Wave PDEs Boundary Conditions (Solutions Guide)
To solve a one-dimensional wave equation, state the string or medium’s domain, wave speed, initial displacement and velocity, and boundary conditions at both ends. Choose a method that fits those conditions, such as a Fourier series for homogeneous endpoints. Then check compatibility and verify that the solution satisfies the equation and every item of given data.
Wave problems can feel difficult because a familiar formula may only work for one set of endpoint rules. The key is to identify those rules before choosing a method. I emphasize this order when working through examples: a correct calculation for fixed ends can still be wrong for a free end.
Problem Setup and Boundary-Condition Types
A boundary condition tells you how the solution behaves at the edge of its spatial domain. For a string on (0<x
Write the full problem before solving:
[
u_{tt}=c^2u_{xx},\qquad 0<x
[ u(x,0)=f(x),\qquad u_t(x,0)=g(x). ]
Here, (u(x,t)) is displacement, (f(x)) is initial displacement, and (g(x)) is initial velocity. The domain length is (L); the time interval may be all (t>0), or a stated finite interval. The parameter (c) sets how fast waves travel.
Boundary conditions specify behavior at (x=0) and (x=L). The main types are Dirichlet, Neumann, and Robin. They are not interchangeable: each changes the allowed vibration modes and therefore the form of the solution.
| Type | Typical condition | Meaning in a string model |
|---|---|---|
| Dirichlet | (u(0,t)=0) | Endpoint displacement is fixed |
| Neumann | (u_x(0,t)=0) | Endpoint slope is zero; this models an ideal free end in the string model |
| Robin | (\alpha u+\beta u_x=0) | Endpoint displacement and slope are linked |
The condition must be given at each end. For example, fixed-fixed means (u(0,t)=u(L,t)=0). A mixed problem could have (u(0,t)=0) and (u_x(L,t)=0). Do not assume both ends have the same type.
Compatibility means the initial data agree with the boundary rules where (t=0) meets an endpoint. With fixed ends, require (f(0)=f(L)=0). For a sufficiently smooth solution, the endpoint velocity must also agree: (g(0)=g(L)=0). Higher smoothness can require further conditions. For example, differentiating a fixed-end condition twice in time and using the PDE gives (f”(0)=f”(L)=0) when those derivatives are required to be continuous.
A mismatch does not always prevent a weaker solution, but it can prevent the smooth solution a problem asks for. Check the stated regularity and the corner conditions before using a series.
Solution Methods, Eigenvalues, and Requirements
Separation of variables breaks a solution into a spatial shape and a time pattern. The endpoint rules determine the spatial eigenfunctions and eigenvalues. For homogeneous boundary conditions, this method leads to Fourier series; for nonzero boundary data, first remove the boundary values or use a suitable eigenfunction expansion.
For fixed ends, suppose (u(0,t)=u(L,t)=0). Seeking (u=X(x)T(t)) gives spatial modes (\sin(n\pi x/L)), for positive integers (n). Each mode oscillates at angular frequency (\omega_n=cn\pi/L). The resulting solution is
[ u(x,t)=\sum_{n=1}^{\infty} \left[A_n\cos(\omega_nt)+B_n\sin(\omega_nt)\right] \sin\left(\frac{n\pi x}{L}\right). ]
The initial displacement sets (A_n), and the initial velocity sets (B_n):
[ A_n=\frac{2}{L}\int_0^L f(x)\sin\left(\frac{n\pi x}{L}\right)\,dx, \qquad B_n=\frac{2}{L\omega_n}\int_0^L g(x)\sin\left(\frac{n\pi x}{L}\right)\,dx. ]
These formulas assume the fixed-end conditions and suitable regularity for the series to represent the data. They are not a general formula for every boundary-value problem.
Worked setup: Let (L=1), (c=2), fixed ends, (f(x)=\sin(\pi x)), and (g(x)=0). Only the first sine mode is present, so
[ u(x,t)=\cos(2\pi t)\sin(\pi x). ]
Indeed, (u_{tt}=-4\pi^2u), while (c^2u_{xx}=4(-\pi^2u)=-4\pi^2u). It has zero displacement at both endpoints, initial displacement (\sin(\pi x)), and zero initial velocity.
For a mixed condition (u(0,t)=0,\ u_x(L,t)=0), the spatial modes are
[ \sin\left(\frac{(n+\tfrac12)\pi x}{L}\right),\qquad n=0,1,2,\ldots, ]
with frequencies (c(n+\tfrac12)\pi/L). This half-step in the frequencies comes directly from imposing both endpoint rules.
Robin conditions connect displacement and slope, often written (\alpha u+\beta u_x=0) at an endpoint. Their eigenvalues depend on the constants and on which condition applies at each end. Solve the corresponding spatial eigenvalue problem; do not insert the fixed-end or free-end frequencies without checking.
| Boundary setup | Spatial modes | Frequency pattern |
|---|---|---|
| Fixed at both ends | (\sin(n\pi x/L)) | (cn\pi/L,\ n\ge1) |
| Zero slope at both ends | (\cos(n\pi x/L)), plus constant | (cn\pi/L,\ n\ge1), plus zero mode |
| Fixed at (0), zero slope at (L) | (\sin((n+\tfrac12)\pi x/L)) | (c(n+\tfrac12)\pi/L) |
| Robin | Determined by endpoint eigenvalue equations | Depends on (\alpha,\beta) and endpoint setup |
Nonzero boundary values
A nonhomogeneous boundary condition has a prescribed value that is not zero, such as (u(0,t)=a). A lifting function is a function chosen to match those boundary values. Subtract it from (u) to form a new unknown with homogeneous endpoint conditions.
For time-independent Dirichlet values (u(0,t)=a) and (u(L,t)=b), one simple choice is
[ \phi(x)=a+\frac{b-a}{L}x,\qquad v=u-\phi. ]
Then (v) is zero at both ends. Since (\phi_{xx}=0) and (\phi) does not depend on time, (v) satisfies the same wave equation. Its initial data become (f-\phi) and (g). If the boundary values depend on time, the transformed equation generally gains a forcing term:
[ v_{tt}-c^2v_{xx}=-(\phi_{tt}-c^2\phi_{xx}). ]
Choose (\phi) to match both endpoints, then solve the forced problem using an eigenfunction method or another method suited to the data.
Verification, Common Errors, and Effective Study Practice
Verification means checking the proposed solution against the PDE, initial data, and both endpoint conditions. It is a required part of solving the problem, not just a final calculation. I find that checking these items in order catches the common error of using a valid series for the wrong boundary conditions.
Use this sequence:
- Record the data: state (L), (c), the time interval, (f), (g), and the condition at each endpoint.
- Check compatibility: evaluate the initial data at the endpoints and compare them with the boundary rules.
- Choose the spatial eigenfunctions: derive them from the endpoint conditions, rather than selecting a familiar series by habit.
- Find the coefficients: use the matching Fourier or eigenfunction expansion and the initial velocity to set the time-dependent coefficients.
- Verify the result: substitute into the PDE, evaluate at (t=0), and check each boundary condition.
For a series solution, checking each term can establish that the PDE and homogeneous boundary conditions hold term by term, when differentiation and summation are justified. To verify the initial data, use the Fourier expansion and its convergence properties. For smooth classical solutions, ask whether the given data are smooth enough for the required derivatives and whether the necessary endpoint compatibility conditions hold.
Common mistakes include using sine modes for Neumann conditions, forgetting the zero mode in a Neumann-Neumann problem, and using the fixed-fixed frequencies for mixed endpoints. Another error is to check only the initial displacement and overlook the initial velocity. If the boundary data are nonzero, a homogeneous-boundary series applied directly to (u) may fail at the endpoints.
A practical study habit is to write the endpoint conditions beside the eigenvalue problem before calculating coefficients. This makes the assumptions visible and helps you spot a mismatch early. If a result is complicated, test one mode or use a numerical plot as a check, but do not treat a plot as proof.
Physical Interpretation and Applications
The boundary conditions describe how the modeled system interacts with its ends. Fixed and free idealizations lead to different mode shapes, while Robin conditions represent a linked response at the boundary. The mathematical solution predicts motion only within the assumptions of the model, including its wave speed and chosen endpoint rules.
For an ideal vibrating string, a fixed endpoint has zero displacement; a free endpoint is often modeled by zero slope. These are simplified idealizations. A real support may move or exert forces in a more complex way, so the correct mathematical condition depends on the physical model.
The modes describe patterns the system can sustain. For fixed ends, the first mode has one half-wave across the length, and higher modes fit more half-waves into the same interval. The frequency grows with (c) and with the mode number (n). Thus, changing the endpoint type can change the frequencies even when the length and wave speed remain the same.
A useful modeling comparison is:
- String held at both ends: fixed-fixed conditions; use sine modes.
- Idealized free end: zero spatial derivative at that end; use a Neumann condition there.
- End attached to a compliant support: a Robin condition may be appropriate if the model relates endpoint displacement and slope.
- Moving or driven endpoint: prescribe its time-dependent value, then use a lifting function or a forced formulation.
These are modeling choices, not universal descriptions of every physical support. State the assumptions, solve the matching mathematical problem, and interpret the result within those limits.
Frequently Asked Questions
These answers summarize the main decisions in a one-dimensional wave problem: what data to state, how endpoint rules affect the method, and what to check before accepting a solution. The concise rules below apply to the standard model (u_{tt}=c^2u_{xx}); more complex systems may require different conditions or methods.
What information is needed to solve a one-dimensional wave equation?
State the spatial domain, wave speed, time interval, initial displacement, initial velocity, and boundary condition at each endpoint.
What is the difference between Dirichlet and Neumann conditions?
A Dirichlet condition specifies the value of (u). A Neumann condition specifies its spatial derivative, (u_x).
When can I use the sine-series solution?
Use the standard sine-series solution for homogeneous fixed-end conditions, (u(0,t)=u(L,t)=0), with suitable initial data and regularity.
What if the boundary values are not zero?
Choose a lifting function that matches them, subtract it from the unknown, then solve the resulting homogeneous-boundary problem, including any forcing term created by the transformation.
Why does a Neumann problem include a zero mode?
A constant spatial function has zero derivative at both ends, so it satisfies homogeneous Neumann conditions. Its time factor is (A_0+B_0t).
What does compatibility mean at (t=0)?
It means the initial displacement and velocity agree with the boundary conditions where the initial line meets each endpoint. Higher smoothness may require additional corner conditions.
How do I check a proposed solution?
Substitute it into the PDE, evaluate its displacement and velocity at (t=0), and check both endpoint conditions for all times.
Are Robin and Neumann conditions the same?
No. Neumann specifies (u_x); Robin links (u) and (u_x). Each produces its own spatial eigenvalue problem and modes.
What is the most common solution error?
Using eigenfunctions that do not satisfy the stated endpoint conditions, such as applying fixed-end sine modes to a Neumann problem.
(This article was written by one of our staff writers, Alan Westbrook. Visit our Meet the Team page.)
