← 2026 Paper 2

UPSC 2026 Maths Optional Paper 2 Q8a — Step-by-Step Solution

15 marks · Section B

Laplace equation: Dirichlet/Neumann, separation of variables · PDEs · asked 5× in 14 yrs · Read the full method →

Question

Two-dimensional harmonic equation in plane polar coordinates (r,θ)(r, \theta) takes the form:

∂2V∂r2+1r∂V∂r+1r2∂2V∂θ2=0\frac{\partial^2 V}{\partial r^2} + \frac{1}{r}\frac{\partial V}{\partial r} + \frac{1}{r^2}\frac{\partial^2 V}{\partial\theta^2} = 0

Deduce that it has solutions of the form (Arn+Br−n)e±inθ(Ar^n + Br^{-n})e^{\pm in\theta}, where AA, BB, nn are constants. Determine VV if it satisfies the two-dimensional harmonic equation in the region 0≤r≤a0 \leq r \leq a, 0≤θ≤2π0 \leq \theta \leq 2\pi and satisfies the conditions:

(i) VV remains finite as r→0r \to 0,

(ii) V=∑ncncos⁡(nθ)V = \displaystyle\sum_{n} c_n\cos(n\theta) on r=ar = a.

Technique

Separation of variables V=R(r)Θ(θ)V = R(r)\Theta(\theta) turns Laplace’s equation on the disc into an angular oscillator plus a Cauchy–Euler radial equation r2R′′+rR′−n2R=0r^2R'' + rR' - n^2R = 0, whose trial solution R=rmR = r^m gives m=±nm = \pm n — that is exactly the printed form. Two things then do all the work: single-valuedness in θ\theta forces nn to be an integer, and n=0n = 0 is a repeated indicial root, so it carries the separate pair {1, log⁡r}\{1,\ \log r\} that the printed form does not contain. Finiteness at the origin kills r−nr^{-n} and log⁡r\log r; matching Fourier coefficients on r=ar = a fixes the rest.

Solution

Step 1 — Separate the variables

Look for a non-trivial product solution

V(r,θ)=R(r) Θ(θ),R≢0, Θ≢0.V(r,\theta) = R(r)\,\Theta(\theta),\qquad R\not\equiv 0,\ \Theta\not\equiv 0 .

Substituting into Vrr+1rVr+1r2Vθθ=0V_{rr} + \dfrac1r V_r + \dfrac1{r^2}V_{\theta\theta} = 0,

R′′Θ+1rR′Θ+1r2R Θ′′=0.R''\Theta + \frac{1}{r}R'\Theta + \frac{1}{r^2}R\,\Theta'' = 0 .

Multiply through by r2RΘ\dfrac{r^2}{R\Theta} (legitimate wherever RΘ≠0R\Theta \neq 0) and separate:

r2R′′+rR′R  =  −Θ′′Θ.\frac{r^2R'' + rR'}{R} \;=\; -\frac{\Theta''}{\Theta} .

The left side depends on rr alone and the right on θ\theta alone, so each equals a common constant. Call it λ\lambda:

Θ′′+λ Θ=0,(1)\Theta'' + \lambda\,\Theta = 0, \tag{1} r2R′′+rR′−λR=0.(2)r^2R'' + rR' - \lambda R = 0. \tag{2}

Step 2 — The radial equation is Cauchy–Euler; deduce r±nr^{\pm n}

Equation (2) is homogeneous of Cauchy–Euler (equidimensional) type: substituting the trial solution R=rmR = r^m gives

m(m−1)rm+mrm−λrm=0⟹m2=λ  (indicial equation).m(m-1)r^{m} + m r^{m} - \lambda r^{m} = 0 \quad\Longrightarrow\quad \boxed{m^2 = \lambda}\ \ \text{(indicial equation)} .

Write n=λn = \sqrt{\lambda}, so that λ=n2\lambda = n^2 and m=±nm = \pm n.

R(r)=Arn+Br−n.R(r) = A r^{n} + B r^{-n}. Θ(θ)=e inθande−inθ.\Theta(\theta) = e^{\,in\theta}\quad\text{and}\quad e^{-in\theta}.

Multiplying,

  V=(Arn+Br−n)e±inθ  \boxed{\;V = \bigl(Ar^{n} + Br^{-n}\bigr)e^{\pm i n\theta}\;}

which is the form asserted in the question. ■\qquad\blacksquare

Replacing nn by −n-n merely interchanges A↔BA \leftrightarrow B and the two exponentials, so no generality is lost by taking n≥0n \ge 0.

Step 3 — The exceptional case n=0n = 0 (which the printed form omits)

If n=0n = 0 the indicial equation m2=0m^2 = 0 has a repeated root, and rn,r−nr^{n}, r^{-n} both collapse to the single solution 11. The second solution is obtained from (2) directly: with λ=0\lambda = 0,

r2R′′+rR′=0  ⟺  (rR′)′=0  ⟺  rR′=const  ⟺  R=A0+B0log⁡r.r^2R'' + rR' = 0 \iff \bigl(rR'\bigr)' = 0 \iff rR' = \text{const} \iff R = A_0 + B_0\log r .

Likewise (1) becomes Θ′′=0\Theta'' = 0, giving Θ=C0+D0θ\Theta = C_0 + D_0\theta. So the n=0n=0 family is

V=(A0+B0log⁡r)(C0+D0θ),V = \bigl(A_0 + B_0\log r\bigr)\bigl(C_0 + D_0\theta\bigr),

not covered by (Arn+Br−n)e±inθ(Ar^n + Br^{-n})e^{\pm in\theta}. This matters here, because the boundary data in (ii) contains the term n=0n = 0.

Step 4 — Single-valuedness forces n∈Zn \in \mathbb{Z} and kills the θ\theta-linear term

The point (r,θ)(r,\theta) and the point (r,θ+2π)(r,\theta + 2\pi) are the same physical point of the disc. A potential must therefore be single-valued:

V(r,θ+2π)=V(r,θ)for all r,θ.V(r,\theta + 2\pi) = V(r,\theta)\quad\text{for all }r,\theta .

For the n≠0n \neq 0 family this requires ein(θ+2π)=einθe^{in(\theta + 2\pi)} = e^{in\theta}, i.e.

e2πin=1⟹2πin∈2πi Z⟹n∈Z.e^{2\pi i n} = 1 \quad\Longrightarrow\quad 2\pi i n \in 2\pi i\,\mathbb{Z} \quad\Longrightarrow\quad n \in \mathbb{Z}.

(Note this rules out complex λ\lambda automatically: a non-real nn would give a factor e−(Im⁡n)θe^{-(\operatorname{Im}n)\theta}, which is not 2π2\pi-periodic. In particular λ≥0\lambda \ge 0.)

For the n=0n = 0 family, single-valuedness forces D0=0D_0 = 0, removing the θ\theta term.

Hence, taking n=0,1,2,…n = 0,1,2,\dots and forming real combinations cos⁡nθ, sin⁡nθ\cos n\theta,\ \sin n\theta of e±inθe^{\pm in\theta}, the general single-valued solution on the punctured disc 0<r≤a0 < r \le a is

V(r,θ)=A0+B0log⁡r+∑n=1∞(Anrn+Bnr−n)cos⁡nθ+∑n=1∞(An′rn+Bn′r−n)sin⁡nθ.(3)V(r,\theta) = A_0 + B_0\log r + \sum_{n=1}^{\infty}\bigl(A_n r^{n} + B_n r^{-n}\bigr)\cos n\theta + \sum_{n=1}^{\infty}\bigl(A_n' r^{n} + B_n' r^{-n}\bigr)\sin n\theta. \tag{3}

Step 5 — Impose condition (i): finiteness as r→0r \to 0

As r→0+r \to 0^{+},

log⁡r→−∞andr−n→+∞  (n≥1),\log r \to -\infty \qquad\text{and}\qquad r^{-n}\to +\infty\ \ (n \ge 1),

while rn→0r^{n} \to 0 and the constant stays bounded. Since the terms of (3) are mutually orthogonal in θ\theta, each Fourier mode must be separately bounded, and boundedness of VV near the origin therefore forces

B0=0,Bn=Bn′=0(n≥1).B_0 = 0,\qquad B_n = B_n' = 0 \quad (n \ge 1).

So

V(r,θ)=A0+∑n=1∞rn(Ancos⁡nθ+An′sin⁡nθ).(4)V(r,\theta) = A_0 + \sum_{n=1}^{\infty} r^{n}\bigl(A_n\cos n\theta + A_n'\sin n\theta\bigr). \tag{4}

Condition (i) is not decorative — it is what makes the problem well-posed. Without it, log⁡(r/a)\log(r/a) and (anr−n−a−nrn)cos⁡nθ\bigl(a^{n}r^{-n} - a^{-n}r^{n}\bigr)\cos n\theta are harmonic on 0<r≤a0 < r \le a and vanish identically on r=ar = a; they could be added to any solution without disturbing (ii). Finiteness at the origin is exactly the condition that removes them.

Step 6 — Impose condition (ii): match Fourier coefficients on r=ar = a

Take the index set in (ii) to be n=0,1,2,…n = 0,1,2,\dots, the n=0n=0 term being the constant c0c_0 (since cos⁡0=1\cos 0 = 1); assume ∑n∣cn∣<∞\sum_n |c_n| < \infty so the boundary series converges uniformly. Setting r=ar = a in (4),

A0+∑n=1∞an(Ancos⁡nθ+An′sin⁡nθ)  =  c0+∑n=1∞cncos⁡nθ,0≤θ≤2π.A_0 + \sum_{n=1}^{\infty} a^{n}\bigl(A_n\cos n\theta + A_n'\sin n\theta\bigr) \;=\; c_0 + \sum_{n=1}^{\infty} c_n\cos n\theta,\qquad 0\le\theta\le 2\pi .

Both sides are uniformly convergent Fourier series on [0,2π][0,2\pi], so their coefficients agree. Extracting them by orthogonality — multiply by 11, by cos⁡mθ\cos m\theta, by sin⁡mθ\sin m\theta and integrate over [0,2π][0,2\pi], using

∫02π ⁣cos⁡mθcos⁡nθ dθ=πδmn,∫02π ⁣sin⁡mθsin⁡nθ dθ=πδmn,∫02π ⁣sin⁡mθcos⁡nθ dθ=0  (m,n≥1)\int_0^{2\pi}\!\cos m\theta\cos n\theta\,d\theta = \pi\delta_{mn},\quad \int_0^{2\pi}\!\sin m\theta\sin n\theta\,d\theta = \pi\delta_{mn},\quad \int_0^{2\pi}\!\sin m\theta\cos n\theta\,d\theta = 0\ \ (m,n\ge1)

— gives

A0=c0,anAn=cn,anAn′=0.A_0 = c_0,\qquad a^{n}A_n = c_n,\qquad a^{n}A_n' = 0 .

Hence An′=0A_n' = 0 for every nn (no sine modes: the data is even in θ\theta, and so therefore is VV), and An=cn/anA_n = c_n/a^{n}.

Step 7 — The solution, its validity, and uniqueness

V(r,θ)=c0+∑n=1∞cn(ra)ncos⁡nθ  =  ∑n≥0cn rnancos⁡nθ.V(r,\theta) = c_0 + \sum_{n=1}^{\infty} c_n\left(\frac{r}{a}\right)^{n}\cos n\theta \;=\; \sum_{n\ge 0} c_n\,\frac{r^{n}}{a^{n}}\cos n\theta .

It really is harmonic. For 0≤r≤ρ<a0 \le r \le \rho < a we have ∣cn(r/a)n∣≤∣cn∣(ρ/a)n\bigl|c_n (r/a)^n\bigr| \le |c_n|(\rho/a)^n, and after kk term-by-term differentiations each term is bounded by ∣cn∣ nkρ−k(ρ/a)n|c_n|\,n^{k}\rho^{-k}(\rho/a)^{n}; since ρ/a<1\rho/a < 1 and ∑∣cn∣<∞\sum |c_n| < \infty, these majorants are summable. By the Weierstrass MM-test the differentiated series converge uniformly on every closed sub-disc, so term-by-term differentiation is valid. Each term rncos⁡nθr^{n}\cos n\theta satisfies the equation by Step 2, hence so does VV.

It is the only such solution. VV is harmonic on the open disc and continuous on the closed disc (the boundary series converges uniformly), and by the maximum principle for harmonic functions the difference of two such solutions is harmonic with zero boundary values, hence identically zero. So the answer is unique.

Sanity value at the centre. V(0,θ)=c0V(0,\theta) = c_0, which is precisely the mean of the boundary data 12π∫02π∑ncncos⁡nθ dθ=c0\frac{1}{2\pi}\int_0^{2\pi}\sum_n c_n\cos n\theta\,d\theta = c_0 — the mean-value property of harmonic functions. ■\qquad\blacksquare

Remark (closed integral form). Since a2−r2a2−2arcos⁡(θ−ϕ)+r2=1+2∑n≥1(ra)ncos⁡n(θ−ϕ)\dfrac{a^2-r^2}{a^2-2ar\cos(\theta-\phi)+r^2} = 1 + 2\sum_{n\ge1}\left(\dfrac{r}{a}\right)^{n}\cos n(\theta-\phi), the same answer can be written as the Poisson integral

V(r,θ)=12π∫02π(a2−r2) f(ϕ)a2−2arcos⁡(θ−ϕ)+r2 dϕ,f(ϕ)=∑ncncos⁡nϕ,V(r,\theta) = \frac{1}{2\pi}\int_0^{2\pi}\frac{(a^2-r^2)\,f(\phi)}{a^2-2ar\cos(\theta-\phi)+r^2}\,d\phi,\qquad f(\phi) = \sum_n c_n\cos n\phi,

which is a useful cross-check but is not needed for the marks.

Answer

  V=(Arn+Br−n)e±inθ  (n≠0),V=A0+B0log⁡r  (n=0);n∈Z by single-valuedness  \boxed{\;V=\bigl(Ar^{n}+Br^{-n}\bigr)e^{\pm in\theta}\ \ (n\neq0),\qquad V=A_0+B_0\log r\ \ (n=0);\qquad n\in\mathbb{Z}\ \text{by single-valuedness}\;}   V(r,θ)  =  c0  +  ∑n=1∞cn(ra) ⁣ncos⁡nθ  =  ∑n≥0cn rnan cos⁡nθ,0≤r≤a.  \boxed{\;V(r,\theta) \;=\; c_0 \;+\; \sum_{n=1}^{\infty} c_n\left(\frac{r}{a}\right)^{\!n}\cos n\theta \;=\; \sum_{n\ge0} c_n\,\frac{r^{n}}{a^{n}}\,\cos n\theta,\qquad 0\le r\le a .\;}
We post more of this — worked solutions, CSAT trap breakdowns, guide chapters — a few times a week on Telegram. Free, no sign-in. Join

This solution is part of the Maths Coverage Map — 14 years, mapped. Get the take-away PDF free.