Modes of oscillation

N N plastic beads of mass m m are placed on a circle of radius R R so that they can move on the circle without friction. The beads are connected by identical massless springs. In the equilibrium state the springs are of equal lengths and unstressed. We number the beads sequentially from 1 to N N . The energy change of the spring between the bead at position j j and position j 1 j-1 is E = 1 2 k ( x j x j 1 ) 2 E=\frac{1}{2}k (x_j-x_{j-1})^2 , where x j x_j is the displacement of bead j j from its equilibrium position. (The first and last bead is also connected with a similar spring.)

If we randomly kick the beads they will oscillate around their equilibrium position. In the general case this oscillation is rather complex, but there are a few frequencies and initial conditions when the motion of each bead is a simple harmonic oscillation. What will be the the lowest possible frequency of oscillation f f for the collection of beads? Provide the answer for N = 10 N=10 , m = 1 m=1 g, R = 10 R=10 cm and k = 0.1 k=0.1 N/m to three significant figures, in units of frequency (Hz).

Notes:

-- f = 0 f=0 is a solution to the equations of motions, but we are not considering it an oscillation.

-- The highest frequency is easy to determine. Initially we move the even-numbered beads clockwise and the odd-numbered beads counter-clockwise by the same amount. When the two sets of beads are released, they will oscillate against each other. The center of every spring will be motionless. Therefore we may just fix it for good, and the system will fall apart into N N independent units. The angular frequency will be ω = 4 k / m = 20 s e c 1 \omega=\sqrt{4k/m}=20sec^{-1} , where the factor 4 = 2 × 2 4=2\times 2 is coming from two sources. First, every spring is cut to half of its original length, so it is twice as stiff. Second, every bead is connected to two springs. The frequency is f = ω / ( 2 π ) = 3.18 f=\omega/(2\pi)=3.18 Hz.


The answer is 0.984.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Laszlo Mihaly
Nov 13, 2017

The equation of motion of bead j j is

m x ¨ j = k ( x j x j 1 x j + 1 ) m\ddot x_j= -k(x_j-x_{j-1}-x_{j+1})

We look for the solution in the form of

x j = A e i ( κ j ω t ) x_j= A e^{ i(\kappa j-\omega t)}

and we get

m ω 2 = k ( 2 e i κ e i κ ) = 2 k ( 1 cos κ ) -m \omega^2 = -k (2-e^{i\kappa}-e^{-i\kappa})=-2k(1-\cos \kappa)

Here A A and κ \kappa are free parameters. The parameter κ \kappa is constrained by the fact that if we go around the circle, we need to get the same displacement: x j = x j + N x_j=x_{j+N} . Therefore e i κ N = 1 e^{i\kappa N}=1 and κ = 2 π n / N \kappa=2\pi n/N , where n n is an integer. The lowest frequency is with n = 1 n=1

ω = 2 k m ( 1 cos 2 π N ) = 6.18 s e c 1 \omega= \sqrt{\frac{2k}{m}\left(1-\cos \frac{2\pi}{N}\right)}= 6.18 sec^{-1}

f = ω / 2 π = 0.984 f=\omega/2 \pi= 0.984 Hz

We get the highest frequency with n = 5 n=5 . For n = 0 n=0 we get zero frequency; that corresponds to the uniform sliding of all beads in the same direction.

You have shown that solutions exist of the form you want. How are you sure that there are no solutions of a different form (which might give more extreme values for ω \omega ?)

Mark Hennings - 3 years, 6 months ago

Log in to reply

One can count the number of normal modes. For N =2M (even) the parameter n is from 0 to M. There is 1 eigenvector for n=0 and n=M each. There are 2 eigenvectors for the other values. That gives N modes, as expected. For N=2M+1 there are two eigenvectors for n=M, and that gives N modes again.

Laszlo Mihaly - 3 years, 6 months ago

Log in to reply

Indeed. You do not go into this in your proof...

Mark Hennings - 3 years, 6 months ago

Log in to reply

@Mark Hennings I am coming more from the physics side to this. BTW, the trick with the periodic boundary condition is used extensively in the description of lattice vibrations in condensed matter physics, except it is applied to N = 1 0 23 N=10^{23} particle.

Laszlo Mihaly - 3 years, 6 months ago
Mark Hennings
Nov 14, 2017

Suppose that N = 2 M N = 2M is even, and consider the N × N N \times N symmetric matrix U ( N ) U(N) , where U ( N ) u v = { 1 u v 1 m o d N 0 otherwise U(N)_{uv} \; = \; \left\{ \begin{array}{lll} 1 & \hspace{1cm} & |u-v| \equiv 1 \; \mathrm{mod}\; N \\ 0 & & \mbox{otherwise} \end{array} \right. Then y \mathbf{y} will be an eigenvector of U ( N ) U(N) with eigenvalue λ \lambda provided that y u 1 + y u + 1 = λ y u 2 u N 1 y_{u-1} + y_{u+1} \;= \; \lambda y_u \hspace{2cm} 2 \le u \le N-1 and in addition provided that y N + y 2 = λ y 1 y N 1 + y 1 = λ y N y_N + y_2 \; = \; \lambda y_1 \hspace{3cm} y_{N-1} + y_1 \; = \; \lambda y_N If we write λ = 2 cos θ \lambda = 2\cos\theta then the main recurrence relation is satisfied by y j = e i j θ 1 j N y_j \; = \; e^{ij\theta} \hspace{2cm} 1 \le j \le N and the additional two equations will also be true provided that e i N θ = 1 e^{iN\theta} = 1 , and so provided that θ = 2 α π N = α π M \theta = \tfrac{2\alpha\pi}{N} = \tfrac{\alpha\pi}{M} for some integer α \alpha . Since U ( N ) U(N) is a symmetric matrix, it has a basis of eigenvectors:

  • The eigenvalue λ 0 = 2 \lambda_0 = 2 has associated eigenvector y ( 0 ) \mathbf{y}(0) , where y ( 0 ) j = 1 1 j N \mathbf{y}(0)_j \; = \; 1 \hspace{2cm} 1 \le j \le N
  • For any 1 α M 1 1 \le \alpha \le M-1 , the eigenvalue λ α = 2 cos ( α π M ) \lambda_\alpha = 2\cos\big(\tfrac{\alpha\pi}{M}\big) has two associated eigenvectors y ( α , 1 ) \mathbf{y}(\alpha,1) and y ( α , 2 ) \mathbf{y}(\alpha,2) , where y ( α , 1 ) j = cos ( π i α j M ) y ( α , 2 ) j = sin ( π i α j M ) 1 j M 1 \mathbf{y}(\alpha,1)_j \; = \; \cos\big(\tfrac{\pi i\alpha j}{M}\big) \hspace{1cm} \mathbf{y}(\alpha,2)_j \; = \; \sin\big(\tfrac{\pi i\alpha j}{M}\big) \hspace{2cm} 1 \le j \le M-1
  • The eigenvalue λ M = 2 \lambda_M = -2 has associated eigenvector y ( M ) \mathbf{y}(M) , where y ( M ) j = ( 1 ) j 1 j N \mathbf{y}(M)_j \; = \; (-1)^j \hspace{2cm} 1 \le j \le N

Assuming that there is a spring joining the first and N N th particles, the kinetic energy of the system is T = 1 2 m j = 1 N x ˙ j 2 T \; = \; \tfrac12m\sum_{j=1}^N \dot{x}_j^2 and the kinetic energy of the system is V = 1 2 k j = 1 N 1 ( x j + x j + 1 ) 2 + 1 2 k ( x N x 1 ) 2 V \; = \; \tfrac12k\sum_{j=1}^{N-1} (x_j + x_{j+1})^2 + \tfrac12k(x_N - x_1)^2 Lagrange's equations for this system can be written as m x ¨ 1 + k ( x N + 2 x 1 x 2 ) = 0 m x ¨ 2 + k ( x 1 + 2 x 2 x 3 ) = 0 m x ¨ 3 + k ( x 2 + 2 x 3 x 4 ) = 0 m x ¨ N 1 + k ( x N 2 + 2 x N 1 + x N ) = 0 m x ¨ N + k ( x N 1 + 2 x N x 1 ) = 0 \begin{aligned} m\ddot{x}_1 + k(-x_N + 2x_1 - x_2) & = \; 0 \\ m\ddot{x}_2 + k(-x_1 + 2x_2 - x_3) & = \; 0 \\ m\ddot{x}_3 + k(-x_2 + 2x_3 - x_4) & = \; 0\\ \cdots & \hspace{0.5cm} \cdots \\ m\ddot{x}_{N-1} + k(-x_{N-2} + 2x_{N-1} + x_N)& = \; 0 \\ m\ddot{x}_N + k(-x_{N-1} + 2x_N - x_1) & = \; 0 \end{aligned} or, more compactly, as m x ¨ + 2 k x k U ( N ) x = 0 m \ddot{\mathbf{x}} + 2k\mathbf{x} - kU(N)\mathbf{x} \; = \; \mathbf{0} We look for normal modes of oscillation of the form x = e i ω t y \mathbf{x} \,= \, e^{i\omega t}\mathbf{y} for a constant vector y \mathbf{y} , which means we require that m ω 2 y + 2 k y k U ( N ) y = 0 -m\omega^2\mathbf{y} + 2k\mathbf{y} - kU(N)\mathbf{y} \; = \; \mathbf{0} or, equivalently, U ( N ) y = ( 2 m ω 2 k ) y U(N)\mathbf{y} \; = \; \left(2 - \tfrac{m\omega^2}{k}\right)\mathbf{y} so that y \mathbf{y} is an eigenvector of U ( N ) U(N) with associated eigenvalue 2 m ω 2 k 2 - \tfrac{m\omega^2}{k} . Thus the possible values of ω \omega are given by the equation 2 m ω 2 k = λ α = 2 cos ( α π M ) 2 - \tfrac{m\omega^2}{k} = \lambda_\alpha = 2\cos\big(\tfrac{\alpha\pi}{M}\big) for 0 α M 0 \le \alpha \le M , so that ω = 2 sin ( α π 2 M ) k m = 2 sin ( α π N ) k m 0 α M \omega \; = \; 2\sin\big(\tfrac{\alpha\pi}{2M}\big)\sqrt{\tfrac{k}{m}} \; = \; 2\sin\big(\tfrac{\alpha\pi}{N}\big)\sqrt{\tfrac{k}{m}} \hspace{2cm} 0 \le \alpha \le M The largest possible value of ω \omega is given by α = M \alpha = M , and is 2 k m 2\sqrt{\tfrac{k}{m}} . The value ω = 0 \omega=0 can be ignored, since it corresponds to y j = 1 y_j=1 for all j j , which happens when no oscillations occur at all, and the particles just neatly rotate around the circle.

Thus the smallest possible value of ω \omega is 2 sin ( π N ) k m 2\sin\big(\tfrac{\pi}{N}\big)\sqrt{\tfrac{k}{m}} .

In this case, N = 10 N=10 , k = 0.1 k=0.1 , m = 0.001 m=0.001 , this leads to the smallest value of ω \omega being 20 sin π 10 = 6.180339887 20\sin\tfrac{\pi}{10} = 6.180339887 , and the smallest possible frequency being 0.9836316431 \boxed{0.9836316431} Hz.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...