Characteristics and transport
Follow information along curves and locate the point where a smooth transport description can fail.
On this page
The question for today
What you’ll understand
- Derive characteristic equations using the chain rule.
- Solve constant-speed transport with initial and inflow data.
- Identify characteristic crossing as a limit of classical nonlinear solutions.
Transport equations describe quantities carried by a prescribed or solution-dependent motion. Their key simplification is geometric: along the correct curve, a PDE becomes an ordinary differential equation. The hard part is not memorizing a translated profile; it is choosing those curves, mapping them back to the observation point, and checking that this map remains invertible. The same viewpoint explains why a boundary needs data only where information enters and why nonlinear waves can develop a shock.
Choose a path that matches the derivative
Consider u_t+a(x,t)u_x=b(x,t,u). Along a path X(t), the chain rule gives d u(X(t),t)/dt=u_t+X'(t)u_x. Choosing X'=a converts the PDE into the scalar ODE U'=b(X,t,U). Initial values for X and U come from a curve carrying the prescribed data. If the coefficient a is smooth enough to give a unique flow and the source is locally Lipschitz in U, this characteristic system supplies a local construction. Its existence interval and regularity still require checking.
For constant a=c and zero source, X(t)=ξ+ct and U(t)=f(ξ). Solving ξ=x−ct gives u(x,t)=f(x−ct). The minus sign describes a profile moving right when c is positive: the peak initially at ξ arrives at ξ+ct. No stretching or smoothing occurs because every point moves at the same speed. On the whole line, translation preserves every Lp norm for which the initial profile belongs to Lp, and transports nonsmooth profiles as well when the equation is interpreted distributionally.
Initial curves and inflow boundaries carry information
On a finite interval with positive speed, characteristics enter through the left endpoint and leave through the right. Prescribing the left boundary history is therefore natural; prescribing an independent right boundary history can overdetermine the problem. At a given interior point, trace the characteristic backward. It either reaches the initial line first or reaches the inflow boundary first. That encounter determines which datum to use. At the corner, compatible values are needed if a continuous classical solution is required.
A general data curve must also be noncharacteristic: its tangent cannot coincide with the direction along which the PDE already specifies change. Giving values along a characteristic may fail to determine neighboring characteristics, or may contradict the ODE along that same curve. Geometrically, solving the problem uses the initial-curve parameter and elapsed characteristic time as local coordinates. A nonzero Jacobian allows inversion. Checking this condition turns a tempting parametric formula into a legitimate local solution rather than a collection of unrelated trajectories.
Solution-dependent speed can destroy invertibility
For inviscid Burgers transport u_t+u u_x=0, the value remains constant along a characteristic but now sets its speed. Starting from ξ, the path is X=ξ+t f(ξ). Its derivative with respect to ξ is 1+t f'(ξ). If an initial slope is negative, faster material on the left catches slower material on the right, and this derivative can vanish. Before that time, the inverse map gives a classical solution; at crossing, assigning both carried values to one point is impossible.
A shock is not obtained by continuing the multivalued classical formula as though it were a function. One instead returns to the conservation law and an admissibility condition, studied later in this module. In the waves lab, set amplitude ratio to zero and play the remaining sin(2x−t) component. Its phase travels at speed one half, an exact constant-speed transport example. Bringing back the second component changes interference, not speed through the local value. The lab does not implement Burgers characteristics, boundary inflow, or shock formation.
Work it through
Follow a decaying transported signal
Solve on the line with , and evaluate u(3,1).
The characteristic through initial point ξ is X(t)=ξ+2t. To reach x=3 at t=1, its starting point must be ξ=3−2=1.
Along the path, U'=−U and U(0)=exp(−ξ²). Thus and .
At (3,1), u=e^(−1)e^(−1)=e^(−2)≈0.135335. Direct differentiation gives u_t+2u_x=−u, and setting t=0 recovers the prescribed Gaussian.
The signal translates at speed two while its carried amplitude decays with time constant one.
The living notebook
Explore the model
Interactive lab · waves
Read a traveling phase
Set amplitude ratio a to zero and play sin(2x−t): a crest moves right at speed 1/2. Restore the second component and change phase difference to see interference between equal-speed waves. This is linear transport of sinusoidal profiles, not nonlinear characteristic crossing or a shock solver.
- Sum at x = 0
- 0.866
- Resultant amplitude
- 1.732
- Phase difference
- 60°
Two waves share the same frequency and wave number. Playback advances t in sin(2x − t) + a sin(2x − t + phase). Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forFollowing constant phase connects a moving picture to its characteristic speed.
Pause & reason
Check your understanding
Your turn
Try it yourself
For Burgers transport on the line, take . Find the first possible breakdown time of the classical characteristic map.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Differentiate X(ξ,t)=ξ+t f(ξ) and use the minimum value of f′.
Compare with the solution
Since , the minimum slope is −1 at ξ=0. Thus stays positive for 0≤t<1 and first vanishes at t=1. The classical slope formula u_x=f′/(1+t f′) becomes unbounded there. A continuation after this time needs a weak solution with entropy admissibility, not a multivalued graph.
Keep the idea
Take it with you
- Characteristics turn directional change into an ODE.
- Trace a characteristic backward to find the relevant initial or inflow datum.
- A nonlinear characteristic formula is valid only while its coordinate map remains invertible.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.