Chapter 10 · Introduction

Chapter 10 — Partial Differential Equations

Equations for functions of several variables: how characteristics carry the solution of a first-order equation, how the sign of one discriminant sorts second-order equations into wave-like, heat-like and Laplace-like, and how separation of variables, d'Alembert's formula and transforms solve the three classical equations.

An ordinary differential equation follows one quantity along one variable — time, usually. Most of nature is spread out in space as well: the temperature along a rod changes both from point to point and from moment to moment, the displacement of a guitar string depends on where you look and when, the concentration of a pollutant in a river is carried downstream as it decays, and the steady temperature in a plate depends on where you are and on nothing else. Equations relating such functions of several variables to their partial derivatives are partial differential equations (PDEs). They are harder than ODEs in one essential way: their solutions are fixed not by a few numbers but by whole functions — initial profiles, boundary values, data along a curve — and which data give exactly one solution depends on the equation.

This chapter follows that thread through the GATE syllabus. For first-order equations the method of characteristics reduces the PDE to ODEs along special curves; the same curves explain which data curves are allowed (transversality) and why a smooth wave can steepen and break. For second-order equations in two variables a single discriminant $\delta=B^2-4AC$ sorts every equation into one of three families — hyperbolic, parabolic, elliptic — with the wave, heat and Laplace equations as models and canonical forms that make the classification concrete. Separation of variables and Fourier series then solve the three model equations on rectangles, discs and intervals. Finally, on the whole line, d'Alembert's formula shows waves travelling at finite speed, the heat kernel shows heat spreading instantly, and the Fourier and Laplace transforms turn PDEs into ODEs or algebra.

First-order PDEs→Characteristics→Lagrange's method→Cauchy problem & transversality→Breaking→$\delta=B^2-4AC$→Canonical forms→Separation of variables→Laplace, heat, wave→d'Alembert & Duhamel→Heat kernel & transforms
🎯 By the end of this chapter you can…
  • classify first-order equations, solve linear ones by characteristics and quasilinear ones by Lagrange's method, and find the integral surface through a given curve;
  • decide with the transversality condition whether a Cauchy problem has a unique solution, and compute the breaking time of Burgers-type equations;
  • classify a second-order equation in each region of the plane, find its characteristics, reduce it to canonical form, and write the general solution of constant-coefficient equations;
  • solve Laplace's equation in a rectangle and a disc, and the heat and wave equations on an interval, by separation of variables and Fourier series, and use the maximum principle, the mean value property and energy arguments;
  • solve the wave equation on the line by d'Alembert's formula, describe domains of dependence and influence, and treat forcing terms by Duhamel's principle;
  • solve the heat equation on the line with the heat kernel, and use Fourier and Laplace transforms for problems on the line and the half-line.
Teaser — One Bump, Three Equations

The same initial bump $u(x,0)=\cos^2\frac{\pi x}{2}$ for $|x|\le1$, $0$ elsewhere (dashed), evolves under three equations on the whole line: transport $u_t+u_x=0$, heat $u_t=u_{xx}$, and the wave equation $u_{tt}=u_{xx}$ released from rest. Press Play and compare. Which equation moves the bump without changing it? Which splits it, and which flattens it? In which one does the bump instantly feel "everywhere"?

t—
peak height—
peak position(s)—
area ∫ u dx—
u at x = 5—

How this is examined

GATE MA asks PDE questions in all three formats, and most of them reward quick, accurate use of a small number of results. MCQs typically ask for the general solution of a first-order linear equation, the value at a point of the solution through given data, the region in which a second-order equation is hyperbolic or elliptic, the characteristics or the canonical form of an equation, or the value of a d'Alembert solution. MSQs test judgement: which data curves give a unique solution of a Cauchy problem, which functions solve a given equation, which statements about harmonic functions (maximum principle, mean value property, Neumann compatibility) or about the heat and wave equations (smoothing, finite speed, conservation of energy) are true. NATs ask for a number: the value of a separated solution of the Laplace, heat or wave equation at a point, a Fourier coefficient, a breaking time, the value of a solution given by the heat kernel or by a transform. The same material appears in IIT JAM and, with more emphasis on proofs, in CSIR NET Part C.

Sections in this chapter

✅ Before you start

You need partial derivatives, the chain rule, Jacobians and the inverse and implicit function theorems from §1.1 — changes of variables are everywhere in this chapter — and the divergence theorem from §2.2 for flux and energy arguments. The ODE toolkit of §6.1 is used constantly: linear first-order equations, the Picard–Lindelöf theorem (characteristics never cross for linear equations because of it), constant-coefficient and Cauchy–Euler equations (the radial part of Laplace's equation in a disc). Separation of variables leads to the Sturm–Liouville problems of §6.3, and the Laplace transform of §6.2 returns in §10.3. Harmonic functions were met in §5.1; uniform convergence and the M-test (§4.2) justify term-by-term differentiation of series, and orthonormal bases and Parseval's identity (§8.3) are the abstract setting of Fourier series. If this is your first visit, the welcome page explains how each section and its 🎯 GATE corner are organised.