Intuition The big pictureWhen you couple two pendulums (or masses on springs), they no longer swing independently — push one, and energy sloshes back and forth. This looks horribly complicated. The magic claim is:
Every small-oscillation system, no matter how tangled, can be re-described as a set of independent simple harmonic oscillators called normal modes . Each mode has its own single frequency. The messy real motion is just a superposition of these clean modes.
WHY this matters: it turns a system of coupled differential equations into separate, trivially-solvable 1D oscillator equations. This is the 80/20 of vibration physics — molecules, bridges, crystals, circuits all use it.
Take the classic: two equal masses m m m
Let outer springs have constant k k k coupling spring k c k_c k c x 1 , x 2 x_1, x_2 x 1 , x 2
Definition Coupled oscillatorA system whose equations of motion contain cross terms mixing the coordinates, e.g. x ¨ 1 \ddot{x}_1 x ¨ 1 x 2 x_2 x 2
Why use energy? Forces get sign-confusing fast. Lagrangian mechanics never makes a sign error.
Kinetic energy:
T = 1 2 m x ˙ 1 2 + 1 2 m x ˙ 2 2 T = \tfrac12 m\dot{x}_1^2 + \tfrac12 m\dot{x}_2^2 T = 2 1 m x ˙ 1 2 + 2 1 m x ˙ 2 2
Potential energy — Why this step? each spring stores 1 2 k ( stretch ) 2 \tfrac12 k(\text{stretch})^2 2 1 k ( stretch ) 2 ( x 2 − x 1 ) (x_2 - x_1) ( x 2 − x 1 ) V = 1 2 k x 1 2 + 1 2 k x 2 2 + 1 2 k c ( x 2 − x 1 ) 2 V = \tfrac12 k x_1^2 + \tfrac12 k x_2^2 + \tfrac12 k_c (x_2 - x_1)^2 V = 2 1 k x 1 2 + 2 1 k x 2 2 + 2 1 k c ( x 2 − x 1 ) 2
Lagrangian L = T − V L = T - V L = T − V d d t ∂ L ∂ x ˙ i = ∂ L ∂ x i \frac{d}{dt}\frac{\partial L}{\partial \dot{x}_i} = \frac{\partial L}{\partial x_i} d t d ∂ x ˙ i ∂ L = ∂ x i ∂ L
m x ¨ 1 = − k x 1 + k c ( x 2 − x 1 ) m\ddot{x}_1 = -k x_1 + k_c(x_2 - x_1) m x ¨ 1 = − k x 1 + k c ( x 2 − x 1 ) m x ¨ 2 = − k x 2 − k c ( x 2 − x 1 ) m\ddot{x}_2 = -k x_2 - k_c(x_2 - x_1) m x ¨ 2 = − k x 2 − k c ( x 2 − x 1 )
Why this step? The ∂ V / ∂ x 1 \partial V/\partial x_1 ∂ V / ∂ x 1 k x 1 kx_1 k x 1 ( x 2 − x 1 ) 2 (x_2-x_1)^2 ( x 2 − x 1 ) 2 k c k_c k c both equations with opposite sign — that is the coupling.
Forecast: if normal modes exist, there are special motions where both masses oscillate at the same frequency ω \omega ω x j ( t ) = A j cos ( ω t + ϕ ) x_j(t) = A_j \cos(\omega t + \phi) x j ( t ) = A j cos ( ω t + ϕ )
Substitute. Each x ¨ j = − ω 2 x j \ddot{x}_j = -\omega^2 x_j x ¨ j = − ω 2 x j algebraic equations:
− m ω 2 A 1 = − ( k + k c ) A 1 + k c A 2 -m\omega^2 A_1 = -(k+k_c)A_1 + k_c A_2 − m ω 2 A 1 = − ( k + k c ) A 1 + k c A 2 − m ω 2 A 2 = + k c A 1 − ( k + k c ) A 2 -m\omega^2 A_2 = +k_c A_1 - (k+k_c)A_2 − m ω 2 A 2 = + k c A 1 − ( k + k c ) A 2
In matrix form ( K − ω 2 M ) A = 0 (\mathbf{K} - \omega^2 \mathbf{M})\mathbf{A} = 0 ( K − ω 2 M ) A = 0
( k + k c − m ω 2 − k c − k c k + k c − m ω 2 ) ( A 1 A 2 ) = 0 \begin{pmatrix} k+k_c - m\omega^2 & -k_c \\ -k_c & k+k_c - m\omega^2 \end{pmatrix}\begin{pmatrix}A_1\\A_2\end{pmatrix} = 0 ( k + k c − m ω 2 − k c − k c k + k c − m ω 2 ) ( A 1 A 2 ) = 0
Expand:
( k + k c − m ω 2 ) 2 − k c 2 = 0 ⟹ k + k c − m ω 2 = ± k c (k+k_c - m\omega^2)^2 - k_c^2 = 0 \implies k+k_c - m\omega^2 = \pm k_c ( k + k c − m ω 2 ) 2 − k c 2 = 0 ⟹ k + k c − m ω 2 = ± k c
So:
ω 1 2 = k m ω 2 2 = k + 2 k c m \boxed{\omega_1^2 = \frac{k}{m}}\qquad \boxed{\omega_2^2 = \frac{k+2k_c}{m}} ω 1 2 = m k ω 2 2 = m k + 2 k c
Plug each ω \omega ω A 2 / A 1 A_2/A_1 A 2 / A 1
Mode 1 (ω 1 2 = k / m \omega_1^2 = k/m ω 1 2 = k / m k + k c − m ω 1 2 = k c k+k_c - m\omega_1^2 = k_c k + k c − m ω 1 2 = k c k c A 1 − k c A 2 = 0 ⇒ A 2 = A 1 k_c A_1 - k_c A_2 = 0 \Rightarrow A_2 = A_1 k c A 1 − k c A 2 = 0 ⇒ A 2 = A 1 In-phase / symmetric : masses move together, coupling spring never stretches, so it doesn't feel k c k_c k c k / m \sqrt{k/m} k / m
Mode 2 (ω 2 2 = ( k + 2 k c ) / m \omega_2^2 = (k+2k_c)/m ω 2 2 = ( k + 2 k c ) / m = − k c = -k_c = − k c A 2 = − A 1 A_2 = -A_1 A 2 = − A 1 Anti-phase / antisymmetric : masses move oppositely, coupling spring is maximally worked → stiffer → higher frequency.
Intuition Why symmetric is slowerIn the symmetric mode the middle spring is a passenger — it never stretches. Less restoring force ⇒ lower ω \omega ω k + 2 k c k+2k_c k + 2 k c ω \omega ω before algebra.
Define new variables that decouple the equations:
η 1 = 1 2 ( x 1 + x 2 ) η 2 = 1 2 ( x 1 − x 2 ) \eta_1 = \tfrac{1}{\sqrt2}(x_1 + x_2)\qquad \eta_2 = \tfrac{1}{\sqrt2}(x_1 - x_2) η 1 = 2 1 ( x 1 + x 2 ) η 2 = 2 1 ( x 1 − x 2 )
Why these? They are the symmetric and antisymmetric combinations — exactly the eigenvectors. Add and subtract the two EOMs:
Add: m ( x ¨ 1 + x ¨ 2 ) = − k ( x 1 + x 2 ) ⇒ η ¨ 1 = − ω 1 2 η 1 m(\ddot{x}_1+\ddot{x}_2) = -k(x_1+x_2) \Rightarrow \ddot{\eta}_1 = -\omega_1^2 \eta_1 m ( x ¨ 1 + x ¨ 2 ) = − k ( x 1 + x 2 ) ⇒ η ¨ 1 = − ω 1 2 η 1
Subtract: m ( x ¨ 1 − x ¨ 2 ) = − ( k + 2 k c ) ( x 1 − x 2 ) ⇒ η ¨ 2 = − ω 2 2 η 2 m(\ddot{x}_1-\ddot{x}_2) = -(k+2k_c)(x_1-x_2) \Rightarrow \ddot{\eta}_2 = -\omega_2^2 \eta_2 m ( x ¨ 1 − x ¨ 2 ) = − ( k + 2 k c ) ( x 1 − x 2 ) ⇒ η ¨ 2 = − ω 2 2 η 2
Definition Normal coordinatesCoordinates η a \eta_a η a completely decoupled , each obeying a simple harmonic oscillator equation η ¨ a = − ω a 2 η a \ddot{\eta}_a = -\omega_a^2 \eta_a η ¨ a = − ω a 2 η a
Worked example Energy sloshing (beats)Start with mass 1 pulled aside, mass 2 at rest: x 1 ( 0 ) = a x_1(0)=a x 1 ( 0 ) = a x 2 ( 0 ) = 0 x_2(0)=0 x 2 ( 0 ) = 0 Step: find η \eta η η 1 ( 0 ) = a / 2 , η 2 ( 0 ) = a / 2 \eta_1(0)=a/\sqrt2,\ \eta_2(0)=a/\sqrt2 η 1 ( 0 ) = a / 2 , η 2 ( 0 ) = a / 2 Why? just plug into definitions. Velocities zero ⇒ ϕ a = 0 \phi_a=0 ϕ a = 0 η 1 = a 2 cos ω 1 t \eta_1 = \tfrac{a}{\sqrt2}\cos\omega_1 t η 1 = 2 a cos ω 1 t η 2 = a 2 cos ω 2 t \eta_2 = \tfrac{a}{\sqrt2}\cos\omega_2 t η 2 = 2 a cos ω 2 t x 1 = a 2 ( cos ω 1 t + cos ω 2 t ) = a cos ( ω 2 − ω 1 2 t ) cos ( ω 1 + ω 2 2 t ) x_1 = \tfrac{a}{2}(\cos\omega_1 t + \cos\omega_2 t) = a\cos\!\big(\tfrac{\omega_2-\omega_1}{2}t\big)\cos\!\big(\tfrac{\omega_1+\omega_2}{2}t\big) x 1 = 2 a ( cos ω 1 t + cos ω 2 t ) = a cos ( 2 ω 2 − ω 1 t ) cos ( 2 ω 1 + ω 2 t ) Result: a fast oscillation modulated by a slow envelope — beats . Energy fully transfers to mass 2 and back. Why this matters: weak coupling (k c ≪ k k_c\ll k k c ≪ k ω 1 ≈ ω 2 \omega_1\approx\omega_2 ω 1 ≈ ω 2
Worked example Exciting one pure modeWant only the symmetric mode? Set x 1 ( 0 ) = x 2 ( 0 ) = a x_1(0)=x_2(0)=a x 1 ( 0 ) = x 2 ( 0 ) = a η 2 ( 0 ) = 0 \eta_2(0)=0 η 2 ( 0 ) = 0 ω 1 = k / m \omega_1=\sqrt{k/m} ω 1 = k / m Why: picking an eigenvector as the initial condition gives single-frequency motion — the whole point of normal coordinates.
Common mistake "Each mass has its own frequency"Why it feels right: uncoupled, mass 1 would oscillate at one rate and mass 2 at another, so surely they keep them?
Fix: Coupling forces shared frequencies. In a normal mode all parts oscillate at ONE common ω \omega ω sum of these shared-frequency modes — not a single frequency per mass.
Common mistake Forgetting the factor of 2 inω 2 2 = ( k + 2 k c ) / m \omega_2^2 = (k+2k_c)/m ω 2 2 = ( k + 2 k c ) / m Why it feels right: you see k c k_c k c V V V k + k c k+k_c k + k c Fix: the antisymmetric stretch is ( x 1 − x 2 ) = 2 x 1 (x_1-x_2)=2x_1 ( x 1 − x 2 ) = 2 x 1 effective contribution doubles. Always derive from the secular determinant, never guess.
Common mistake Normalizing eigenvectors wrong / mixingM \mathbf{M} M Why it feels right: treating it like an ordinary eigenvalue problem K A = λ A \mathbf{K}\mathbf{A}=\lambda\mathbf{A} KA = λ A Fix: it is a generalized problem K A = ω 2 M A \mathbf{K}\mathbf{A}=\omega^2\mathbf{M}\mathbf{A} KA = ω 2 MA M \mathbf{M} M A a T M A b = 0 \mathbf{A}_a^T\mathbf{M}\mathbf{A}_b=0 A a T M A b = 0 a ≠ b a\neq b a = b M = m 1 \mathbf{M}=m\mathbb{1} M = m 1
Recall Feynman: explain to a 12-year-oldImagine two swings tied together with a rubber band. If you push them so they swing the same way at the same time, the rubber band just goes along — it's relaxed, and they swing at their normal lazy speed. If you push them opposite ways, the rubber band keeps stretching and snapping back, fighting them, so they swing faster. Those two special ways of swinging are "normal modes." Any other crazy swinging you start is really just a mix of these two special ways happening at once. By spotting the special ways, the confusing motion becomes two simple ones.
Mnemonic SLOW = Same, FAST = FightS ymmetric mode is S low (spring S leeps). Anti-symmetric is fast (spring fights ). ω 2 \omega^2 ω 2 = k / m =k/m = k / m = ( k + 2 k c ) / m =(k+2k_c)/m = ( k + 2 k c ) / m
What defines a normal mode? A collective motion in which every coordinate oscillates at a single common frequency in a fixed amplitude ratio (an eigenvector of
K − ω 2 M \mathbf{K}-\omega^2\mathbf{M} K − ω 2 M ).
What equation gives the normal-mode frequencies? The secular equation
det ( K − ω 2 M ) = 0 \det(\mathbf{K}-\omega^2\mathbf{M})=0 det ( K − ω 2 M ) = 0 .
For two equal masses with outer k k k k c k_c k c ω 1 2 = k / m \omega_1^2=k/m ω 1 2 = k / m (symmetric) and
ω 2 2 = ( k + 2 k c ) / m \omega_2^2=(k+2k_c)/m ω 2 2 = ( k + 2 k c ) / m (antisymmetric).
Why is the symmetric mode lower in frequency? Both masses move together so the coupling spring never stretches; only
k k k provides restoring force.
What is a normal coordinate? A linear combination of the original coordinates that obeys an uncoupled SHM equation
η ¨ a = − ω a 2 η a \ddot\eta_a=-\omega_a^2\eta_a η ¨ a = − ω a 2 η a .
Normal coordinates for two equal masses? η 1 , 2 = 1 2 ( x 1 ± x 2 ) \eta_{1,2}=\frac{1}{\sqrt2}(x_1\pm x_2) η 1 , 2 = 2 1 ( x 1 ± x 2 ) .
How many normal modes does an N-DOF system have? Exactly N.
What causes 'beats' in coupled oscillators? Superposition of two modes with nearly equal frequencies; energy slowly transfers between masses with envelope frequency
( ω 2 − ω 1 ) / 2 (\omega_2-\omega_1)/2 ( ω 2 − ω 1 ) /2 .
How do you excite a single pure mode? Choose initial conditions equal to that mode's eigenvector (e.g.
x 1 = x 2 x_1=x_2 x 1 = x 2 for symmetric).
What orthogonality do eigenvectors obey? Mass-weighted:
A a T M A b = 0 \mathbf{A}_a^T\mathbf{M}\mathbf{A}_b=0 A a T M A b = 0 for
a ≠ b a\neq b a = b .
Lagrangian Mechanics — how we derived the EOMs without force-sign errorsSimple Harmonic Motion — each normal coordinate IS an SHMEigenvalues and Eigenvectors — secular equation = generalized eigenvalue problemBeats and Superposition — observable consequence of two close modesSmall Oscillations — Taylor-expanding V V V K \mathbf{K} K Phonons and Lattice Vibrations — N→∞ limit of coupled oscillators in solids
Cross terms mixing coords
Lagrangian L equals T minus V
Coupled equations of motion
Algebraic matrix equation
K minus w2 M times A equals 0
Secular equation det equals 0
Superposition of real motion
Intuition Hinglish mein samjhoSocho do masses ek spring se jude hain — dono walls se bhi springs se attached. Agar tum ek mass ko hilao, energy doosre mein chali jaati hai, aur motion bahut complicated lagta hai. Lekin ek shaandaar baat hai: koi bhi aisa coupled system hamesha kuch independent simple oscillators ka mixture hota hai — inhe normal modes kehte hain. Har mode ki apni ek single frequency hoti hai.
Yahan do special modes nikalte hain. Symmetric mode mein dono masses ek hi direction mein same amount move karte hain, isse beech wali (coupling) spring kabhi stretch hi nahi hoti — woh sota rehta hai — isliye frequency sirf k / m \sqrt{k/m} k / m Antisymmetric mode mein dono opposite move karte hain, beech wali spring sabse zyada fight karti hai, isliye stiffer feel hoti hai aur frequency zyada: ( k + 2 k c ) / m \sqrt{(k+2k_c)/m} ( k + 2 k c ) / m
Trick yeh hai ki hum naye coordinates banate hain: η 1 = ( x 1 + x 2 ) / 2 \eta_1=(x_1+x_2)/\sqrt2 η 1 = ( x 1 + x 2 ) / 2 η 2 = ( x 1 − x 2 ) / 2 \eta_2=(x_1-x_2)/\sqrt2 η 2 = ( x 1 − x 2 ) / 2 normal coordinates mein equations completely alag-alag ho jaate hain — har ek bas ek seedha SHM ban jaata hai. Phir actual motion in dono SHM ka simple sum hai. Agar tum sirf ek mass ko push karke chhodo, dono modes thode-thode excite ho jaate hain aur energy slowly idhar-udhar slosh karti hai — yahi beats hain. Yeh idea molecules, bridges, crystals — sab vibrations samajhne ka 80/20 hai.
