Home / Learn / Exam prep / GATE Physics / Lesson 3.1
All lessons

Module 3: Electromagnetic theory · Lesson 3.1 · about 60 min

Boundary value problems and Laplace's equation

Uniqueness with Dirichlet, Neumann and mixed data, separation of variables in Cartesian, spherical and cylindrical coordinates, Legendre expansions on spheres and shells, spheres and cylinders in uniform fields, fields at corners and edges, and the symmetry tricks that solve a problem in one line.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 3 · about 60 minutes. This lesson turns Laplace’s equation from a formula into a working tool. You will learn exactly which boundary data fix a solution, how to build solutions by separating variables in the three coordinate systems GATE uses, how to read off charges, dipole moments and fields from the coefficients, and the handful of symmetry tricks that save whole pages of algebra. Conductors, the basic uniqueness theorem and the simple slot problem are in the IIT JAM lesson on conductors, Laplace’s equation and images; look there first if they are new.

In a charge-free region the potential is fixed completely by what happens on the boundary. So any function that obeys Laplace's equation and fits the boundary is the answer, and separation of variables is a systematic way of building that function out of simple pieces.

1. What data fix the answer

Poisson’s equation ∇2V=−ρ/ϵ0 holds inside a region; on its closed boundary S we may give:

  • Dirichlet data: V itself on S (a conductor held at a known potential);
  • Neumann data: the normal derivative ∂V/∂n=−En on S (a known surface charge just inside a conductor, or a known flux);
  • mixed data: V on part of S and ∂V/∂n on the rest.

Why any of these is enough. Let V1 and V2 both solve the problem and put U=V1−V2. Then ∇2U=0 inside, and Green’s first identity with ∇⋅(U∇U)=|∇U|2+U∇2U gives

∫region|∇U|2dτ=∮SU∂U∂nda

On each patch of S either U=0 (Dirichlet) or ∂U/∂n=0 (Neumann), so the right side vanishes. A sum of squares that integrates to zero is zero everywhere: ∇U=0, so the two fields are identical. With pure Neumann data U can still be a constant; with any Dirichlet patch it is zero.

Two consequences matter in exams:

  1. Too much data is fatal. Giving both V and ∂V/∂n on all of S (Cauchy data) over-determines the problem: in general no solution exists.
  2. The second uniqueness theorem. For a set of conductors, giving the total charge on each one (plus ρ between them) also fixes the field. The same identity, applied with each conductor an equipotential carrying zero net difference charge, does the job. How the charge spreads over each surface comes out of the solution.

Neumann data must also be consistent with Gauss’s law: ∮∂V/∂nda=−Qenc/ϵ0.

Harmonic function
a solution of ∇²V = 0; it has no local maxima or minima, and its value at any point is the average over any sphere around it
Dirichlet / Neumann
the boundary gives the potential itself / the normal field
Uniqueness
one solution only, so a correct guess is the answer, however it was found
Separation of variables
try V as a product of functions of one coordinate each; Laplace's equation then splits into ordinary equations linked by separation constants
Physical meaning
The boundary is a set of instructions; the potential inside is the smoothest function that follows them. There is never a choice left over.

2. Separation in Cartesian coordinates

Try V=X(x)Y(y) in two dimensions. Dividing ∇2V=0 by XY:

X′′X+Y′′Y=0⇒X′′X=k2,Y′′Y=−k2

Each ratio depends on a different variable, so each is a constant, and the two constants are equal and opposite. One direction oscillates, the other grows or decays. The boundary decides which: the direction with two grounded walls gets the sines, because only sin⁡(nπy/a) vanishes at both y=0 and y=a.

Worked example: one Fourier mode. A rectangle 0≤x≤a, 0≤y≤b is grounded on three sides and held at V0sin⁡(πx/a) on the side y=b. The sines must go along x; along y we need a function that vanishes at y=0, so sinh⁡:

V(x,y)=V0sin⁡πxasinh⁡(πy/a)sinh⁡(πb/a)

It is harmonic, it fits all four sides, so by uniqueness it is the answer. For a square, at the centre: V0sinh⁡(π/2)/sinh⁡π=0.199V0. The potential sags far below the “straight line” guess of 0.5V0: harmonic functions are dominated by their largest boundary values only nearby.

Any boundary function. If the edge is held at a general f(x), expand it in a Fourier sine series, f(x)=∑cnsin⁡(nπx/a) with cn=2a∫0af(x)sin⁡nπxadx, and give each term its own sinh⁡. For a constant V0, cn=4V0/nπ for odd n and zero for even n. Higher modes die faster, as e−nπx/a, so a few terms usually suffice away from the edge.

Two equal walls. If the walls x=±b are both at V0 (and y=0,a grounded), the solution must be even in x: use cosh⁡(nπx/a)/cosh⁡(nπb/a). Using sinh⁡ there is the commonest slip.

Three dimensions. In a box, V=X(x)Y(y)Z(z) with X′′/X=−α2, Y′′/Y=−β2 forces Z′′/Z=α2+β2. Two directions oscillate and the third decays faster than in 2D: for the face mode sin⁡(πx/a)sin⁡(πy/a) the decay constant is π2/a, and in a cube the centre value is only 0.107V0.

3. Spherical coordinates: Legendre polynomials

With azimuthal symmetry, try V=R(r)Θ(θ). The angular equation is Legendre’s equation, whose solutions finite at both poles are the Legendre polynomials Pl(cos⁡θ): P0=1, P1=cos⁡θ, P2=12(3cos2⁡θ−1), P3=12(5cos3⁡θ−3cos⁡θ). The radial equation ddr(r2dRdr)=l(l+1)R is solved by rl and r−(l+1). So

V(r,θ)=∑l=0∞(Alrl+Blrl+1)Pl(cos⁡θ)

Inside a sphere (origin included) only rl survives; outside (potential going to zero) only r−(l+1). The coefficients come from orthogonality, ∫−11PlPl′dx=22l+1δll′.

Rule of thumb: each Legendre mode on the surface travels on its own. If the surface holds V0Pl, the inside is V0(r/R)lPl and the outside V0(R/r)l+1Pl. The jump in Er at the surface gives the charge on it:

σ=ϵ0∑l(2l+1)AlRPl(cos⁡θ)⟺Al=σlR(2l+1)ϵ0

where Al is the surface value of mode l. Read it either way: given the potential, find the charge; given a shell’s charge σlPl, find its potential. For σ=σ0cos⁡θ this gives the famous uniform field −σ0/3ϵ0 inside.

Worked example. A surface is held at V0cos2⁡θ. First rewrite: cos2⁡θ=13+23P2. Then

Vin=V03+2V03r2R2P2(cos⁡θ),Vout=V03Rr+2V03R3r3P2(cos⁡θ)

The centre is at V0/3, the surface carries total charge 4πϵ0RV0/3 (from the 1/r term), and the far field is that of this charge plus a quadrupole. Forgetting that cos2⁡θ contains an l=0 piece is the classic error.

Worked example: two hemispheres. Northern hemisphere at +V0, southern at −V0. The boundary function is odd, so only odd l appear: cl=(2l+1)V0∫01Pldx, giving c1=32V0, c3=−78V0. Far away, the dipole term 32V0(R/r)2cos⁡θ means p=6πϵ0R2V0.

A sphere in a uniform field

The most-asked separation problem: a sphere (conductor or linear dielectric) in a uniform field E0zˆ. Far away V→−E0rcos⁡θ=−E0rP1, so only l=1 can appear anywhere.

Guess the form. Outside: V = −E₀ r cos θ + B cos θ / r² (the uniform field plus an induced dipole). Inside: V = −A r cos θ (a uniform field, finite at the centre).

Match V at r = R. −E₀R + B/R² = −AR.

Match the normal D. For a dielectric, εr∂Vin/∂r = ∂Vout/∂r at r = R: −εrA = −E₀ − 2B/R³.

Solve. A = 3E₀/(εr + 2) and B = E₀R³(εr − 1)/(εr + 2). For a conductor (εr → ∞, or simply V = 0 on the sphere) A = 0 and B = E₀R³: a dipole p = 4πε₀R³E₀ and σ = 3ε₀E₀ cos θ.

Try the switch in the figure: as ϵr rises the field inside weakens and the lines bend towards the sphere, until the metal swallows them completely, meeting its surface at right angles. The dipole part alone shows the r−2 term that decorates the uniform field. The polarisation side of the same problem, with bound charge, is in fields in matter.

4. Cylindrical problems in two dimensions

For long objects with no z dependence, separation in polar coordinates (s,ϕ) gives

V(s,ϕ)=a0+b0ln⁡s+∑n≥1(ansn+bnsn)(cncos⁡nϕ+dnsin⁡nϕ)

The n=0 pair is a constant and the line-charge potential ln⁡s; requiring single-valuedness makes n an integer.

Worked example: a metal cylinder across a field. Far away V→−E0scos⁡ϕ; on the cylinder V=0. Only n=1 with cos⁡ϕ is needed:

V=−E0(s−R2s)cos⁡ϕ,σ=−ϵ0∂V∂s|R=2ϵ0E0cos⁡ϕ

The induced charge integrates to zero, the outside picture is a 2D line dipole of strength 2πϵ0R2E0 per length, and the surface field peaks at 2E0 (against 3E0 for a sphere). A surface held at V0cos⁡2ϕ gives V0(s/R)2cos⁡2ϕ=V0(x2−y2)/R2 inside.

Corners and edges

Two conducting planes at the same potential meet along an edge, enclosing a charge-free wedge of angle β. With V=0 on ϕ=0 and ϕ=β, the separated solutions are snπ/βsin⁡(nπϕ/β). Near the edge the n=1 term dominates:

V∝sπ/β,E∝sπ/β−1

Inside a right-angled corner (β=π/2) the field vanishes linearly; outside a right-angled metal corner (β=3π/2) it diverges as s−1/3; at the edge of a thin sheet (β=2π) as s−1/2. This is the quantitative reason charge piles up on sharp edges and lightning rods work, and why the 2D electrostatics solver needs a fine grid near corners.

“A sphere held at V0cos2⁡θ has V0/2 at its centre, because the average of cos2⁡ is one half.” One half is the average over the angle θ. The centre value is the average over solid angle, and the sin⁡θ in dΩ=sin⁡θdθdϕ gives the equator, where cos2⁡θ is small, more weight. The right answer is V0/3. In Legendre language, only the l=0 coefficient reaches the centre.

5. One-line tricks

Separation is the honest method; these shortcuts are what make GATE questions quick.

  • Mean value. The potential at the centre of any charge-free sphere is the average over its surface.
  • Superposition by symmetry. A cube with one face at V0 and five grounded: add the six rotated versions to get a cube entirely at V0, so the centre is at V0/6. A long square pipe with one wall at V0: centre V0/4. This even sums awkward Fourier series exactly.
  • The axis trick. For an azimuthally symmetric problem, find V on the z axis (often easy), expand in powers of z, and replace zl by rlPl(cos⁡θ) (or z−(l+1) by r−(l+1)Pl). A ring of radius a gives V=Q4πϵ0∑lalrl+1Pl(0)Pl(cos⁡θ) for r>a.
  • Green reciprocity. The charge induced on grounded conductors by a point charge equals that induced by any distribution with the same symmetry-averaged position: smear the point into a shell or a sheet. A charge between grounded concentric spheres a<r<b induces −qarb−rb−a on the inner one.
  • Fields of harmonic functions. Each component of E→ is harmonic too, but |E→|2 is subharmonic: the field strength is largest on the boundary.

Worked example. Two grounded concentric spheres of radii 1 cm and 3 cm have a charge q at r=2 cm. The inner sphere gets −q⋅12⋅12=−0.25q and the outer −q⋅32⋅12=−0.75q. No images, no series.

  • Dirichlet, Neumann or mixed data each fix the solution (Neumann up to a constant); Cauchy data over-determine it. Total charges on conductors also fix the field.
  • Cartesian: one direction oscillates, the other goes as sinh⁡/cosh⁡/exponentials. Fourier coefficients cn=2a∫fsin⁡(nπx/a)dx; a constant gives 4V0/nπ for odd n. In 3D the decay constant is kx2+ky2.
  • Spherical, azimuthal symmetry: ∑(Alrl+Blr−(l+1))Pl(cos⁡θ). Each surface mode travels alone; σl=(2l+1)ϵ0Al/R.
  • Sphere in E0: inside 3E0/(ϵr+2); conductor σ=3ϵ0E0cos⁡θ, p=4πϵ0R3E0. Cylinder across E0: σ=2ϵ0E0cos⁡ϕ.
  • 2D polar: a0+b0ln⁡s+∑(ansn+bns−n)(cos⁡nϕ,sin⁡nϕ). Wedge of angle β: E∝sπ/β−1.
  • Tricks: mean value, symmetric superposition (V0/6, V0/4), the axis trick, Green reciprocity.

Practise: exam questions on this lesson

These start at warm-up level and climb to top-rank questions. Check each answer, then read the full solution. More on the GATE Physics practice page.

Next: the method of images builds solutions for spheres, cylinders and planes from fictitious charges instead of series. Then multipole expansion reads the Legendre coefficients as charge, dipole and quadrupole moments. For the BSc foundations, see potential and electrostatic energy and Fourier series.

Check yourself

1A thin spherical surface is held at V0 cos²θ and encloses no charge. What is the potential at its centre?
2Outside a sphere held at V0 P2(cos θ), with no charge outside, how does the potential fall off?
3Near the sharp edge of a thin charged metal sheet, how does the surface charge density vary with distance s from the edge?

Sources

  • J. D. Jackson, Classical Electrodynamics, 3rd ed. (1999), chapters 1 to 3
  • D. J. Griffiths, Introduction to Electrodynamics, 4th ed. (2017), chapter 3
  • W. R. Smythe, Static and Dynamic Electricity, 3rd ed. (1968), chapters 4 and 5
  • G. B. Arfken, H. J. Weber and F. E. Harris, Mathematical Methods for Physicists, 7th ed. (2013), chapters 9 and 15
  • GATE Physics (PH) syllabus, Electromagnetic theory section, published by the organising institute each year

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Not affiliated with the IITs, IISc or the GATE organising institute. The general aptitude part of GATE is not covered here. Check the official GATE website for the current pattern. All questions are original.