derived in terms of a function , known

as Rayleigh’s dissipation function, and defined as

• Obtain components of the frictional force by:

etc. Or,

• Physical Interpretation of ₣ : Work done by system against friction:

• = -2 dt
⇒ Rate of energy dissipation due to friction:
(dWf /dt) = -2
 Velocity-Dependent Potentials & the Dissipation Function

• The component of the generalized force resulting from the force of friction
is then given by

(Since )

• Lagrange’s Equations, with frictional (dissipative) forces:

(j = 1,2,3, ..n)
Velocity-Dependent Potentials & the Dissipation Function
 Simple Applications of the Lagrangian Formulation

• Lagrangian formulation: 2 scalar functions, T & V

• Newtonian formulation: MANY vector forces & accelerations.
(Advantage of Lagrangian over Newtonian!)

• “Recipe” for application of the Lagrangian method:

– Choose appropriate generalized coordinates
– Write T & V in terms of these coordinates
– Form the Lagrangian L = T - V
– Apply: Lagrange’s Eqtns:

(j = 1,2,3, … n)

– Equivalently D’Alembert’s Principle:

(j = 1,2,3, … n)
• Sometimes, T & V are easily obtained in generalized coordinates qj &
velocities qj & sometimes not.
• If not, write in Cartesian coordinates & transform to generalized
coordinates. Use transformation eqtns:
ri = ri (q1,q2,q3,.,t) (i = 1,2,3,…n)

⇒ vi ≡ (dri/dt) = ∑j(∂ri/∂qj)(dqj/dt) + (∂ri/∂t) =

• Squaring gives:
• where

• Always:

Or:

T0 ≡ M0 → independent of generalized velocities

T1 ≡ → linear in generalized velocities

T2 ≡ → quadratic in generalized velocities

 NOTE: From previous equations, if (∂ri/∂t) = 0 (if transformation
equations do not contain time explicitly), then
T0 = T 1 = 0 ⇒ T = T2

⇒ If the transformation equations from Cartesian to generalized

coordinates do not contain the time explicitly, the kinetic energy is a
homogeneous, quadratic function of the generalized velocities.