Quasilinear Equations and Shocks
In the
method of characteristics,
every family of characteristics we met was well-behaved: the curves were fixed by where they
started, and no two of them ever touched. That good behaviour had a hidden assumption baked into
it — the speed a characteristic travels at never depended on the very quantity
u it was carrying. Break that assumption, and the whole tidy picture can
fall apart in a genuinely new way: fast-moving parts of the solution can catch up to slow-moving
parts ahead of them, their characteristics collide, and a sharp jump — a
shock — erupts out of a perfectly smooth starting condition.
Equations where the speed depends on u itself are called
quasilinear; the famous toy example is the inviscid Burgers' equation
u_t + u\,u_x = 0,
a model of a fluid where each point simply moves at a speed equal to its own height. Following the
same logic as before, along curves with \frac{dx}{dt} = u we still get
\frac{du}{dt} = 0 — u is still constant along
a characteristic. The twist is that each characteristic now carries its own private speed
u(x_0, 0), so the lines are no longer parallel, and nothing stops them
crossing.
The mechanism: a traffic jam in miniature
The cleanest way to feel why this happens is to picture cars on a motorway instead of a fluid.
Traffic-density models are built on exactly this kind of quasilinear PDE, with one simple physical
rule: denser traffic moves slower. A car in an open, sparse stretch of road can
cruise; a car packed into a dense stretch has to crawl.
Now imagine a region of fast, sparse traffic sitting just behind a region of dense, slow traffic.
Each little clump of cars is a "characteristic," coasting along at its own local speed. The fast
clump behind is gaining on the slow clump ahead the whole time — and when it catches up, the smooth
density profile can't describe what happens next with a single value at a single point: cars pile
up, and a sudden, sharp boundary appears between clear road and jam. That boundary is the shock,
appearing out of nowhere even though every car obeyed a perfectly smooth rule the entire time.
Notice what's essential here and what isn't. It doesn't matter that the "stuff" is cars — swap in
molecules of air, and "denser traffic moves slower" becomes "more compressed air pushes back
harder," which is exactly the mechanism behind a shock wave in a gas. The specific substance
changes; the mathematics of characteristics catching each other up does not.
Worked example: pinpointing where a shock is born
Take the decreasing initial profile u_0(x_0) = 2 - 0.5\,x_0 used in the
figure below: taller (faster) values sit to the left, shorter (slower) values sit to the right —
exactly the setup for a pile-up. The characteristic launched from x_0
is the line
x(t) = x_0 + u_0(x_0)\,t = x_0 + (2 - 0.5\,x_0)\,t.
Take two of them, starting at x_0 = a and x_0 = b
with a \ne b, and ask when they meet:
a + (2 - 0.5a)t \;=\; b + (2 - 0.5b)t \;\;\Longrightarrow\;\; (a - b)\big(1 - 0.5t\big) = 0.
Since a \ne b, the only way to satisfy this is
1 - 0.5t = 0, i.e. t = 2 — and that answer
didn't depend on which pair a, b we picked! For this particular
(linear) profile, every characteristic passes through the very same point at
t = 2, namely x = 2t = 4. The whole fan of
starting values collapses into a single instant and a single place: the shock is born at exactly
(x, t) = (4, 2), precisely where the geometry below marks it.
No — that clean, universal t = 2 is special to a perfectly
linear decreasing profile. For a general (curved) decreasing profile
u_0(x_0), different pairs of characteristics cross at different times,
and the shock is born at the very first crossing — the earliest time any two
characteristics collide, not the average or the last. Finding that earliest time in general takes
a bit more calculus than this page needs, but the idea is identical: race the characteristics
forward and watch for the first collision.
The lines that crash together
Step through the picture: feet on the x-axis launch characteristics
whose slopes are set by the initial height there. Because the left points are taller (faster), the
lines lean in and meet — and the first crossing is where the shock is born, matching the algebra
above exactly.
The physical reality
This is not a mathematical curiosity confined to a blackboard. Genuine shocks — sudden jumps that
emerge from smoothly varying starting conditions — show up throughout physics and engineering:
- Sonic booms. A supersonic aircraft compresses the air ahead of it faster than
sound can carry the disturbance away, and the compression waves pile up into the sharp crack you
hear on the ground.
- Traffic jams. Exactly the mechanism above: a smoothly varying stream of cars
develops a sudden density jump — the visible "wall" at the back of a jam — with no crash required
to start it.
- Gas dynamics. Compressor blades, engine cylinders, and re-entering spacecraft
all generate shock waves where a smooth pressure disturbance steepens into a near-discontinuous
front.
- Water waves. A gently sloping ocean swell can steepen as it runs into shallow
water, its front edge outrunning its back edge until the wave breaks — the same
faster-catches-slower story, playing out in water instead of air or asphalt.
In every case, the ingredients are identical: a smooth starting state, a speed that depends on the
local value of the very quantity being transported, and enough time for the fast parts to run down
the slow parts ahead of them.
Once two characteristics of a quasilinear equation cross, the naive rule "follow each curve and read
off u" stops making sense: at the crossing point, it hands you
two different values of u for the same
(x, t). A function can't genuinely take two values at once, so this
multi-valuedness isn't a strange new kind of answer — it's the method telling you that its own
starting assumption, a smooth single-valued solution, has broken down.
The tempting fix is to just pick one of the two values (the bigger one? the average?) and move on.
Don't. The genuinely correct solution beyond the crossing time requires a proper
shock-tracking approach — one that inserts a moving jump and enforces a physical conservation law
across it — which is beyond this introductory page's scope. The lesson to take away is the warning
sign itself: characteristics crossing means "stop, the smooth picture no longer applies here,"
not "pick an answer and carry on."
Yes — startlingly literally. Strip away the units and both are the same quasilinear PDE mechanism:
a quantity (air pressure, car density) moves at a speed that depends on its own value, faster parts
catch up to slower parts, characteristics cross, and a sudden jump appears. The sonic boom you hear
as a wall of sound and the abrupt "wall" at the back of a traffic jam are, mathematically, the
same phenomenon — a shock born from crossing characteristics — playing out in two utterly
different physical media, one in compressible air and one in a stream of cars.
That's the quiet power of studying the equation rather than the substance: once you recognise the
pattern u_t + u\,u_x = 0 (or something shaped like it), you can spot the
same shock-forming story in gas dynamics, traffic engineering, water waves, and beyond — all from one
page of characteristics.