Classical Mechanics Fall 2011 Chapter 7: Lagrange’s...

13
1 Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations 1. Introduction to Lagrange’s Equations Let us consider a single particle that moves under the influence of conservative forces. There are no constraints on the motion of the particle; however, its path must be consistent with Newton’s second law. The kinetic energy of the particle is T = 1 2 m x 2 + y 2 + z 2 ( ) , (7.1) and the potential energy is U = U ( r ) = U ( x, y, z ). (7.2) The Lagrangian for this system is defined as L = T U . (7.3) Let us relate the Lagrangian to Newton’s second law. Since the forces are conservative, F x = U x = L x . (7.4) Further, T x = L x = m x = p x . (7.5) By Newton’s second law, F x = p x . (7.6) In terms of L, Newton’s second law is given by L x = d dt L x . (7.7) This equation is called Lagrange’s equation. Similar equations can be derived for the y and z components. Clearly, Lagrange’s equations (for all Cartesian coordinates) are equivalent to Newton’s second law. Eq. (7.7) has exactly the same form as the EulerLagrange equation that we have previously seen. It follows that δ S = δ L dt t 1 t 2 = 0, (7.8)

Transcript of Classical Mechanics Fall 2011 Chapter 7: Lagrange’s...

Page 1: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  1  

Classical Mechanics Fall 2011

Chapter 7: Lagrange’s Equations

1.  Introduction  to  Lagrange’s  Equations  

Let  us  consider  a  single  particle  that  moves  under  the  influence  of  conservative  forces.  There  are  no  constraints  on  the  motion  of  the  particle;  however,  its  path  must  be  consistent  with  Newton’s  second  law.  The  kinetic  energy  of  the  particle  is    

  T = 1

2m x2 + y2 + z2( ),   (7.1)  

and  the  potential  energy  is    

  U =U(r ) =U(x, y, z).   (7.2)  

The  Lagrangian  for  this  system  is  defined  as  

  L = T −U.   (7.3)  

Let  us  relate  the  Lagrangian  to  Newton’s  second  law.  Since  the  forces  are  conservative,  

  Fx = − ∂U

∂x= ∂L∂x.   (7.4)  

Further,    

  ∂T∂ x

= ∂L∂ x

= mx = px .   (7.5)  

By  Newton’s  second  law,    

  Fx = px .   (7.6)  

In  terms  of  L,  Newton’s  second  law  is  given  by    

  ∂L∂x

= ddt

∂L∂ x.   (7.7)  

This  equation  is  called  Lagrange’s  equation.  Similar  equations  can  be  derived  for  the  y  and  z  components.  Clearly,  Lagrange’s  equations  (for  all  Cartesian  coordinates)  are  equivalent  to  Newton’s  second  law.  

Eq.  (7.7)  has  exactly  the  same  form  as  the  Euler-­‐Lagrange  equation  that  we  have  previously  seen.  It  follows  that      

  δS = δ Ldt

t1

t2

∫ = 0,   (7.8)  

Page 2: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  2  

i.e.,  the  integral  S  (called  the  action  integral)  is  stationary  for  any  path  followed  by  the  particle  within  the  time  interval  t1  to  t2.  Eq.  (7.8)  is  called  Hamilton’s  principle.  Hamilton’s  principle  is  usually  regarded  as  fundamental  in  mechanics  and  Lagrange’s  equations  are  derived  from  it.  

Generalized  Coordinates  

It  is  often  much  easier  to  use  coordinate  systems  other  than  the  Cartesian  one  to  solve  problems.  Do  Lagrange’s  equations  change  when  we  move  from  the  Cartesian  coordinate  system  to  another  one?  Let  us  assume  that  we  can  fully  describe  the  configuration  (position,  orientation,  etc.)  of  the  system  in  terms  of  another  set  of  coordinates.  A  set  of  quantities  that  can  completely  describe  the  configuration  or  state  of  a  system  is  called  a  set  of  generalized  coordinates.  These  could  be  other  common  coordinate  systems  such  as  the  spherical  polar  system,  or  they  may  be  other  quantities  that  are  not  geometric  in  nature.  For  an  unconstrained  particle  moving  in  three-­‐dimensional  space,  three  generalized  coordinates  are  necessary  to  describe  its  configuration,  which  is  just  the  position  in  the  case  of  a  single  particle.  The  position  of  the  particle  can  be  uniquely  specified  in  terms  these  three  generalized  coordinates.  In  other  words,  each  Cartesian  component  of  the  position  can  be  uniquely  expressed  in  terms  of  these  coordinates:  

  x = x(q1,q2,q3); y = y(q1,q2,q3); z = z(q1,q2,q3).   (7.9)  

In  these  coordinates,  the  action  integral  is    

  S = L(q1,q2,q3, q1, q2, q3)dt

t1

t2

∫ .   (7.10)  

Clearly,  the  value  of  the  definite  integral  will  not  change  with  a  change  of  coordinates.  Thus,  the  action  integral  is  still  stationary  and  Lagrange’s  equations  have  exactly  the  same  form  in  the  new  (and  any)  coordinate  system:  

 

∂L∂qi

= ddt

∂L∂ qi

. [i = 1,2,3]   (7.11)  

In  analogy  to  Newton’s  second  law  in  Cartesian  coordinates,  the  quantity   ∂L ∂qi  is  defined  as  the  ith  component  of  the  generalized  force  and   ∂L ∂ qi  is  the  ith  component  of  the  generalized  momentum.  To  make  these  ideas  more  concrete,  let  us  use  the  Lagrangian  formalism  for  a  particle  moving  in  a  plane.  We  will  specify  the  particle’s  position  using  polar  coordinates.  

Prelude:  In-­‐class  Problem:  Taylor,  Problem  7.5  

Two-­‐Dimensional  Motion  of  a  Single  Particle  Using  Polar  Coordinates  

The  kinetic  energy  of  the  particle  is    

Page 3: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  3  

  T = 1

2m r2 + r2 φ 2( ).   (7.12)  

The  potential  energy  is    

  U =U(r,φ).   (7.13)  

The  Lagrangian  is  therefore    

  L = 1

2m r2 + r2 φ 2( )−U(r,φ).   (7.14)  

Applying  Lagrange’s  equation  for  the  radial  coordinate  gives    

 

∂L∂r

= ddt

∂L∂ r

, which gives

mr φ 2 − ∂U∂r

= mr, or, − ∂U∂r

= mr −mr φ 2.   (7.15)  

Since  Fr = −∂U ∂r,  we  see  that  the  generalized  force  for  the  radial  coordinate  is  the  sum  of  the  negative  of  the  centripetal  force  (i.e.,  the  centrifugal  force)  and  the  radial  component  of  the  conservative  force.  Of  course,  we  obtained  exactly  the  same  equation  using  Newton’s  second  law  in  the  form  of  the  second  equation  in  Eq.  (7.15).  

For  the  polar-­‐angle  coordinate,  the  Lagrange  equation  is    

 

∂L∂φ

= ddt

∂L∂ φ.   (7.16)  

This  gives  

    − ∂U∂φ

= ddt

mr2 φ( ).   (7.17)  

Now,  in  polar  coordinates  

 

∇U = ∂U

∂rr̂ + 1

r∂U∂φ

φ̂   (7.18)  

Thus,  Fφ = − 1r ∂U ∂φ( )  and  so  Eq.  (7.17)  becomes  

  rFφ =dldt,   (7.19)  

where   l = mr2 φ  is  the  angular  momentum.  Thus,  the  generalized  momentum  in  this  case  is  

the  angular  momentum  and  the  generalized  force  is  the  torque.  One  could  say  that  the  

Page 4: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  4  

generalized  force  and  momentum  are  the  quantities  that  arise  naturally  from  the  generalized  coordinates  used  to  describe  the  system.  

The  Lagrangian  formalism  can  be  easily  extended  to  systems  consisting  of  more  than  one  unconstrained  particles.  Thus,  if  we  have  a  system  of  N  particles,  we  would  need  three  generalized  coordinates  per  particle,  and  so  a  total  of  3N  generalized  coordinates  would  be  needed.  The  corresponding  Lagrange’s  equations  are    

 

∂L∂qi

= ddt

∂L∂ qi

. [i = 1,2,...., 3N ]   (7.20)  

The  kinetic  and  potential  energies  in  the  Lagrangian  are  the  total  values  for  the  system.  

2.  Lagrange’s  Equations:  Constrained  Motion  

A  particle  moving  on  a  horizontal  table  is  constrained  to  move  in  two  dimensions  because  of  the  action  of  the  normal  force.  The  normal  force  is  called  a  constraint  force.    Constraints  reduce  the  number  of  generalized  coordinates  needed  to  describe  the  motion  of  the  system.  For  example,  to  describe  the  motion  of  a  single  particle  moving  on  a  horizontal  table,  only  two  coordinates  are  needed;  the  number  of  coordinates  have  been  reduced  from  three  to  two.  In  general,  for  a  system  of  N  particles,  the  x-­‐component  of  the  position  of  the  ith  particle  is  given  by    

  xi = xi q1,q2,....qn ,t( ), [i = 1,2,...,N ]   (7.21)  

where  the  number  of  generalized  coordinates  n  <  3N  in  a  constrained  system.  The  other  Cartesian  coordinates  can  be  similarly  expressed  in  terms  of  the  generalized  coordinates.    The  number  of  degrees  of  freedom  of  a  system  is  the  number  of  independent  coordinates  necessary  to  specify  an  arbitrary  displacement.  Put  another  way,  a  system  has  p  degrees  of  freedom  if  exactly  p  independent  parameters  are  needed  to  specify  the  configuration  of  the  system  at  any  point  in  time.  If  the  number  of  degrees  of  freedom  equals  the  number  of  generalized  coordinates  needed  to  describe  the  system,  the  system  is  said  to  be  holonomic.  A  holonomic  system  has  holonomic  constraints,  which  can  be  described  by  equations  relating  the  generalized  coordinates  of  the  unconstrained  system:  

  gj q1,q2,...,q3N ,t( ) = 0, [j = 1,2,..,m]   (7.22)  

where  gj    is  the  jth  functional  relationship  among  the  generalized  coordinates  and  there  are  m  constraints.  The  number  of  degrees  of  freedom  (and  the  number  of  generalized  coordinates  for  the  constrained  system)  is  n  =  3N  –  m.  Though  it  seems  natural  for  the  number  of  generalized  coordinates  to  be  equal  to  the  number  of  degrees  of  freedom,  it  does  not  necessarily  have  to  be  so.  However,  we  will  not  deal  with  non-­‐holonomic  systems  in  this  course.    

Using  the  Lagrangian  formalism  for  holonomic  systems,  one  finds  that  one  can  determine  the  equations  of  motion  without  considering  the  constraint  forces.  This  is  not  possible  with  Newtonian  dynamics.  The  equations  of  motion  will  involve  the  independent  generalized  

Page 5: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  5  

coordinates  (and  their  time  derivatives)  used  to  describe  the  system  in  the  presence  of  constraints  and  so  will  be  consistent  with  the  constraints.  

Consider  Eq.  (7.21)  for  a  single  particle.  The  n  generalized  coordinates  are  consistent  with  the  constraints.  The   x -­‐component  of  the  velocity  can  be  expressed  in  terms  of  the  generalized  coordinates  via  the  chain  rule:    

  x = ∂x

∂q1

dq1dt

+ ∂x∂q2

dq2dt

+ ...+ ∂x∂qn

dqndt

+ ∂x∂t

= ∂x∂qi

dqidt

+ ∂x∂ti=1

n

∑ .   (7.23)  

If  we  assume  natural  coordinates,   x  is  not  an  explicit  function  of  time  and  so  ∂x ∂t = 0  in  Eq.  (7.23).  We  can  define  the  generalized  velocity  corresponding  to  the  ith  generalized  coordinate  as  

  qi =

dqidt.   (7.24)  

Thus,   x is  a  function  of  both  the  generalized  coordinates  and  velocities:  

  x = x q1,q2,...qn , q1, q1,..qn( ).   (7.25)  

 One  can  obtain  similar  expression  for   y  and   z .  One  can  then  evaluate  the  kinetic  energy  

  T = 1

2m x2 + y2 + z2( )   (7.26)  

in  terms  of  the  generalized  coordinates  and  velocities  using  Eq.  (7.25)  and  its  analogs  for   y  and   z .  One  can  take  the  derivative  of  T  with  respect  to  the  generalized  velocities,  and  identifying  the  ith  component  of  the  generalized  force  as  

  Qi = Fx∂x∂qi

+ Fy∂y∂qi

+ Fz∂z∂qi

= − ∂U∂qi,   (7.27)  

one  finds,  upon  using  Newton’s  second  law,  that    

 

ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟= ∂L∂qi.   (7.28)  

Thus,  Lagrange’s  equations  have  exactly  the  same  form  if  the  motion  is  constrained.  Note  that  Eq.  (7.28)  is  valid  for  any  choice  of  generalized  coordinates  (the  number  of  coordinates  must  be  equal  to  the  number  of  degrees  of  freedom;  i.e.,  the  system  must  be  holonomic)  because  the  Lagrangian  is  a  scalar  quantity,  which  is  invariant  under  coordinate  transformations.    Again,  Lagrange’s  equations  can  be  straightforwardly  extended  to  systems  of  more  than  one  particles.  In  this  case,  the  kinetic  energy  and  potential  energy  in  the  Lagrangian  are  the  total  values  for  the  system.  

 

Page 6: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  6  

 

3.  Applications  of  Lagrange’s  Equations  

Taylor,  Problem  7.23  

The  motion  of  the  larger  cart  is  specified,  so  the  coordinate  X  is  specified  and  represents  a  constraint.  We  need  only  the  coordinate  x  to  describe  the  motion  of  the  smaller  cart.  We  must  remember  to  write  down  the  Lagrangian  with  respect  to  an  inertial  reference  frame.  This  is  because  Newton’s  second  law  has  the  usual  form  only  in  inertial  frames.  Thus,  even  though  the  coordinates  themselves  may  be  defined  with  respect  to  non-­‐inertial  frames  (as  is  x  in  this  problem),  the  Lagrangian  itself  must  be  constructed  in  an  inertial  frame.  

The  velocity  of  the  smaller  cart  (relative  to  the  inertial  frame  attached  to  the  floor)  is  

v = d dt X + x( ) = X + x.  As  both  carts  move,  the  extension  of  the  spring  relative  to  the  floor  is  simply  x.  Thus,  the  Lagrangian  for  the  smaller  cart  is    

  L = 1

2m x + X( )2 − 12 kx

2.  

Lagrange’s  equation  for  the  coordinate  x  is  

 

ddt

∂L∂ x

⎛⎝⎜

⎞⎠⎟ −

∂L∂x

= 0,  

i.e.,      

 

m ddtx + X( ) + kx = 0.

m x + X( ) +mω 02x = 0.

x +ω 02x = − X =ω 2Acosωt,

 

which  is  the  required  form,  with   B =ω 2A .  The  equation  of  motion  represents  the  forced  harmonic  oscillator  without  damping.  What  would  happen  to  the  cart  eventually  in  this  ideal  case?  

Taylor,  Problem  7.38  

(a)  In  spherical  polar  coordinates,  the  velocity  is  given  by    

 

v = r = ddt

rr̂( ) = rr̂ + r ̂r.   (7.29)  

Now,  any  infinitesimal  change  in   r̂  must  be  perpendicular  to  it  (the  magnitude  cannot  change!)  and  so  the  tip  of  the   r̂ unit  vector  must  move  over  the  surface  of  a  sphere.  It  follows  that    

Page 7: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  7  

  dr̂ = dθθ̂ + sinθdφφ̂.  

Thus,    

  dr̂dt

= dθdt

θ̂ + sinθ dφdt

φ̂,  

or,    

  ̂r = θθ̂ + sinθ φφ̂.   (7.30)  

Thus,  the  velocity  in  spherical  coordinates  is  

  v = rr̂ + r θθ̂ + r sinθ φφ̂.   (7.31)  

(a)  The  Lagrangian  is    

  L = 1

2m r2 + r2 θ 2 + r2 sin2θ φ 2( )−mgr cosθ .  

(Remember  that   z = r cosθ .)  For  a  cone,  the  polar  angle  θ  is  fixed:  θ =α .  Thus,   θ = 0 .  Therefore,  the  Lagrangian  becomes  

  L = 1

2m r2 + r2 sin2α φ 2( )−mgr cosα . ----- (i)  

(b)  The  equations  of  motion  can  be  found  from  Lagrange’s  equations.  For  the  radial  coordinate,  we  have  

 

ddt

∂L∂ r

⎛⎝⎜

⎞⎠⎟ =

∂L∂r.  

This  yields  

  mr = mr sin2α φ 2 −mgcosα .    -­‐-­‐-­‐-­‐-­‐  (ii)  

For  the  φ  coordinate,  we  have  

 

ddt

∂L∂ φ

⎛⎝⎜

⎞⎠⎟= ∂L∂φ.  

This  gives  

 

ddt

mr2 sin2α φ( ) = 0.  

We  can  rewrite  the  above  equation  as  

Page 8: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  8  

 

ddt

mρ2 φ( ) = 0,  

where   ρ = r sinα .  Now,    

                          lz = mρ2 φ, ----- (iii)    

so  the  equation  of  motion  becomes  

  dlzdt

= 0,  

or,  

  lz = constant.  

Using  Eq.  (iii)  to  substitute  for   φ  in  Eq.  (ii),  we  obtain  

  mr = mr sin2α lz

mr2 sin2α⎛⎝⎜

⎞⎠⎟2

−mgcosα ,  

or,    

  mr = lz

2

mr3 sin2α−mgcosα . ---- (iv)  

If   lz = 0,  Eq.  (iv)  becomes  

  mr = −mgcosα .  

This  tells  us  that  the  radial  force  (away  from  the  apex  of  the  cone)  equals  the  radial  component  of  the  weight  (up  is  positive,  so  the  component  pointing  away  from  the  apex  of  the  cone  is  negative).  This  makes  perfect  sense  if  the  z-­‐component  of  the  angular  momentum  is  zero.  

The  particle  can  remain  in  a  horizontal  circular  path  if   r = 0 .  Substituting  this  in  Eq.  (iv)  and  solving  for  r  yields  

  r = r0 =lz2

m2gsin2α cosα⎛⎝⎜

⎞⎠⎟

1/3

.  

(c)  We  substitute   r = r0 + ε(t)  into  Eq.  (iv)  yielding  

  mε = lz

2

m r0 + ε( )3 sin2α−mgcosα . ---- (v)  

Now,  

Page 9: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  9  

  r0 + ε( )−3 = r0−3 1+ εr0( )−3 ≈ r0−3 1− 3ε

r0( ).  Substituting  this  approximate  result  into  Eq.  (v)  gives  

  mε = lz

2

mr03 sin2α

1− 3 εr0

⎛⎝⎜

⎞⎠⎟−mgcosα . ------ (vi)  

But,  as  established  above,    

  lz2

mr03 sin2α

= mgcosα ----- (vii)  

is  the  condition  for  a  horizontal  circular  orbit.  Thus  Eq.  (vi)  becomes  

  ε = −

3lz2

m2r04 sin2α

ε = 0. ----(viii)  

 

Eq.  (viii)  is  the  harmonic  oscillator  equation  in  ε ;  therefore,  the  particle  oscillates  about  a  stable  circular  path  with  angular  frequency  

  Ω =3lz2

m2r04 sin2α

= 3gcosαr0

.  

[The  last  expression  is  obtained  by  using    Eq.  (vii).]  

In-­‐class  Problem:  Taylor,  Problem  7.29.  

 

4.  Ignorable  Coordinates  and  Conservation  Laws  

Consider  a  Lagrangian  expressed  in  terms  of  generalized  coordinates   q1,q2,....,qn :  

  L = L(q1,q2,...,qn , q1, q2,..., qn ,t).   (7.32)  

The  Lagrange  equation  for  the  ith  coordinate  is    

 

ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟= ∂L∂qi.   (7.33)  

If   L  is  not  an  explicit  function  of  qi,  then  the  coordinate  qi    is  said  to  be  ignorable  or  cyclic.  If  a  coordinate  qi  is  ignorable,  then    

Page 10: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  10  

 

∂L∂qi

= 0. (Ignorable coordinate)   (7.34)  

Substituting  this  in  Eq.  (7.33)  gives  

 

ddt

∂L∂ qi

= 0.   (7.35)  

This  indicates  that  the  generalized  momentum   pi ≡

∂L∂ qi

 is  conserved.    (Unfortunately,  we  

are  using  the  same  symbol  for  momentum  and  generalized  momentum.  This  is  not  the  best  state  of  affairs,  but  it  is  the  norm.)  Thus,  if  the  Lagrangian  is  invariant  under  changes  in  a  coordinate,  then  the  corresponding  generalized  momentum  is  conserved.  This  very  important  relationship  between  the  invariance  of  the  Lagrangian  and  conservation  laws  is  called  Noether’s  theorem.  For  example,  if  the  Lagrangian  is  invariant  under  spatial  translations  in  a  given  direction,  then  the  component  of  the  total  momentum  in  that  direction  is  conserved.    Noether’s  theorem  expresses  the  beautiful  relationship  between  symmetry  and  conservation  laws  in  physics.    

It  is  also  useful  to  note  (though  I  will  not  derive  it  in  these  notes)  that  if  the  Lagrangian  does  not  depend  explicitly  on  time,  a  quantity  called  the  Hamiltonian  is  conserved.  The  Hamiltonian  is  defined  as    

  H = pi

i=1

n

∑ qi − L.   (7.36)  

Further,  if  the  relationship  between  the  Cartesian  coordinates  and  the  generalized  coordinates  is  time  independent,  i.e.,    

  xi = xi (q1,q2,...,qn ); yi = yi (q1,q2,...,qn ); zi = zi (q1,q2,...,qn ),   (7.37)  

then  the  Hamiltonian  is  simply  the  total  energy:  

  H = T +U.   (7.38)  

Thus,  if  Eq.  (7.37)  is  satisfied,  and  the  Lagrangian  is  independent  of  time,  the  total  energy  is  conserved.  

5.  Lagrange  Multipliers  and  Constraint  Forces  

One  of  the  benefits  of  the  Lagrangian  formalism  is  that  one  can  obtain  equations  of  motion  without  needing  to  deal  with  constraint  forces.  However,  if  one  needs  to  calculate  the  constraint  forces,  the  Lagrangian  method  can  also  accommodate  this.  

As  indicated  before,  holonomic  constraints  can  be  expressed  as  

  gj q1,q2,...,qN ,t( ) = 0, [j = 1,2,..,m]   (7.39)  

Page 11: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  11  

where  the  gj  are  algebraic  relationships  among  the  generalized  coordinates.  In  Eq.    (7.39),  there  are  N  generalized  coordinates  in  the  unconstrained  system  and  m  constraints.  Eq.  (7.39)  may  be  alternatively  expressed  as  

  f j q1,q2,...,qN ,t( ) = constant. [j = 1,2,..,m]   (7.40)  

An  example  of  a  holonomic  constraint  is  a  particle  moving  on  the  surface  of  a  sphere.  Using  Cartesian  coordinates,  the  equation  of  constraint  is  

  x2 + y2 + z2 = a2.   (7.41)  

Since  one  coordinate  can  be  expressed  in  terms  of  the  other  two  via  Eq.  (7.41),  there  are  only  two  independent  coordinates.  

In  differential  form,  the  holonomic  constraint  Eq.  (7.40)  can  be  expressed  as  

 ∂ f j∂qii

N

∑ dqi = 0, (j = 1,2,...,m)   (7.42)  

where  for  simplicity,  we  assume  that    the  constraint  equations  are  not  explicitly  dependent  on  time.  Hamilton’s  principle  tells  us  that    

  δ Ldt1

2

∫ = dt ∂L∂qi

− ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟

⎣⎢

⎦⎥

i=1

N

∑ δ1

2

∫ qi = 0.   (7.43)  

Taking   dqi ≈ δqi  in  Eq.  (7.42),  we  can  incorporate  Eq.  (7.42)  into  Eq.  (7.43):  

  δ Ldt1

2

∫ = dt ∂L∂qi

− ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟+ λ j

∂ f j∂qij=1

m

∑⎡

⎣⎢

⎦⎥

i=1

N

∑ δ1

2

∫ qi = 0,   (7.44)  

where  λ j is  an  undetermined  Lagrange  multiplier.  Because  of  the  constraints,  the  displacements  δqi  are  not  all  independent;  they  are  related  by  the  m  constraint  equations  in  Eq.  (7.42).  We  can  separate  Eq.  (7.44)  into  independent  and  non-­‐independent  parts.  There  are  N  –m  independent  coordinates;  thus,  we  have  

 

δ Ldt1

2

∫ = dt ∂L∂qi

− ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟+ λ j

∂ f j∂qij=1

m

∑⎡

⎣⎢

⎦⎥

i=1

N−m

∑ δ1

2

∫ qi

+ dt ∂L∂qi

− ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟+ λ j

∂ f j∂qij=1

m

∑⎡

⎣⎢

⎦⎥

i=N−m+1

N

∑ δ1

2

∫ qi = 0.   (7.45)  

We  can  now  choose  the  Lagrange  multipliers  λ j  such  that  the  second  integral  is  zero.  Thus,  the  quantity  in  the  square  brackets  is  zero.  Since  the  “independent”  δqi  displacements  are  arbitrary,  it  follows  that  the  first  integral  must  also  vanish.  Hence,  we  have    

Page 12: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  12  

 

ddt

∂L∂ qi

⎛⎝⎜

⎞⎠⎟= ∂L∂qi

+ λ j

∂ f j∂qij=1

m

∑ . (i = 1,2,...,N ); (j = 1,2,...,m)   (7.46)  

There  are  N  equations  for  the  generalized  coordinates  qi  and  including  the  multipliers,  N +m  unknowns.  But  there  are  m  equations  of  constraint,  so  the  system  can  be  solved  completely.  The  multipliers  λ j  are  related  to  the  constraint  forces.  

Example:  A  particle  of  mass  m  is  placed  at  the  top  of  a  smooth  hemisphere  of  radius  a.  Find  the  normal  force  exerted  on  the  particle  by  the  sphere.  If  the  particle  is  disturbed  and  slides  on  the  sphere,  find  the  vertical  distance  below  the  top  at  which  the  particle  leaves  the  sphere.  

We  will  use  spherical  polar  coordinates.  The  Lagrangian  is    

L = 1

2m r2 + r2 θ 2 + r2 sin2θ φ 2( )−mgr cosθ .  

Note  that   L  is  independent  of  φ ,  which  means  the  z-­‐component  of  the  angular  momentum  is  conserved.  Since  its  initial  value  is  zero  (because   φinitial = 0 ),  it  is  always  zero.  Thus,  we  can  rewrite  the  Lagrangian  as  

L = 1

2m r2 + r2 θ 2( )−mgr cosθ .  

The  constraint  equation  is  r  =  a.  Thus,   f (r,θ ) = r = a (a constant).  Hence,  

∂ f∂r

= 1  and   ∂ f∂θ

= 0.  Therefore,  we  need  only  one  multiplier,  which  we  will  call  λ .  

The  Lagrange  equation  for  the  radial  coordinate  is  therefore  (note  that   r = 0 and  r  =  a)  

ma θ2 −mgcosθ + λ = 0. ---- (i)  

The  Lagrange  equation  for  the  angular  coordinate  is  

ma2 θ = mgasinθ . --- (ii)  

We  can  integrate  Eq.  (ii)  to  obtain  (fill  in  the  details!)  

θ 2 = 2g

a1− cosθ( ).  -­‐-­‐-­‐-­‐  (iii)  

Using  Eq.  (iii)  to  substitute  for   θ2  in  Eq.  (i)  yields  

λ = mg(3cosθ − 2).  

The  quantity  λ  is  the  normal  force.  Setting  λ = 0  gives  the  angle  of  departure  

 

N  

mg  

Page 13: Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equationsfaculty.chas.uni.edu/~Shand/Classical_Mechanics... · Classical Mechanics Fall 2011 Chapter 7: Lagrange’s Equations

  13  

θ = θ f = arccos23

⎛⎝⎜

⎞⎠⎟ .  The  required  vertical  distance  is  

 h = r(1− cosθ f ) =r3.