Wave Mechanics: Right- and Left-Traveling SolutionsWaveMechanicsRightAndLeftTravelingSolutions2026-09-12 15:11:482026-09-12 15:11:48Topicwave mechanicsone-dimensional wave equationright-moving waveleft-moving wavecharacteristic coordinatescharacteristicstraveling-wave decompositiontransport equationfactorizationd'Alembert solutionsuperposition% this is the default PhysicsLibrary preamble. as your knowledge
% of TeX increases, you will probably want to edit this, but
% it should be fine as is for beginners.
% almost certainly you want these
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{amsfonts}
% used for TeXing text within eps files
%\usepackage{psfrag}
% need this for including graphics (\includegraphics).
% If you use \includegraphics{figure} with no file extension, PhysicsLibrary
% will prefer figure.png before falling back to figure.eps.
\usepackage{graphicx}
\DeclareGraphicsExtensions{.png,.jpg,.jpeg,.pdf,.eps}
% for neatly defining theorems and propositions
%\usepackage{amsthm}
% making logically defined graphics
%\usepackage{xypic}
% there are many more packages, add them here as you need them
% define commands here\section*{Wave Mechanics: Right- and Left-Traveling Solutions}
WM15 verified that every sufficiently smooth profile of the form
\begin{equation}
F(x-ct)
\end{equation}
satisfies the one-dimensional wave equation
\begin{equation}
\boxed{u_{tt}=c^2u_{xx}},
\end{equation}
and that the same is true for every sufficiently smooth profile of the form
\begin{equation}
G(x+ct).
\end{equation}
WM16 now asks the deeper structural question:
\begin{equation}
\boxed{\text{Why does the wave equation naturally contain two propagation directions?}}
\end{equation}
The answer is that the second-order wave operator can be separated into two first-order propagation operators. Those two operators correspond to information moving at speeds $+c$ and $-c$. In characteristic coordinates, the wave equation then reduces to a particularly simple mixed-derivative equation whose sufficiently smooth solutions have the form
\begin{equation}
\boxed{u(x,t)=F(x-ct)+G(x+ct).}
\end{equation}
This is the structural form underlying the classical d'Alembert solution of the one-dimensional wave equation \cite{French1971,Crawford1968,Feynman47,MIT803L10}.
WM16 derives this two-family structure. WM17 will take the next step: given an initial displacement and initial velocity, determine the specific functions $F$ and $G$.
\section{One equation supports two directions}
The homogeneous one-dimensional wave equation is
\begin{equation}
u_{tt}-c^2u_{xx}=0.
\end{equation}
WM15 showed directly that
\begin{equation}
u_R(x,t)=F(x-ct)
\end{equation}
moves toward increasing $x$, while
\begin{equation}
u_L(x,t)=G(x+ct)
\end{equation}
moves toward decreasing $x$.
The two families are illustrated below.
\begin{center}
\includegraphics{WM16_fig01_two_traveling_families.png}
\vspace{0.45em}
\textbf{Figure.}
The same shape-preserving idea produces two propagation families. A profile depending on $x-ct$ moves toward increasing $x$; a profile depending on $x+ct$ moves toward decreasing $x$.
\end{center}
The speed magnitude is the same in both cases. Only the direction changes.
\section{The first-order equations retain direction information}
For a right-moving wave
\begin{equation}
u_R(x,t)=F(x-ct),
\end{equation}
WM15 found
\begin{equation}
u_t=-cu_x.
\end{equation}
Therefore
\begin{equation}
\boxed{u_t+cu_x=0\qquad\text{for a pure right-moving wave}.}
\end{equation}
For a left-moving wave
\begin{equation}
u_L(x,t)=G(x+ct),
\end{equation}
we have
\begin{equation}
u_t=+cu_x,
\end{equation}
so
\begin{equation}
\boxed{u_t-cu_x=0\qquad\text{for a pure left-moving wave}.}
\end{equation}
These are first-order transport equations. Unlike the second-order wave equation, they remember the propagation direction.
The distinction can be summarized as
\begin{equation}
\boxed{
\begin{array}{ll}
u_t+cu_x=0 & \Longrightarrow \text{right-moving},\\[0.4em]
u_t-cu_x=0 & \Longrightarrow \text{left-moving}.
\end{array}}
\end{equation}
This interpretation is standard in treatments of traveling-wave solutions and characteristics \cite{Feynman47,MIT803L10}.
\section{Factor the wave operator}
Because $c$ is constant and the partial derivatives commute for a sufficiently smooth function, we may write
\begin{align}
\left(\frac{\partial}{\partial t}-c\frac{\partial}{\partial x}\right)
\left(\frac{\partial}{\partial t}+c\frac{\partial}{\partial x}\right)u
&=u_{tt}+cu_{tx}-cu_{xt}-c^2u_{xx}\\
&=u_{tt}-c^2u_{xx}.
\end{align}
Thus the wave equation can be factored as
\begin{equation}
\boxed{
\left(\frac{\partial}{\partial t}-c\frac{\partial}{\partial x}\right)
\left(\frac{\partial}{\partial t}+c\frac{\partial}{\partial x}\right)u=0.}
\end{equation}
This factorization is the PDE analogue of factoring an algebraic expression such as
\begin{equation}
a^2-b^2=(a-b)(a+b).
\end{equation}
It hints that two first-order propagation mechanisms are embedded inside the second-order wave equation.
\section{Characteristic coordinates}
Introduce the coordinates
\begin{equation}
\boxed{\xi=x-ct}
\end{equation}
and
\begin{equation}
\boxed{\eta=x+ct.}
\end{equation}
The coordinate $\xi$ stays constant along a right-moving feature. The coordinate $\eta$ stays constant along a left-moving feature.
If
\begin{equation}
\xi=\text{constant},
\end{equation}
then
\begin{equation}
x=ct+\text{constant},
\end{equation}
so
\begin{equation}
\frac{dx}{dt}=+c.
\end{equation}
If
\begin{equation}
\eta=\text{constant},
\end{equation}
then
\begin{equation}
x=-ct+\text{constant},
\end{equation}
so
\begin{equation}
\frac{dx}{dt}=-c.
\end{equation}
\begin{center}
\includegraphics{WM16_fig02_characteristic_coordinates.png}
\vspace{0.45em}
\textbf{Figure.}
The two families of characteristic lines in the $x$-$t$ plane. Solid lines carry constant $\xi=x-ct$ and move toward increasing $x$; dashed lines carry constant $\eta=x+ct$ and move toward decreasing $x$.
\end{center}
These lines are called \emph{characteristics}. They organize how information propagates through the solution.
\section{Rewrite the wave equation in characteristic coordinates}
Let
\begin{equation}
u(x,t)=U(\xi,\eta).
\end{equation}
Since
\begin{equation}
\xi_x=1,
\qquad
\eta_x=1,
\end{equation}
we obtain
\begin{equation}
u_x=U_{\xi}+U_{\eta}.
\end{equation}
Differentiating again,
\begin{equation}
u_{xx}=U_{\xi\xi}+2U_{\xi\eta}+U_{\eta\eta}.
\end{equation}
Similarly,
\begin{equation}
\xi_t=-c,
\qquad
\eta_t=+c,
\end{equation}
so
\begin{equation}
u_t=-cU_{\xi}+cU_{\eta}.
\end{equation}
Differentiating again,
\begin{equation}
u_{tt}=c^2U_{\xi\xi}-2c^2U_{\xi\eta}+c^2U_{\eta\eta}.
\end{equation}
Therefore
\begin{align}
u_{tt}-c^2u_{xx}
&=c^2U_{\xi\xi}-2c^2U_{\xi\eta}+c^2U_{\eta\eta}\\
&\quad-c^2\left(U_{\xi\xi}+2U_{\xi\eta}+U_{\eta\eta}\right)\\
&=-4c^2U_{\xi\eta}.
\end{align}
The wave equation therefore becomes
\begin{equation}
-4c^2U_{\xi\eta}=0.
\end{equation}
For $c\neq 0$,
\begin{equation}
\boxed{U_{\xi\eta}=0.}
\end{equation}
This is much simpler than the original PDE.
\section{Integrate the transformed equation}
The equation
\begin{equation}
U_{\xi\eta}=0
\end{equation}
means
\begin{equation}
\frac{\partial}{\partial \xi}\left(U_{\eta}\right)=0.
\end{equation}
Therefore $U_{\eta}$ cannot depend on $\xi$. It can depend only on $\eta$:
\begin{equation}
U_{\eta}=H(\eta).
\end{equation}
Integrating with respect to $\eta$ gives
\begin{equation}
U(\xi,\eta)=G(\eta)+F(\xi),
\end{equation}
where the ``constant of integration'' with respect to $\eta$ may still be an arbitrary function of $\xi$.
Returning to $x$ and $t$,
\begin{equation}
\boxed{u(x,t)=F(x-ct)+G(x+ct).}
\end{equation}
Thus, on a suitable domain and for sufficiently smooth functions, the general solution of the homogeneous one-dimensional constant-speed wave equation is the sum of one right-moving profile and one left-moving profile \cite{Feynman47,MIT803L10}.
\section{What ``general solution'' means here}
The statement
\begin{equation}
u(x,t)=F(x-ct)+G(x+ct)
\end{equation}
is stronger than the verification result in WM15.
WM15 showed:
\begin{equation}
\boxed{\text{if }u=F(x-ct)\text{ or }u=G(x+ct),\text{ then the PDE is satisfied}.}
\end{equation}
WM16 shows, under the usual smoothness assumptions for a classical solution,
\begin{equation}
\boxed{\text{every solution can be represented locally by a sum of these two families}.}
\end{equation}
The arbitrary functions $F$ and $G$ have not yet been determined. That requires initial or boundary data.
This distinction is essential:
\begin{equation}
\boxed{\text{PDE structure tells us the form; data choose the particular solution}.}
\end{equation}
\section{The physical displacement is the sum of the two components}
Suppose
\begin{equation}
u_R(x,t)=F(x-ct)
\end{equation}
and
\begin{equation}
u_L(x,t)=G(x+ct).
\end{equation}
Then the physical field is
\begin{equation}
u(x,t)=u_R(x,t)+u_L(x,t).
\end{equation}
The addition is point by point.
\begin{center}
\includegraphics{WM16_fig03_two_components_sum.png}
\vspace{0.45em}
\textbf{Figure.}
At any fixed time, the observed displacement is the point-by-point sum of a right-moving component and a left-moving component. Linearity allows both components to coexist without changing the governing equation.
\end{center}
Because the wave equation is linear, the two components propagate independently in the ideal model even while their sum may display interference.
\section{Standing waves fit naturally into the two-family picture}
Take equal-amplitude sinusoidal components
\begin{align}
u_R(x,t)&=A\cos(kx-\omega t),\\
u_L(x,t)&=A\cos(kx+\omega t).
\end{align}
For the ideal wave equation,
\begin{equation}
\omega=ck.
\end{equation}
Adding the two components gives
\begin{align}
u(x,t)
&=A\cos(kx-\omega t)+A\cos(kx+\omega t)\\
&=2A\cos(kx)\cos(\omega t).
\end{align}
Therefore the standing waves studied earlier are not a separate species of solution. They are a particular superposition of equal right- and left-moving components.
This connects the standing-wave material of WM10 directly to the two-family structure of the wave equation.
\section{A compact structural map}
\begin{center}
\includegraphics{WM16_fig04_solution_structure.png}
\vspace{0.45em}
\textbf{Figure.}
The logical chain from the second-order wave equation to its right- and left-moving solution families. WM17 will determine the two arbitrary functions from initial data.
\end{center}
\section{Worked example 1: identify both traveling components}
Consider
\begin{equation}
u(x,t)=\exp\left[-(x-4t)^2\right]
+\frac{1}{2}\exp\left[-(x+4t-2)^2\right].
\end{equation}
Identify the direction and speed of each component and state the wave equation it satisfies.
The first term is
\begin{equation}
F(x-4t),
\end{equation}
so it moves toward increasing $x$ at speed
\begin{equation}
\boxed{4\,\text{m/s}}.
\end{equation}
The second term can be written as
\begin{equation}
G(x+4t),
\end{equation}
where
\begin{equation}
G(\eta)=\frac{1}{2}e^{-(\eta-2)^2}.
\end{equation}
Therefore it moves toward decreasing $x$ at the same speed magnitude,
\begin{equation}
\boxed{4\,\text{m/s}}.
\end{equation}
Because each term separately satisfies the wave equation and the equation is linear, their sum satisfies
\begin{equation}
\boxed{u_{tt}=16u_{xx}.}
\end{equation}
\section{Worked example 2: a non-obvious solution written as two traveling pieces}
Consider
\begin{equation}
u(x,t)=xt.
\end{equation}
Since
\begin{equation}
u_{tt}=0
\end{equation}
and
\begin{equation}
u_{xx}=0,
\end{equation}
this function satisfies
\begin{equation}
u_{tt}=c^2u_{xx}
\end{equation}
for any constant $c$.
Can it really be written in right- and left-moving form?
Use the identity
\begin{equation}
(x+ct)^2-(x-ct)^2=4cxt.
\end{equation}
Therefore
\begin{equation}
xt=\frac{(x+ct)^2}{4c}-\frac{(x-ct)^2}{4c}.
\end{equation}
Define
\begin{equation}
G(\eta)=\frac{\eta^2}{4c}
\end{equation}
and
\begin{equation}
F(\xi)=-\frac{\xi^2}{4c}.
\end{equation}
Then
\begin{equation}
\boxed{xt=F(x-ct)+G(x+ct).}
\end{equation}
This example shows that the two-family representation applies to more than localized pulses and sinusoids.
\section{Worked example 3: decompose a standing wave}
Suppose
\begin{equation}
u(x,t)=6.0\,\text{mm}\cos(3x)\cos(12t),
\end{equation}
with $x$ in meters and $t$ in seconds.
Use
\begin{equation}
2\cos\alpha\cos\beta
=\cos(\alpha-\beta)+\cos(\alpha+\beta).
\end{equation}
Then
\begin{align}
u(x,t)
&=3.0\,\text{mm}\cos(3x-12t)\\
&\quad+3.0\,\text{mm}\cos(3x+12t).
\end{align}
Hence the standing wave is the sum of
\begin{equation}
\boxed{u_R=3.0\,\text{mm}\cos(3x-12t)}
\end{equation}
and
\begin{equation}
\boxed{u_L=3.0\,\text{mm}\cos(3x+12t).}
\end{equation}
The common wave speed is
\begin{equation}
c=\frac{\omega}{k}=\frac{12}{3}=\boxed{4.0\,\text{m/s}}.
\end{equation}
\section{Worked example 4: verify a general two-family expression}
Consider
\begin{equation}
u(x,t)=(x-2t)^3+2(x+2t)^2.
\end{equation}
The first term is a function of $x-2t$ only, and the second is a function of $x+2t$ only. Therefore, by the WM15 result, each separately satisfies
\begin{equation}
u_{tt}=4u_{xx}.
\end{equation}
Linearity then implies that the sum also satisfies
\begin{equation}
\boxed{u_{tt}=4u_{xx}.}
\end{equation}
A direct check gives the same result.
For the first term,
\begin{equation}
\frac{\partial^2}{\partial x^2}(x-2t)^3=6(x-2t)
\end{equation}
and
\begin{equation}
\frac{\partial^2}{\partial t^2}(x-2t)^3=24(x-2t).
\end{equation}
For the second term,
\begin{equation}
\frac{\partial^2}{\partial x^2}\left[2(x+2t)^2\right]=4
\end{equation}
and
\begin{equation}
\frac{\partial^2}{\partial t^2}\left[2(x+2t)^2\right]=16.
\end{equation}
Thus
\begin{equation}
u_{tt}=24(x-2t)+16
\end{equation}
while
\begin{equation}
4u_{xx}=4\left[6(x-2t)+4\right]=24(x-2t)+16.
\end{equation}
Therefore the PDE is satisfied everywhere.
\section{Worked example 5: infer speed from the first-order relation}
At one point in a known pure one-way wave, measurements give
\begin{equation}
u_t=-0.24\,\text{m/s}
\end{equation}
and
\begin{equation}
u_x=0.0030.
\end{equation}
Suppose the wave is known to be purely right-moving. Then
\begin{equation}
u_t=-cu_x.
\end{equation}
Hence
\begin{align}
c&=-\frac{u_t}{u_x}\\
&=-\frac{-0.24\,\text{m/s}}{0.0030}\\
&=\boxed{80\,\text{m/s}}.
\end{align}
The sign relation is consistent with rightward propagation because $u_t$ and $u_x$ have opposite signs.
This diagnostic is valid only when the field is known to contain a single traveling family. If both $F$ and $G$ are present simultaneously, the local ratio $-u_t/u_x$ generally does not equal $c$.
\section{Worked example 6: connect the two-family solution to string mechanics}
A stretched string has tension
\begin{equation}
T=180\,\text{N}
\end{equation}
and linear mass density
\begin{equation}
\mu=0.020\,\text{kg/m}.
\end{equation}
The string wave speed is
\begin{align}
c&=\sqrt{\frac{T}{\mu}}\\
&=\sqrt{\frac{180}{0.020}}\\
&=\sqrt{9000}\\
&\approx \boxed{94.9\,\text{m/s}}.
\end{align}
Suppose a sinusoidal component has
\begin{equation}
k=6.0\,\text{rad/m}.
\end{equation}
Then
\begin{equation}
\omega=ck
\end{equation}
gives
\begin{align}
\omega
&=(94.9)(6.0)\\
&\approx \boxed{569\,\text{rad/s}}.
\end{align}
A possible pair of equal-amplitude traveling components is therefore
\begin{equation}
\boxed{u_R=A\cos(6x-569t)}
\end{equation}
and
\begin{equation}
\boxed{u_L=A\cos(6x+569t).}
\end{equation}
Both components satisfy the same mechanically derived string equation
\begin{equation}
u_{tt}=\frac{T}{\mu}u_{xx}.
\end{equation}
\section{Common mistakes}
\begin{itemize}
\item \textbf{Mistake:} thinking $x-ct$ means left-moving because of the minus sign. Holding the argument constant gives $x=ct+\text{constant}$, so the feature moves toward $+x$.
\item \textbf{Mistake:} applying $u_t=-cu_x$ to a field containing both right- and left-moving components. That first-order equation applies to a pure right-moving component.
\item \textbf{Mistake:} treating the factorization of the wave operator as ordinary scalar multiplication without remembering that the factors are differential operators.
\item \textbf{Mistake:} concluding that a solution must look like a pulse or sinusoid. The arbitrary functions $F$ and $G$ can have many sufficiently smooth shapes.
\item \textbf{Mistake:} assuming the two functions $F$ and $G$ are known once the PDE is written. Initial or boundary data are needed to determine them.
\item \textbf{Mistake:} forgetting the smoothness assumptions behind the classical derivative manipulations.
\end{itemize}
\section{What WM16 establishes}
The one-dimensional constant-speed wave equation
\begin{equation}
u_{tt}=c^2u_{xx}
\end{equation}
contains two characteristic propagation families:
\begin{equation}
\boxed{F(x-ct)}
\end{equation}
and
\begin{equation}
\boxed{G(x+ct).}
\end{equation}
For a sufficiently smooth classical solution on a suitable domain,
\begin{equation}
\boxed{u(x,t)=F(x-ct)+G(x+ct).}
\end{equation}
The two components propagate in opposite directions at the same speed magnitude $c$.
The next problem is not to discover the form of the solution, but to determine the two arbitrary functions from physical data. That is the purpose of WM17: initial conditions and the d'Alembert solution.
\section*{References}
\begin{thebibliography}{9}
\bibitem{French1971}
A.~P. French,
\emph{Vibrations and Waves},
M.I.T. Introductory Physics Series,
W. W. Norton \& Company, 1971.
\bibitem{Crawford1968}
Frank S. Crawford, Jr.,
\emph{Waves},
Berkeley Physics Course, Volume 3,
McGraw-Hill, 1968.
\bibitem{OpenStax162}
William Moebs, Samuel J. Ling, and Jeff Sanny,
\emph{University Physics, Volume 1},
OpenStax, 2016,
Section 16.2, ``Mathematics of Waves.''
\bibitem{Feynman47}
Richard P. Feynman, Robert B. Leighton, and Matthew Sands,
\emph{The Feynman Lectures on Physics, Volume I},
Chapter 47, ``Sound. The Wave Equation.''
\bibitem{Feynman48}
Richard P. Feynman, Robert B. Leighton, and Matthew Sands,
\emph{The Feynman Lectures on Physics, Volume I},
Chapter 48, ``Beats.''
\bibitem{MIT803L10}
Massachusetts Institute of Technology,
\emph{8.03SC Physics III: Vibrations and Waves},
Lecture 10, ``Traveling Waves,''
MIT OpenCourseWare.
\end{thebibliography}