Chapter12

CoupledOscillations

Manyimportantphysicssystemsinvolvedcoupledoscillators.Coupledoscillatorsare
oscillators connectedin such awaythat energycan betransferred betweenthem.The
motionofcoupledoscillatorscanbecomplex,anddoesnothavetobeperiodic.However,
whentheoscillatorscarryoutcomplexmotion,wecanfindacoordinateframeinwhich
eachoscillatoroscillateswithaverywelldefinedfrequency.
A solid is a good example of a system that can be described in terms of coupled
oscillations.The atoms oscillate around their equilibrium positions, and the interaction
between the atoms is responsible for the coupling.To start our study of coupled
oscillations,wewillassumethattheforcesinvolvedarespringlikeforces(themagnitude
oftheforceisproportionaltothemagnitudeofthedisplacementfromequilibrium).

TwoCoupledHarmonicOscillators
Consider asystem of twoobjects of massM.The two objects are attached to two
springswithspringconstants(seeFigure1).Theinteractionforcebetweenthemassesis
representedbyathirdspringwithspringconstant12,whichconnectsthetwomasses.

Figure1.Twocoupledharmonicoscillators.

Wewillassumethatwhenthemassesareintheirequilibriumposition,thespringsarealso
intheirequilibriumpositions.Theforceontheleftmassisequalto

Theforceontherightmassisequalto

Theequationsofmotionarethus


Sinceitisreasonabletoassumethattheresultingmotionhasanoscillatorybehavior,we
considerfollowingtrialfunctions:

Substituting these trial functions into the equations of motion we obtain the following
conditions:

Theseequationsonlywillhaveanontrivialsolutionif

Note:thetrivialsolutionisB1=B2=0.Therequirementforanontrivialsolutionrequires
thattheangularfrequencyofthesystemisequaltooneofthefollowingtwocharacteristic
frequencies(thesocalledeigenfrequencies):

Foreachofthesefrequencies,wecannowdeterminetheamplitudesB1andB2.Letusfirst
consider the eigen frequency1.For this frequency we obtain the following relations
betweenB1andB2:

orB1= B2.For the eigen frequency2we obtain the following relations

betweenB1andB2:

orB1=B2.Themostgeneralsolutionofthecoupledharmonicoscillatorproblemisthus

Anotherapproachthatcanbeusedtosolvethecoupledharmonicoscillatorproblemisto
carryoutacoordinatetransformationthatdecouplesthecoupledequations.Considerthe
twoequationsofmotion.Ifweaddthemtogetherweget

Ifwesubtractfromeachotherweget

Basedonthesetwoequationsitisclearthatinordertodecoupletheequationsofmotion
weneedtointroducethefollowingvariables

Thesolutionstothedecoupledequationsofmotionare

wherethefrequenciesarethecharacteristicfrequenciesdiscussedbefore.Oncewehave
thesesolutionswecandeterminethepositionsofthemassesasfunctionoftime:

Wenotethatthesolution1correspondstoanasymmetricmotionofthemasses,whilethe
solution2corresponds to an asymmetric motion of the masses (see Figure 2).Since
higherfrequenciescorrespondtohigherenergies,theasymmetricmode(outofphase)hasa
higherenergy.


Figure2.Normalmodesofoscillation.

WeakCoupling
Coupled oscillations, involving a weak coupling, are important to describe many
physical systems.For example, in many solids, the force that tie the atoms to their
equilibriumpositionsareverymuchstrongerthantheinteratomiccouplingforces.Inthe
example we discussed in the pervious section, the weak coupling limit requires
that12<<.In this approximation we can show (see text book for details) that our
solutions have a highfrequency component that oscillates inside a slowly varying
component (see Figure 3).The solutions are thus sinusoidal functions with a slowly
varyingamplitude.

Figure3.Examplesofsolutionintheweakcouplinglimit.

Example:

Problem12.1
ReconsidertheproblemoftwocoupledoscillatorsdiscussioninSection12.2inthe
eventthatthethreespringsallhavedifferentforceconstants.Findthetwocharacteristic
frequencies,andcomparethemagnitudeswiththenaturalfrequenciesofthetwooscillators
intheabsenceofcoupling.

The equations of motion are

(1)
We attempt a solution of the form

(2)
Substitution of (2) into (1) yields

(3)
In order for a non-trivial solution to exist, the determinant of coefficients of
vanish. This yields

and

must

(4)
from which we obtain
(5)
This result reduces to
If

for the case

(compare Eq. (12.7)].

were held fixed, the frequency of oscillation of

would be
(6)

while in the reverse case,

would oscillate with the frequency

(7)

Comparing (6) and (7) with the two frequencies,

and

, given by (5), we find

(8)
so that
(9)
Similarly,

(10)
so that
(11)
If

, then the ordering of the frequencies is

(12)

Example:

Problem12.3
Two identical harmonic oscillators (with massesMand natural frequenciesw0) are
coupled such that by adding to the system a massm, common to both oscillators, the
equationsofmotionbecome

Solvethispairofcoupledequations,andobtainthefrequenciesofthenormalmodesofthe
system.

The equations of motion are

(1)
We try solutions of the form
(2)
We require a non-trivial solution (i.e., the determinant of the coefficients of
to zero), and obtain

and

equal

(3)
so that
(4)

and then

(5)
Therefore, the frequencies of the normal modes are

(6)
where

corresponds to the symmetric mode and

to the anti-symmetric mode.

By inspection, one can see that the normal coordinates for this problem are the same as
those for the example of Section 12.2. Another approach to find the normal coordinates is
to try to find ways to add the two equations of motion in such a way that we get an
uncoupled differential equation. Consider what happens when we multiply the first
equation of motion by and add it to the second equation of motion:

This will become an uncoupled equations if

Thisequationcanonlybecorrectif

and

Takingthelastequationforandsubstitutingitintothesecondtolastequationweobtain


Thisshowsthat

andthenormalcoordinatesareproportionaltox2+x1andx2x1.

GeneralProblemofCoupledOscillations
Theresultsofourstudyofthecoupledharmonicoscillatorproblemresultsinanumber
ofdifferentobservations:
o Thecouplinginasystemwithtwodegreesoffreedomresultsintwocharacteristic
frequencies.
o The two characteristic frequencies in a system with two degree of freedom are
pushedtowardslowerandhigherenergiescomparedtothenoncoupledfrequency.
Letusnowconsiderasystemwithncoupledoscillators.Wecandescribethestateofthis
systemintermsofngeneralizedcoordinatesqi.Theconfigurationofthesystemwillbe
describedwithrespecttotheequilibriumstateofthesystem(atequilibrium,thegeneralized
coordinatesare0,andthegeneralizedvelocityandaccelerationare0).Theevolutionofthe
systemcanbedescribedusingLagrange'sequations:

The second term on the lefthand side will contain terms that include the generalized
velocity and the generalized acceleration, and is thus equal to 0 at the equilibrium
position.Lagrange'sequationsthustellsusthat

However,sinceweknowhowtoexpressthekineticenergyofthesystemintermsofthe
generalizedcoordinatesweconcludethat

where


ForthepotentialenergyUweconcludethat

ThepotentialenergycanbeexpandedaroundtheequilibriumpositionusingaTaylorseries
andwefindthat

where

Wethusconcludethat:

Theequationofmotioncannowbewrittenas

Theindexkrunsoveralldegreesoffreedomofthesystem,andwethushavensecond
orderdifferentialequations.Inordertofindthegeneralsolutionwetryatrialsolutionthat
exhibitstheexpectedoscillatorybehavior:

Withthissolution,theequationsofmotionbecome

Thissetofequationswillonlyhasnontrivialsolutionsifthedeterminantofthecoefficient
mustvanish:


Ingeneraltherewillbendifferentvaluesoftheangularfrequency.Thesefrequenciesare
calledthecharacteristicfrequenciesoreigenfrequencies.Dependingonthecoefficients,
some of the characteristic frequencies are the same (this phenomena is called
degeneracy).Foreacheigenfrequencywecandeterminetheratiooftheamplitudes;these
amplitudesdefineanndimensionalvector,alsocalledtheeigenvector.Note:theeigen
vectorhasapureharmonictimedependence.
The general solution of the system is a linear combination of the solutionsqi.Of
course,itisonlytherealpartofthesolutionsthatismeaningful.
The normal coordinates can be determined by finding the appropriate linear
combinationsofsolutionsqithatoscillatesatasinglefrequency.Thesenormalcoordinates
are

The amplitude may be a complex number.The normal coordinates must satisfy the
followingrelation

Since there arenequations of motion, we also expect to seennormal coordinates,

andndecoupledequationsofmotion.
Toillustratethedetailedstepstobefollowedtosolveacoupledoscillatorproblemwe
willexamineExample12.4fromthetextbook.Inthis example,thecoupledpendulum
showninFigure4isexamined.

Figure4.CoupledpendulumofExample12.4.

1. Choosegeneralizedcoordinates.Thepropergeneralizedcoordinatesinthisproblem
arethe angles1and2.The kineticand thepotentialenergy ofthesystemcan be
easilyexpressedintermsoftheseangles.Wemaketheassumptionthatthespringis

massless and there is thus no kinetic energy associated with the motion of the
spring.Thekineticenergyofthesystemisthusjustequaltothekineticenergyofthe
twomasses,andthusequalto

Thepotentialenergyofthesystemisthesumofthepotentialenergyassociatedwiththe
change in the height of the masses and the potential energy associated with the
extensionorcompressionofthespring.Thetotalpotentialenergyisthusequalto

Wehaveusedthesmallangleapproximationinordertoexpressthesinandcosofthe
anglesintermsoftheangles.

2. Determine theAandmtensors.In order to calculate these tensors we use the

expressionsforTandUobtainedinstep1.Sincethekineticenergyobtainedinstep1
doesnotcontainproductsofthegeneralizedvelocityofmass1andthegeneralized
velocityofmass 2,the mass tensorwill beadiagonaltensor.We canseethis by
lookingatthedefinitionofthemasstensorelements:

Themasstensoristhusequalto

TheAtensorisequalto

3. Determinetheeigenfrequencyandtheeigenvectors.Theeigenfrequenciescanbe
determinedbyrequiringthatthedeterminantofthecoefficientsoftheequationsof
motionsvanishes:

Thisrequiresthat

or

Theeigenfrequenciesarethusequalto

Consider the first eigen frequency.For this frequency, the eigen vector is
(a11,a21).Theequationsofmotionforthisfrequencyare

Eachofthesetwoequationstellsusthata11=a21.

Since the eigen vectors are orthogonal, we expect that the eigen vector for the
secondeigenfrequencyisgivenbya12=a22.Wecometothesameconclusionifwe
startfromtheequationsofmotionforthatfrequencyandtheeigenvector(a12,a22):

Eachofthesetwoequationstellsusthata12=a22.

4. Determine the scale factors required to match the initial conditions.In this
example,wedonotneedtomatchinitialconditions(suchastheinitialdisplacementor
theinitialvelocityandwethusdonotneedtodeterminescalefactor).

5. Determinethenormalcoordinates.Thenormalcoordinatesarethosecoordinatesthat
oscillatewithasinglefrequency.Inthecurrentexamplewethusobservethefollowing
normalcoordinates:

Note: the constants in these equations need to be adjusted to match the initial
conditions.
Thesystemwillcarryoutamotionwithnormalfrequency1when2=0.Thisrequires
that1=2and the motion is symmetric. The system will carry out a motion with
normal frequency 2 when1= 0.This requires that1= 2and the motion is
asymmetric.

MolecularVibrations
Our theory of coupled oscillations has many important applications in molecular
physics.Eachatominamoleculehas3degreesoffreedom,andifwearelookingata
moleculewithnatoms,wehaveatotalof3ndegreesoffreedom.Threedifferenttypesof
motioncanbecarriedoutbytheatomsinthemolecule:translation(3degreesoffreedom),
rotation(3degreesoffreedom),andvibration(3n6degreesoffreedom).
Consideralinearmolecule(theequilibriumpositionsofallatomsarelocatedalonga
straightline)withnatoms.ThenumberofdegreesoffreedomassociatedwithVibrational
motionis3n5sincethereareonly2rotationaldegreesoffreedom.Thevibrationsina
linearmoleculecanbelongitudinalvibrations(therearen1degreesoffreedomassociated
withthistypeofvibrations)andtransversevibrations(thereare(3n5)(n1)=(2n4)

degrees offreedomassociatedwiththis typeofvibration).Ifthe vibrations areplanar

vibrations(themotionofallatomsiscarriedoutinasingleplane)wecanspecifyany
transverse vibration in terms of vibrations in two mutually perpendicular planes.The
characteristicfrequenciesineachoftheseplaneswillbethesame(symmetry)andthe
numberofcharacteristicfrequencieswillthusbeequalton2.
Toillustratemolecularvibrationsletusconsiderthedynamicsofatriatomicmolecule
(seeFigure5).

Figure5.Vibrationalmotionofalineartriatomicmolecule.

Inordertodeterminethevibrationalmodesofthissystemwelookatthelongitudinaland
transversalmodesseparately.Sincewearenotinterestedinpuretranslationalmotionwe
canrequirethatthecenterofmassofthesystemisatrest.Thismeansthatwedonothave
3independentpositioncoordinates,butonly2.Forexample,wecaneliminatetheposition
oftheheavyatom:

Inordertodeterminethenormalmodes,wewillfollowthesameprocedureasweusedin
thepreviousexample(note:thisdiffersfromtheapproachusedinthetextbook).

1. Choosegeneralizedcoordinates.Thepropergeneralizedcoordinatesinthisproblem
arethedisplacementsx1andx2.Thekineticandthepotentialenergyofthesystemcan
beeasilyexpressedintermsofthesedisplacements.Thekineticenergyofthesystemis
thusjustequaltothekineticenergyofthethreeatoms,andthusequalto


Thepotentialenergyofthesystemisthesumofthepotentialenergyassociatedwiththe
compressionofthesprings.Thetotalpotentialenergyisthusequalto

2. Determine theAandmtensors.In order to calculate these tensors we use the

expressionsforTandUobtainedinstep1.Sincethekineticenergyobtainedinstep1
doesnotcontainproductsofthegeneralizedvelocityofmass1andthegeneralized
velocityofmass 2,the mass tensorwill beadiagonaltensor.We canseethis by
lookingatthedefinitionofthemasstensorelements:

Themasstensoristhusequalto

TheAtensorisequalto

3. Determinetheeigenfrequencyandtheeigenvectors.Theeigenfrequenciescanbe
determinedbyrequiringthatthedeterminantofthecoefficientsoftheequationsof
motionsvanishes:

Thisrequiresthat

or

Considerthetwosigns.Firstthepositivesign:

Thisisequivalentto

or

Nowconsiderthenegativesign:

Thisisequivalentto


or

Consider the first eigen frequency, and assume the corresponding eigen vector is
(a11,a21).Theequationsofmotionforthisfrequencyare

Substitutingtheexpressionofthefirsteigenfrequencyintheseequationsweobtainfor
eachequationthefollowingexpression:

Thisequationstellsusthata11=a31.Sincetheeigenvectorsareorthogonal,weexpect
thattheeigenvectorforthesecondeigenfrequencyisgivenbya12=a32.

4. Determine the scale factors required to match the initial conditions.In this
example,wedonotneedtomatchinitialconditions(suchastheinitialdisplacementor
theinitialvelocityandwethusdonotneedtodeterminescalefactor).

5. Determinethenormalcoordinates.Thenormalcoordinatesarethosecoordinatesthat
oscillatewithasinglefrequency.Inthecurrentexamplewethusobservethefollowing
normalcoordinates:

Note: the constants in these equations need to be adjusted to match the initial
conditions.
Thesystemwillcarryoutamotionwithnormalfrequency1when2=0.Thisrequires
thatx1=x3andthemotionisasymmetric.Thesystemwillcarryoutamotionwith

normal frequency 2 when1= 0.This requires thatx1=x3and the motion is

symmetric.
Note:thenormalfrequency1isequaltothefrequencyofamassmonaspringwhose
otherendremainsfixed.Thismoderequiresthecentralatomtoremainfixed,andthis
canbeachievedwhenthemotionisasymmetricsincetheforcesexertedbythetwo
springsonthecentralmasscancel.

The transverse vibration of the molecule can be specified in terms of a single

parameter.For this mode of vibration we will get a single "uncoupled" differential
equation with a single corresponding characteristic frequency.The calculation of this
frequencyisshownindetailinthetextbookandwillnotbereproducedhere.

Example:

Problem12.21
Three oscillators ofequal massmare coupled such that the potential energy ofthe
systemisgivenby

where

Find the eigen frequencies by solving the secular equation.What is the physical
interpretationofthezerofrequencymode?

The tensors

and

are:

(1)

(2)
thus, the secular determinant is

(3)
from which
(4)
In order to find the roots of this equation, we first set

and then factor:

(5)
Therefore, the roots are

(6)
Consider the case

. The equation of motion is

(7)

so that
(8)
with the solution
(9)
That is, the zero-frequency mode corresponds to a translation of the system with
oscillation.

TheLoadedString
Agoodmodelofanelasticstringisastringofparticlesofmassm,eachseparatedbya
distanced(seeFigures6and7).Wewillassumethatthetensioninthestringisconstant
andequalto.

Figure6.Theloadedstring.

Figure7.Calculationoftherestoringforceactingonmassj.

Inexaminingthisproblem,wewillmakethefollowingassumptions:
Themassescanonlymoveintheverticaldirection(thusonlythecomponentofthe
tensionintheverticaldirectionmatters).
Thepotentialenergyofthesystemisthepotentialenergyassociatedwiththetension
inthestring.
Weassumethatthedisplacementsfromtheequilibriumpositionsaresmall.
We ignore the gravitational forces acting on the masses (and the associated
gravitationalpotentialenergy).
Inordertocalculatetheforceactingonmassjwecalculatetheverticalcomponentsdueto
thetensionintheleftandrightsectionofthestring:

In the last step we have made the assumption that the vertical displacement is small
comparedtothedistanced.Sincetheforceonmassjdependsnotonlyonthepositionof
massjbut also on the position of massesj 1 andj+ 1.We can use the force on
thenmassestoobtainncoupleddifferentialequationsthatwecantrytosolve.Consider
thefollowingtrialfunction:

Substitutingthisfunctionintoourdifferentialequationweobtain

or


Theamplitudesacanbecomplex.Basedonthetypeofmotionweexpectthesystemto
carryout,wecantrytoparameterizetheamplitudedependenceonjinthefollowingway:

whereaisnowarealnumber.Takingthis expressionforajandsubstitutingitintothe
previousequationweobtain

Thisexpressioncanbeusedtofindthefollowingexpressionfortheangularfrequency:

or

Sincetheremustbeneigenfrequencies,weexpecttofindndistinctvaluesof.
Additional constraints are imposed on the solution by requiring that the boundary
conditionsaremet:
a0=0:This conditionrequires that(note:weonlyconsidertherealpartofthe
amplitude)

or

an+1=0:Thisconditionrequiresthat

Theargumentofthesinfunctionmustthusbeanintegermultipleof:


or

wheres=1,2,3,,n.
Sincetheboundaryconditionsprovideuswithndifferentvaluesoftheparameter,we
expectthattherewillalsobenuniquevaluesoftheangularfrequencyforthissystem:

wheres=1,2,3,,n.
Puttingallthedifferentpiecesofinformationtogetherwecannowwritedownthe
generalsolutionoftheloadedstringproblem:

and

WecanalsousetheLagrangianmethodtofindthenormalmodesofthesystem,butas
wewillsee,thisapproachismuchlesstransparentthantheapproachjustused.Inorderto
apply this procedure we need to determine the kinetic energy and the potential of the
systemintermsofthegeneralizedcoordinates.Inthisparticularproblem,thebestchoice
forthegeneralizedcoordinatesistheverticaldisplacementofthemasses.Intermsofthese
displacementswecanwritethekineticenergyas

Inordertodeterminethepotentialenergyofthesystem,wefirsthavetodeterminethe
potentialenergyofmassj.Sinceweknowtherelationbetweenthepotentialenergyand
theforce,wecanseethatthepotentialenergyisequalto

Note:theindexrunsfromj=1toj=n+1.Therearenomasses atposition0andat
position (n+1)d; these positions are the ends of the string.The displacement at these
locationsisequalto0.
Note:inordertoverifythatthepotentialenergyiscorrect,weneedtoshowthatitsgradient
isrelatedtotheforceonmassj:

Themasstensormforthesystemisgivenby

ThepotentialtensorAforthesystemisgivenby

Theeigenfrequenciescannowbefoundbyrequiringthattheseculardeterminantisequal
to0:

Wecansolvethisequationforbuttheresultsaremoredifficulttointerpretthanthe
resultsobtainedwithoutfirstapproach.