Chapter 3 · Introduction

Chapter 3 — First-Order Partial Differential Equations

Equations for an unknown surface $z(x,y)$ that involve only its first partial derivatives $p=z_x$, $q=z_y$ — solved by turning them into ordinary differential equations along characteristic curves.

In the 1770s Lagrange noticed something remarkable about an equation of the form $P\,z_x+Q\,z_y=R$. It looks like a partial differential equation, but it says only that the vector $(P,Q,R)$ is tangent to the surface $z=z(x,y)$ — so the surface must be made of the curves that follow this vector field, and those curves are found by solving ordinary differential equations. A few years later Monge gave the geometric picture for arbitrary first-order equations: at each point the equation allows a whole cone of tangent planes (the Monge cone), and a solution surface threads its way through these cones. In 1784 the young Paul Charpit, who died that same year, showed how to reduce any first-order equation in two variables to ordinary differential equations by carrying the slopes $p$ and $q$ along as unknowns. Half a century later Hamilton and Jacobi found that the same theory governs light rays and the motion of planets: the Hamilton–Jacobi equation of mechanics (Chapter 8) and the eikonal equation of optics are first-order equations, and their characteristics are the rays and trajectories.

First-order equations also describe things that move and pile up. Riemann (1860) saw that sound waves of large amplitude steepen into shocks; Lighthill, Whitham and Richards (1955–56) modelled traffic on a motorway by a conservation law whose characteristics carry car density, and explained why jams appear out of smooth traffic. In each case the method is the one of this chapter: follow the characteristics from the data, and the solution is whatever the characteristics carry — until two of them cross. Section 3.1 develops Lagrange's method for quasi-linear equations, with the forming of equations by eliminating constants and functions; Section 3.2 treats nonlinear equations (complete and singular integrals, Charpit's method, the standard forms) and the Cauchy problem, which decides when a curve lies on exactly one solution surface. Chapter 4 then moves to second-order equations, where characteristics reappear in the classification into hyperbolic, parabolic and elliptic types.

Eliminate constants / functions→Pp + Qq = R→dx/P = dy/Q = dz/R→φ(u, v) = 0→Surface through a curve→Complete & singular integrals→Charpit's equations→Standard forms→Cauchy problem & transversality
🎯 By the end of this chapter you can…
  • classify a first-order PDE as linear, semilinear, quasi-linear or nonlinear, and form the PDE of a family of surfaces by eliminating arbitrary constants or an arbitrary function;
  • explain why integral surfaces of $Pp+Qq=R$ are unions of characteristic curves, and find the general integral $\varphi(u,v)=0$ by Lagrange's method, using multipliers where needed;
  • find the integral surface through a given curve and the surfaces orthogonal to a given family, and recognise when such a surface fails to exist or is not unique;
  • find complete, general and singular integrals, test two equations for compatibility, and apply Charpit's method, the four standard forms (including Clairaut's form) and, briefly, Jacobi's method;
  • solve the Cauchy problem for a quasi-linear equation by characteristics, check the transversality condition, and compute when the characteristics of an equation like $z_y+zz_x=0$ first cross.
Teaser — Characteristics Carry the Data, Until They Cross

Think of $y$ as time. An initial profile $z(x,0)=g(x)$ is carried along straight characteristic lines in the $(x,y)$-plane (left): for the transport equation $z_y+0.8\,z_x=0$ all lines are parallel, so the hump just slides. For Burgers' equation $z_y+z\,z_x=0$ each point moves with speed equal to its own height, so the top of the hump overtakes the front: the lines converge, and at the breaking time $y^*=-1/\min g'$ they cross and the profile (right) overturns. With increasing data the lines fan out instead. Press Play or drag the time slider.

0.60
time y—
breaking time y*—
profile—

How this is examined

First-order PDEs supply a steady stream of short, sharp CSIR NET questions. In Part B you may be asked to form the equation satisfied by a family such as $z=f(x^2+y^2)$ or $z=ax+by+ab$; to find the general solution of a Lagrange equation, often one that needs the multipliers $(1,1,1)$ or $(x,y,z)$; to evaluate at a given point the integral surface passing through a given curve; to write down a complete integral of an equation of standard form; to find the singular integral of a Clairaut-type equation; or to compute the first time at which the characteristics of a Burgers-type equation intersect. Part C, where several options may be correct, tests understanding: whether a given function is a first integral, whether a Cauchy problem has no solution, exactly one or infinitely many (the transversality condition and its failure when the data curve runs along a characteristic), whether a solution exists for all time, whether a complete integral is unique, whether a proposed surface really is a singular integral, and which of several equations are quasi-linear. Most wrong options are instances of the counterexamples collected in the ⚠️ boxes of the two sections. GATE MA asks similar questions, sometimes with a numerical answer.

Sections in this chapter

✅ Before you start

The whole chapter reduces partial differential equations to ordinary ones, so you need Chapter 1: the existence and uniqueness theorem with its Lipschitz condition (§1.1), envelopes, singular solutions and Clairaut's equation for ODEs (§1.2 — complete and singular integrals of PDEs are their direct analogues), and systems of first-order equations (§1.3), since the characteristic equations form such a system. From the CSIR NET Unit 1 course you need partial derivatives and the chain rule in several variables, Jacobians, the implicit and inverse function theorems, and exact differentials. A clear picture of a surface $z=z(x,y)$, its tangent plane and its normal vector $(z_x,z_y,-1)$ will make every proof in this chapter transparent. If this is your first visit, the welcome page explains how each section is organised.