advanced34 min studyLesson 415 in Mathematics

Scalar conservation laws and characteristics

Follow state-dependent characteristic speeds until the initial-label map loses invertibility and gradients become singular.

On this page

What you’ll understand

  • Derive the characteristic map for a smooth scalar conservation law.
  • Calculate its Jacobian and the resulting spatial derivative.
  • Distinguish loss of smoothness from unbounded solution amplitude.

In linear transport every part of a profile travels at the same speed, so its shape is simply shifted. A nonlinear scalar conservation law assigns a speed according to the current state. Different parts of a profile can therefore spread apart or crowd together. Characteristics remain a precise solution method while their initial labels map invertibly to position. Their failure identifies a breakdown of the smooth description, not permission to continue a multivalued physical density.

Conservation gives a state-dependent transport law

For a smooth scalar u and a twice continuously differentiable flux f, the equation ut+∂xf(u)=0u_t+\partial_x f(u)=0 becomes ut+f′(u)ux=0u_t+f'(u)u_x=0. Along a curve moving with speed f′(u)f'(u), the total derivative of u vanishes. Its state is therefore the initial value attached to the curve. The characteristic is a straight line for an autonomous flux depending only on u, because that state and hence its speed remain constant along the smooth characteristic.

This derivation uses the classical chain rule. After a discontinuity forms, the conservative divergence form remains meaningful distributionally, while the product of a discontinuous speed and a distributional derivative requires additional interpretation. One should not silently treat those expressions as equivalent at shocks. The straight-line construction describes the pre-shock classical solution and supplies information for building a weak continuation, but it does not by itself select that continuation.

x(t;ξ)=ξ+tf′(u0(ξ)),u(x(t;ξ),t)=u0(ξ)x(t;\xi)=\xi+t f'(u_0(\xi)),\qquad u(x(t;\xi),t)=u_0(\xi)
The initial label ξ identifies a smooth characteristic for a flux depending only on u. The position map must remain invertible to define a single-valued classical solution.↔ Scroll wide equations horizontally.

Invertibility tracks the steepening mechanism

Differentiate the characteristic position with respect to its initial label. The resulting Jacobian is one plus time times the flux curvature times the initial slope. While it is positive and the map is globally invertible, the spatial slope equals the initial slope divided by this Jacobian. For a strictly convex flux, a negative initial slope makes characteristics converge, while a positive slope spreads them. The sign concerns the product of flux curvature and profile slope, not merely whether the state is positive or negative.

For Burgers' flux, the speed is u and the curvature is one. If bounded continuously differentiable initial data have bounded derivative with an attained negative minimum, the first positive Jacobian zero occurs at minus the reciprocal of that minimum. This gives the first characteristic gradient-breaking time in this whole-line construction. The states remain in their original bounded range before breaking because they are constant along characteristics. What blows up is their spatial derivative as neighboring labels compress into the same position.

∂ξx=1+tf′′(u0)u0′,ux=u0′1+tf′′(u0)u0′,t∗=−1min⁡ξu0′(ξ) for Burgers when the minimum is negative\partial_\xi x=1+t f''(u_0)u_0',\qquad u_x=\frac{u_0'}{1+t f''(u_0)u_0'},\qquad t_*=-\frac1{\min_\xi u_0'(\xi)}\ \text{for Burgers when the minimum is negative}
The breaking-time statement uses the stated smooth bounded-data assumptions and an attained negative derivative minimum. It is not a formula for arbitrary nonlinear systems.↔ Scroll wide equations horizontally.

Spreading and compression require different continuations

When neighboring speeds separate, the solution can remain smooth or form a continuous fan from discontinuous increasing data. When they collide, continuing every labeled line produces multiple values at one position. A conserved density cannot take all those values simultaneously. Replacing the overlap by a moving jump restores a single-valued weak description, but its speed must be determined from conservation rather than from an arbitrary average chosen for visual smoothness.

The characteristic picture also identifies a finite local propagation bound when the solution stays in a bounded state interval and the flux derivative is bounded there. This scalar whole-line statement is different from a universal speed independent of data: changing the allowed state range may change the maximal speed. Source terms, explicit spatial flux dependence, boundaries, and coupled systems modify the characteristic equations. Each extension needs its own hypotheses before the simple initial-label formulas can be reused.

f(u)=12u2,f′(u)=u,∣x(t;ξ)−ξ∣≤Mtif ∣u0∣≤Mf(u)=\frac12u^2,\qquad f'(u)=u,\qquad |x(t;\xi)-\xi|\le Mt\quad\text{if }|u_0|\le M
Before characteristic breaking, bounded Burgers states give a data-dependent finite-speed bound. This does not define the post-shock admissible solution by itself.↔ Scroll wide equations horizontally.

Bounded values, unbounded slope

For Burgers' equation take u₀(ξ)=1−2tanhξ. Find the first gradient-breaking time and location, and the slope along the characteristic from ξ=0 at t=0.25.

  1. The initial derivative is −2sech²ξ, whose minimum is −2 at ξ=0. The initial values stay between −1 and three, so amplitude is bounded.

  2. The Jacobian first vanishes at t*=1/2. The characteristic from zero has state u₀(0)=1 and position x=t, so the first breaking point is (x,t)=(0.5,0.5).

  3. At t=0.25 on that characteristic, the slope is −2/(1−2·0.25)=−4. As t approaches one half from below, the denominator tends to zero while the state remains one.

ux(x(t;0),t)=−21−2t,u(x(t;0),t)=1u_x(x(t;0),t)=\frac{-2}{1-2t},\qquad u(x(t;0),t)=1
The derivative formula is valid before the characteristic map breaks at t=1/2.↔ Scroll wide equations horizontally.
The result

The first singularity is a gradient catastrophe at time one half, not an infinite state value.

Explore the model

Interactive lab

Interactive lab · function

A flux derivative is a characteristic speed

Pause and Reset, then set coefficient a=0.5 and position x=1, interpreting the horizontal coordinate as state u in the Burgers flux f(u)=u²/2. Predict the tangent slope. For a fresh comparison, Reset and set a=0.5 and x=−1. Compare the two characteristic speeds.

A flux derivative is a characteristic speedThis lab uses only y = ax². Playback traces the curve within the fixed integration interval from 0 to x; it does not estimate numerical error or change the chosen point. Current readings: y at chosen x: 0.608; Tangent slope · 2ax: 1.35; Signed area · ax³/3: 0.182. All parameters are adjustable with the labeled controls below.−2.4−1.201.22.4−9−4.504.59xy(x, ax²)
y = ax²TangentSigned integral
y at chosen x
0.608
Tangent slope · 2ax
1.35
Signed area · ax³/3
0.182
t = 0 s

This lab uses only y = ax². Playback traces the curve within the fixed integration interval from 0 to x; it does not estimate numerical error or change the chosen point. 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 forThe slopes are one and minus one, so these states travel in opposite directions before interactions. The instrument displays a flux graph and tangent; it does not evolve a spatial profile or form a shock.

Check your understanding

For a smooth Burgers solution, what does the first zero of 1+t u₀′(ξ) signal when u₀′ at that label is negative?

Try it yourself

For Burgers data u₀(ξ)=−0.5tanhξ, determine the first gradient-breaking time and the slope on the zero-label characteristic at t=1.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

The derivative minimum occurs at ξ=0. Use the characteristic Jacobian before it vanishes.

Compare with the solution

The derivative is −0.5sech²ξ with minimum −0.5, so t*=2. At the zero label the state is zero and its characteristic stays at x=0. At t=1 its slope is −0.5/(1−0.5)=−1. The initial amplitude stays within one half and remains bounded before breaking, even as the minimum slope steepens toward negative infinity at time two.

Take it with you

  • A smooth scalar state is constant along its own state-dependent characteristic.
  • The characteristic-map Jacobian determines when slopes steepen and invertibility fails.
  • A post-breaking weak continuation needs conservation and an admissibility principle beyond multivalued characteristics.

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.