betaFeedback

WorksMathematics

The effect of finite duration sources on modes and generalization of the d'Alembert solution

Abstract

We investigate the evolution of dispersive waves governed by linear wave equations, where a finite duration source is applied. The resulting wave may be viewed as the superposition of modes before the source is turned on and after it is turned off. We consider the problem of relating the modes after the source term is turned off to the modes before the source term was turned on. We obtain explicit formulas in both the wavenumber and position representations. A number of special cases are considered. Using the methods presented, we obtain a generalization of the d'Alembert solution which applies to linear wave equations with constant coefficients.

Introduction

For propagating dispersive waves governed by linear wave equations with constant coefficients and without any source terms, the wave may be decomposed into modes. However, if there is a source term of finite duration, then for the time that the source term is on, the wave cannot be described by modes. In particular suppose the governing equation is [1, 2]

Lu(x,t)={0for t<−T,f(x,t)for −T<t<0,0for t>0.(1)\mathcal{L}u(x,t)= \begin{cases} 0 & \text{for } t<-T,\\ f(x,t) & \text{for } -T<t<0,\\ 0 & \text{for } t>0. \end{cases} \tag*{(1)}

where u(x,t)u(x,t) is the wave, f(x,t)f(x,t) is the source term, and the linear wave equation, L\mathcal{L}, is given by

L=∑n=0Nxbn∂n∂xn−∑n=0Ntan∂n∂tn(2)\mathcal{L}=\sum_{n=0}^{N_x}b_n\frac{\partial^n}{\partial x^n}-\sum_{n=0}^{N_t}a_n\frac{\partial^n}{\partial t^n} \tag*{(2)}

That is, the wave evolves without the source term up to time t=−Tt=-T, at which time, a source term f(x,t)f(x,t) is applied up to a time t=0t=0, and is then turned off. The resulting wave may be described by modes only before the source term is turned on and after it is turned off. We consider the problem of relating the modes after the source term is turned off to the modes before the source term was turned on. That is, we investigate the effect of the finite duration source term on modes.

In the next section we describe the method of modes and also discuss how to obtain the modes given the initial conditions for the wave.

Method of Modes

For waves governed by a wave equation

Lu(x,t)=0(3)\mathcal{L}u(x,t)=0 \tag*{(3)}

where

L=∑n=0Nxbn∂n∂xn−∑n=0Ntan∂n∂tn(4)\mathcal{L}=\sum_{n=0}^{N_x}b_n\frac{\partial^n}{\partial x^n}-\sum_{n=0}^{N_t}a_n\frac{\partial^n}{\partial t^n} \tag*{(4)}

one substitutes u=eikx−iωtu=e^{ikx-i\omega t} into Eq. (3)

Leikx−iωt=0(5)\mathcal{L}e^{ikx-i\omega t}=0 \tag*{(5)}

to obtain the dispersion relation [3, 4, 5, 6, 7, 8]

∑n=0Nxbn(ik)n−∑n=0Ntan(−iω)n=0(6)\sum_{n=0}^{N_x}b_n(ik)^n-\sum_{n=0}^{N_t}a_n(-i\omega)^n=0 \tag*{(6)}

One may solve Eq. (6) for ω\omega as a function of kk, or for kk as a function of ω\omega; the choice depends on the physical situation. For example, if we pluck a string at a particular time and let go, then one would solve for ω\omega as a function of kk. This corresponds to when u(x,0)u(x,0) is the initial condition. Alternatively, if we are at a fixed position and create a pulse as a function of time, u(0,t)u(0,t), then one would solve for kk as a function of ω\omega. In this paper, we deal with the former case where we solve for ω\omega as a function of kk.

Generally speaking, there is more than one solution to the dispersion relation. Assuming that there are MM solutions, we write

ω=ωℓ(k),for ℓ=1,2,⋯ ,M;(7)\omega=\omega_\ell(k),\qquad\text{for }\ell=1,2,\cdots,M; \tag*{(7)}

each ωℓ(k)\omega_\ell(k) may be complex. When the source term is zero, the wave is then given by the sum of modes

u(x,t)=∑ℓ=1Muℓ(x,t)(8)u(x,t)=\sum_{\ell=1}^{M}u_\ell(x,t) \tag*{(8)}

where uℓ(x,t)u_\ell(x,t) are the modes; each corresponds to a solution of the dispersion relation, ωℓ(k)\omega_\ell(k). The properties and evolution of the modes and methods for obtaining them are described in § 7, where we also discuss how to obtain the initial conditions of the modes from the initial conditions for the wave.

Before the source is turned on, the wave is the superposition of modes

u(x,t)=∑ℓ=1MuℓB(x,t),for t<−T(9)u(x,t)=\sum_{\ell=1}^{M}u_\ell^{B}(x,t),\qquad\text{for }t<-T \tag*{(9)}

where uℓB(x,t)u_\ell^{B}(x,t) are the mode before the source is turned on.

While the source is on, we write the solution as

u(x,t)=μ(x,t)for −T<t<0(10)u(x,t)=\mu(x,t)\qquad\text{for }-T<t<0 \tag*{(10)}

After the source is turned off, we can again express the solution in terms of modes

u(x,t)=∑ℓ=1MuℓA(x,t)for t>0(11)u(x,t) = \sum_{\ell=1}^{M} u_\ell^A(x,t) \quad\text{for } t > 0 \tag*{(11)}

where uℓAu_\ell^A are the modes after the source is tuned off.

As mentioned in the Introduction, in this paper, we address the question of how modes are affected by the source term. That is, we consider the relation of the modes after the source uℓAu_\ell^A to the modes before the source uℓBu_\ell^B.

Notation:

All integrals range from −∞-\infty to ∞\infty unless otherwise noted.

u(x,t)u(x,t) and S(k,t)S(k,t) are the wave expressed in position space and wavenumber space (which is discussed in the next section).

uℓBu_\ell^B and uℓAu_\ell^A are the modes before and after the source, respectively.

SℓB(x,t)S_\ell^B(x,t) and SℓA(x,t)S_\ell^A(x,t) are the modes before and after the source expressed in wavenumber space, respectively.

f(x,t)f(x,t) and F(k,t)F(k,t) are the source term, expressed in position and wavenumber space, respectively.

up(k,t)u_p(k,t) and Sp(k,t)S_p(k,t) are particular solutions in the time interval −T<t<0-T < t < 0.

φ(k,t)=S(k,t)for −T<t<0(12)\varphi(k,t) = S(k,t) \quad\text{for } -T < t < 0 \tag*{(12)}
μ(x,t)=u(x,t)for −T<t<0(13)\mu(x,t) = u(x,t) \quad\text{for } -T < t < 0 \tag*{(13)}

As usual, expressions such as ∂∂tS2A(k,t)∣t=T\left.\frac{\partial}{\partial t}S_2^A(k,t)\right|_{t=T} mean ∂∂tS2A(k,t)∣t=T\frac{\partial}{\partial t}S_2^A(k,t)\big|_{t=T}.

For expressions such as

∂∂teiω2tS1B(k,t=0)(14)\frac{\partial}{\partial t}e^{i\omega_2t}S_1^B(k,t=0) \tag*{(14)}

we mean that the differentiation is done on eiω2tS1B(k,t)e^{i\omega_2t}S_1^B(k,t) and after the differentiation tt is set to zero only on the term indicated by t=0t=0.

We define wavenumber operator, K\mathcal{K} in the position representation by

K=1i∂∂x(15)\mathcal{K} = \frac{1}{i}\frac{\partial}{\partial x} \tag*{(15)}

Wavenumber formulation

The wavenumber representation makes it easier to relate the modes after the source term is tuned off to the modes before the source term is turned on. The fundamental reason is that in the wavenumber formulation the equation of motion for modes is relatively simple as we now show. The wave in wavenumber space, S(k,t)S(k,t), is

S(k,t)=12π∫u(x,t)e−ikx dx(16)S(k,t) = \frac{1}{\sqrt{2\pi}}\int u(x,t)e^{-ikx}\,dx \tag*{(16)}

and each mode is defined by

uℓ(x,t)=12π∫Sℓ(k,0)eikx−iωℓ(k)t dk(17)u_\ell(x,t) = \frac{1}{\sqrt{2\pi}}\int S_\ell(k,0)e^{ikx-i\omega_\ell(k)t}\,dk \tag*{(17)}

where Sℓ(k,0)S_\ell(k,0) is the initial wavenumber mode, obtained from the spatial mode uℓ(x,0)u_\ell(x,0) by way of

Sℓ(k,0)=12π∫uℓ(x,0)e−ikx dx(18)S_\ell(k,0) = \frac{1}{\sqrt{2\pi}}\int u_\ell(x,0)e^{-ikx}\,dx \tag*{(18)}

If one defines the time-dependent wavenumber mode as [9]

Sℓ(k,t)=e−iωℓ(k)tSℓ(k,0)(19)S_\ell(k,t) = e^{-i\omega_\ell(k)t}S_\ell(k,0) \tag*{(19)}

then uℓ(x,t)u_\ell(x,t) and Sℓ(k,t)S_\ell(k,t) form Fourier transform pairs between xx and kk for all time

uℓ(x,t)=12π∫Sℓ(k,t)eikx dk(20)u_\ell(x,t) = \frac{1}{\sqrt{2\pi}}\int S_\ell(k,t)e^{ikx}\,dk \tag*{(20)}
Sℓ(k,t)=12π∫uℓ(x,t)e−ikx dx(21)S_\ell(k,t) = \frac{1}{\sqrt{2\pi}}\int u_\ell(x,t)e^{-ikx}\,dx \tag*{(21)}

Each of the modes in wavenumber space evolve in a simple manner, namely Eq. (19), and satisfy the equation of motion [10]

i∂∂tSℓ(k,t)=ωℓ(k)Sℓ(k,t)(22)i\frac{\partial}{\partial t}S_\ell(k,t) = \omega_\ell(k)S_\ell(k,t) \tag*{(22)}

In the wavenumber representation, the governing equation of the wave corresponding to Eq. (1) is

References

  1. [1]P. Morse, H. Feshbach, Methods of Theoretical Physics Part I & II, McGraw-Hill, 1953.DOI
  2. [2]Y. Pinchover, J. Rubinstein, An Introduction to Partial Differential Equations, Cambridge University Press, 2005.DOI
  3. [3]K. Graff, Wave Motion in Elastic Solids, Oxford University Press, 1975.
  4. [4]J. D. Jackson, Classical Electrodynamics, Wiley, 1992.
  5. [5]J. Lighthill, Waves in Fluids, Cambridge University Press, 1978.
  6. [6]P. H. Morse and K. U. Ingard, Theoretical Acoustics, McGraw-Hill, 1968.
  7. [7]I. Tolstoy, C. S. Clay, Ocean Acoustics: Theory and Experiment in Underwater Sound, Acoustical Society of America, 1966.
  8. [8]G. B. Whitham, Linear and Nonlinear Waves, Wiley, 1974.
  9. [9]P. Loughlin and L. Cohen, “A Wigner approximation method for wave propagation,” J. Acoust. Soc. Amer. vol. 118, no. 3, pp. 1268-1271, 2005.DOI
  10. [10]L. Cohen, “Phase-Space Differential Equations for Modes”. In: Schulze, BW., Wong, M.W. (eds) Pseudo-Differential Operators: Complex Analysis and Partial Differential Equations. Operator Theory: Advances and Applications, vol 205. Birkhäuser Basel, 2009.DOI
  11. [11]J. S. Ben-Benjamin and L. Cohen, “Pulse propagation and windowed wave functions”, Journal of Modern Optics, vol. 61, iss. 1, pp. 36-42, 2014.
  12. [12]J. L. Davis , Mathematics of Wave Propagation, Princeton University Press, Princeton NJ, 2000.
  13. [13]D. R. Bland, Wave Theory and Applications, Oxford University Press, New York, 1988.

Paper details

vlnc:W_vd438gjf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents