% vim: tw=50 % 21/10/2022 10AM \noindent But $\frac{\ddot{T}}{T}$ depends only on $t$, and $\frac{X''}{X}$ depends only on $x$! \\ So both sides must be equal to a constant, say $-\lambda$, so \[ X'' + \lambda X = 0 \tag{3.6} \] \[ \ddot{T} + \lambda c^2 T = 0 \tag{3.7} \] \subsection{Boundary conditions and normal modes} Three possibilities for $\lambda$ (+, 0, -) in spatial ODE (3.6) but restricted by (3.1) \begin{enumerate}[(i)] \item $\lambda < 0$. Take $\chi^2 = -\lambda$ then \[ X(x) = Ae^{\chi x} + B e^{-\chi x} = \tilde{A} \cosh \chi x + \tilde{B} \sinh \chi x \] but boundary conditions imply $X(0) = X(L) = 0 \implies \tilde{A} = \tilde{B} = 0$ (only trivial solution works). \item $\lambda = 0$ then $X(x) = Ax + B$ but then by boundary conditions $A = B = 0$. \item $\lambda > 0$, then $X(x) = A \cos \sqrt{\lambda} x + B\sin \sqrt{\lambda} x$. Here, the boundary conditions (3.1) imply $A = 0$ and $B \sin\sqrt{\lambda} L = 0$, so $\sqrt{\lambda} L = n\pi$. So \[ X_n(x) = B_n \sin \frac{n\pi x}{L}, \quad \lambda_n = \left( \frac{n\pi}{L} \right)^2 \tag{3.8} \] i.e. eigenfunctions and eigenvalues of the system. \end{enumerate} These are \emph{normal modes} because spatial shape in $x$ does not change in time (amplitude may vary). \begin{itemize} \item Fundamental mode ($n = 1$): $\lambda_1 = \frac{\pi^2}{L^2}$. Lowest frequency vibration or first harmonic. \begin{center} \includegraphics[width=0.6\linewidth] {images/a2019e9c512211ed.png} \end{center} \item Second mode ($n = 2$): $\lambda_2 = \frac{4\pi^2}{L^2}$ second harmonic or overtone. \begin{center} \includegraphics[width=0.6\linewidth] {images/a8a9e812512211ed.png} \end{center} \item Third mode $n = 3$ etc \begin{center} \includegraphics[width=0.6\linewidth] {images/af4d8d68512211ed.png} \end{center} \end{itemize} \subsection{Initial conditions and temporal solution} Substitute eigenvalues $\lambda_n = \left( \frac{n\pi}{L} \right)^2$ into time ODE (3.7) \[ \ddot{T} + \frac{n^2 \pi^2 c^2}{L^2} T = 0 \] which has solutions \[ T_n(t) = C_n \cos \frac{n\pi ct}{2} + D_n \sin \frac{n\pi ct}{L} \tag{3.9} \] Thus a specific solution to (3.4) satisfying boundary conditions (3.1) is \begin{align*} y_n(x, t) &= T_n(t) X_n(x) \\ &= \left( C_n \cos \frac{n\pi ct}{L} + D_n \sin \frac{n\pi ct}{L} \right) \sin \frac{n\pi x}{L} \end{align*} (absorbing $B_n$ into $C_n$ and $D_n$). Exercise: verify that this is a solution. \myskip Since the wave equation (3.4) is linear (and boundary conditions (3.1) are homogeneous) we can add the solutions together to find \emph{general string solution} \begin{flashcard}[general-string-solution] \prompt{General string solution?} \[ \boxed{y(x, t) = \cloze{\sum_{n = 1}^\infty \left( C_n \cos \frac{n\pi ct}{L} + D_n \sin \frac{n\pi c t}{L} \right) \sin \frac{n\pi x}{L}} \tag{3.10}} \] \end{flashcard} By construction (3.10) satisfies boundary conditions (3.1), so now impose initial conditions (3.2): \\ For $t = 0$ we have \[ y(x, 0) = p(x) = \sum_{n = 1}^\infty C_n \sin \frac{n\pi x}{L} \] by (3.10) and also by (3.10): \[ \pfrac{y}{t} (x, 0) = q(x) = \sum_{n = 1}^\infty \frac{n\pi c}{L} D_n \sin \frac{n\pi x}{L} \] So the coefficients are those for Fourier sine series given by (1.12): \[ C_n = \frac{2}{L} \int_0^L p(x) \sin \frac{n\pi x}{L} \dd x \] \[ D_n = \frac{2}{2\pi c} \int_0^L q(x) \sin \frac{n\pi x}{L} \dd x \tag{3.11}\] Hence (3.10-11) is the solution to (3.4) satisfying (3.1-2). \begin{example*} Pluck string at $x = 3$, drawing it back as \[ y(x, 0) = p(x) = \begin{cases} x(1 - \xi) & 0 \le x \le \xi \\ \xi (1 - x) & \xi < x \le 1 \end{cases} \] \[ \pfrac{y}{t} (x, 0) = q(x) = 0 \] Then with Fourier series (1.8) \[ C_n = \frac{2\sin n\pi \xi}{(n\pi)^2}, \quad D_n = 0 \] so we have solution \[ y(x, t) = \sum_{n = 1}^\infty \frac{2}{(n\pi)^2} \sin n\pi\xi \sin n\pi x \cos n\pi ct \] Take $\xi = \half$ then $C_{2m} = 0$, $C_{2m - 1} = \frac{2(-1)^{m + 1}}{((2m - 1)\pi)^2}$. For a guitar, $\frac{1}{4} \le \xi \le \frac{1}{3}$, for a violin, $\xi \approx \frac{1}{7}$. \end{example*} \subsubsection*{Separation of Variables Methodology} \begin{enumerate}[(1)] \item Obtain linear PDE for system (with boundary conditions and initial conditions) \item Separate variables to tield decoupled ODEs \item Impose homogeneous boundary conditions to find eigenvalues and eigenfunctions \item Use these eigenvalues (constants of separation) to find eigenfunctions in the other variables. \end{enumerate} \subsubsection*{Aside: Solution in characteristic coordinates} Recall sine / cosine summation identities which means our general solution (3.10) becomes \begin{align*} y(x, t) &= \half \sum_{n = 1}^\infty \left[ C_n \sin \frac{n\pi}{L} (x - ct) + D_n \cos \frac{n\pi}{L} (x - ct) \right. \\ &~~~~~~~~~~~+ \left. C_n \sin \frac{n\pi}{L} (x + ct) + D_n \cos \frac{n\pi}{L} (x + ct) \right] \\ &\equiv f(x - ct) + g(x + ct) \tag{3.12} \end{align*} The standing wave solution (3.10) is made up of a right-moving wave (along the characteristic $x - ct = \eta$, constant) and a left-moving wave ($x + ct = \xi$, constant) i.e. a general solution with arbitrary $f$, $g$ (see later). \myskip Special case: $g(x) = 0$ in (3.1), then $f = g = \half p$ at $t = 0$.