UPSC 2026 Maths Optional Paper 2 Q7a — Step-by-Step Solution
15 marks · Section B
Second-order linear PDEs with constant coefficients (CF, PI) · PDEs · asked 13× in 14 yrs · Read the full method →
Question
Solve the partial differential equation
(2D2−5DD′+2D′2)z=24(y−x)+sin(y−x)
where D≡∂x∂, D′≡∂y∂, z=z(x,y).
Technique
A homogeneous linear PDE with constant coefficients (every term is of order 2), so the complementary function comes from the auxiliary equation 2m2−5m+2=0 with z=ϕ(y+mx). The trigger for the particular integral is that both pieces of the right-hand side are functions of the single combination u=y−x, i.e. of y+mx with m=−1. For such a right-hand side a homogeneous operator of degree n collapses to the constant F(m,1) times dn/dun, so the PI is obtained by n plain integrations — provided F(−1,1)=0, which must be checked because F(−1,1)=0 is exactly the resonant case where the shortcut fails.
Solution
Step 1 — Factorise the operator
Treat F(D,D′)=2D2−5DD′+2D′2 as a quadratic form in D,D′:
2D2−5DD′+2D′2=(2D−D′)(D−2D′).
Check by expansion: (2D−D′)(D−2D′)=2D2−4DD′−DD′+2D′2=2D2−5DD′+2D′2. ✓
The operator is homogeneous of degree 2, so the equation is a homogeneous linear PDE with constant coefficients and the standard theory applies: general solution = complementary function + particular integral, with two arbitrary functions.
Step 2 — The complementary function
The convention, stated (this is a mark). For a homogeneous operator, put z=ϕ(y+mx) with ϕ arbitrary and twice differentiable. Then
Hence ϕ(y+mx) solves F(D,D′)z=0iffm is a root of the auxiliary equation
F(m,1)=2m2−5m+2=0.
Solving: 2m2−5m+2=(2m−1)(m−2)=0, so
m1=2,m2=21.
The roots are real and distinct, so no repeated-root modification (no factor x) is needed, and
zc=ϕ1(y+2x)+ϕ2(y+2x),
with ϕ1,ϕ2 arbitrary twice-differentiable functions. Since ϕ2(y+2x)=ψ(2y+x) on writing ψ(τ)=ϕ2(τ/2) — and ψ is arbitrary exactly when ϕ2 is — the CF may equally be written
zc=ϕ(y+2x)+ψ(2y+x)
Two arbitrary functions, matching the order of the equation. ✓
Step 3 — Why the right-hand side is special, and the lemma that exploits it
The right-hand side is
R(x,y)=24(y−x)+sin(y−x),
which is a function of the single variableu=y−x=y+mx with m=−1. This is the whole point of the question.
Lemma. If F(D,D′) is homogeneous of degree n with constant coefficients and G is n-times differentiable, then
F(D,D′)G(y+mx)=F(m,1)G(n)(y+mx).
Proof. Each term of F is ckDn−kD′k, and Dn−kD′kG(y+mx)=mn−kG(n)(y+mx), because each D contributes a factor m and each D′ a factor 1. Summing over k gives (∑kckmn−k)G(n)=F(m,1)G(n). □
(Equation (∗) of Step 2 is this lemma with n=2, G=ϕ.)
Corollary (the shortcut, with its hypothesis). If the right-hand side is g(y+mx) and F(m,1)=0, then a particular integral is
zp=F(m,1)1ntimes∫⋯∫g(u)dunu=y+mx,
the constants of integration being discardable because they generate polynomials in u of degree <n, which already lie in the complementary function.
The hypothesis must be checked here, not assumed. With m=−1,
F(−1,1)=2(−1)2−5(−1)(1)+2(1)2=2+5+2=9=0.
Equivalently: m=−1 is not a root of 2m2−5m+2=0 (the roots are 2 and 21), so the direction y−x is not a characteristic of the operator and the case is non-resonant. Had −1 been a root, the two-fold integration would have produced 0=R and the method would have collapsed, requiring the usual x⋅(⋅) modification instead.
Discardability of the integration constants, verified explicitly. A general linear function α(y−x)+β does lie in the CF: taking ϕ(τ)=Aτ, ψ(τ)=Bτ gives A(y+2x)+B(2y+x)=(A+2B)y+(2A+B)x, and A=−α,B=α reproduces α(y−x); the constant β is absorbed by a constant ϕ. So dropping them costs nothing.
Step 4 — The particular integral
By the Corollary with n=2, m=−1, F(−1,1)=9 and g(u)=24u+sinu:
∫(24u+sinu)du=12u2−cosu,∫(12u2−cosu)du=4u3−sinu.
Hence
zp=91(4u3−sinu)u=y−x=94(y−x)3−91sin(y−x).
Direct verification by substitution (do this in the hall — it is two lines and it insures the whole PI). Write u=y−x, so ∂u/∂x=−1, ∂u/∂y=+1. For any twice-differentiable f,
D2f(u)=f′′(u),DD′f(u)=−f′′(u),D′2f(u)=f′′(u),
so
(2D2−5DD′+2D′2)f(u)=(2+5+2)f′′(u)=9f′′(u).
With f(u)=94u3−91sinu we get f′′(u)=38u+91sinu, and
9f′′(u)=24u+sinu=24(y−x)+sin(y−x).✓
(Cross-check of the trigonometric part by the standard operator rule. For sin(ax+by) one replaces D2→−a2, DD′→−ab, D′2→−b2. Here a=−1, b=1, giving 2(−1)−5(1)+2(−1)=−9=0, so
2D2−5DD′+2D′21sin(y−x)=−91sin(y−x),
in agreement.)
Step 5 — General solution
z=ϕ(y+2x)+ψ(2y+x)+94(y−x)3−91sin(y−x),
ϕ,ψ arbitrary twice-differentiable functions. ■
Step 6 — Remark: a second, equally correct PI for the polynomial part
Expanding F−1 in powers of D′/D (the routine textbook route) gives, for the part 24(y−x),
and indeed 2D2(6x2y+3x3)−5DD′(6x2y+3x3)=2(12y+18x)−5(12x)=24(y−x). ✓
This looks different from 94(y−x)3, but the two differ by a complementary function — as they must:
94(y−x)3−(6x2y+3x3)=−94(y+2x)3+91(2y+x)3.
Both are therefore correct particular integrals and both give the same general solution. Writing this identity out (one line) pre-empts the examiner who has only the other form on the marking scheme.
Answer
z=ϕ(y+2x)+ψ(2y+x)+94(y−x)3−91sin(y−x)
with ϕ,ψ arbitrary twice-differentiable functions (auxiliary roots m=2,21; PI legitimate because F(−1,1)=9=0).