UPSC 2015 Maths Optional Paper 2 Q6a — Step-by-Step Solution
15 marks · Section B
Quasilinear first-order PDEs (Lagrange's method) · PDEs · asked 9× in 13 yrs · Read the full method →
Question
Solve for the general solution pcos(x+y)+qsin(x+y)=z, where p=∂z/∂x and q=∂z/∂y.
Technique
Lagrange auxiliary system; clever substitution u=x+y, v=x−y to decouple; first integral via du/(cosu+sinu)=dz/z (giving the half-angle tangent form); second integral via du/(cosu+sinu)=dv/(cosu−sinu) (giving v−ln∣cosu+sinu∣).
Solution
Lagrange’s method. Auxiliary system:
cos(x+y)dx=sin(x+y)dy=zdz.
Step 1 — First integral
Add the first two ratios:
cos(x+y)+sin(x+y)dx+dy=zdz.
Let u=x+y, du=dx+dy:
cosu+sinudu=zdz.
The LHS denominator: cosu+sinu=2sin(u+π/4) (or equivalently 2cos(u−π/4)).
Let w=x−y (independent variable), then x+y=u is independent. We have dw in the numerator, 2cos(u+π/4) in the denominator (depends on u, not w). Hmm — these aren’t immediately separable.
Different approach: integrate dx−dy and dz/z directly?
cos(x+y)−sin(x+y)d(x−y)=zdz.
To proceed, eliminate using the relation from Step 1: u=x+y is implicitly tied to z via (⋆). Then cosu−sinu can be expressed in terms of z, and we have d(x−y) in terms of z and the differential.
Alternative: use the first integral (⋆) to express u in terms of z, then cosu−sinu becomes a function of z alone. Integrating d(x−y)=f(z)dz/z gives x−y as a function of z, a second integral.
This is getting algebraically complex. Let me try a cleaner approach.
Step 2’ — Alternative second integral via the differential of u−2y or similar
Try: zdz=cosudx=sinudy. So dz/z along the characteristics has two expressions; equate them?
Easier: subtract again. cosudx−sinudy=sinucosusinudx−cosudy. Set this equal to some other ratio?
Let me try sinucosu−sinucosusinudx−cosudy=0... — degenerate.
Cleaner: use multipliers sinu,−cosu:
sinucosu−cosusinusinudx−cosudy=0sinudx−cosudy.
So sinudx−cosudy must equal 0 along the characteristics. Let’s check:
sinudx−cosudy: along dx/cosu=dy/sinu=kdt (parameterise), dx=kcosudt, dy=ksinudt. Then sinudx−cosudy=k(sinucosu−cosusinu)dt=0 ✓.
So sinudx−cosudy=0 along characteristics. But u=x+y depends on (x,y), so this isn’t an exact differential immediately. Let’s try:
d(sinux−cosuy⋅stuff)… let me try d(xsinu+ycosu):
d(xsinu+ycosu)=sinudx+xcosudu+cosudy−ysinudu=(sinudx+cosudy)+(xcosu−ysinu)du.
Closer but extra term (xcosu+ysinu)du=0 along characteristics.
This problem is genuinely awkward. Let me try yet another approach.
Step 2” — Multipliers (1,1,0) then (1,−1,0)
(1,1,0): numerator dx+dy, denom cosu+sinu. Already used (Step 1).
(1,−1,0): numerator dx−dy, denom cosu−sinu.
So cosu−sinudx−dy=zdz (using the third ratio).
Wait — that uses the property that if A/a=B/b=C/c, then for any λ,μ: (λA+μB)/(λa+μb)=C/c (provided denom nonzero). So cosu−sinudx−dy=zdz.
Now d(x−y)=dx−dy, but the denom cosu−sinu=cos(x+y)−sin(x+y) depends on (x+y). So integration isn’t straightforward unless we change variable.
Hmm — note that we already have Step 1’s relation expressing u in terms of z (and C1). So along a characteristic, u is a function of z, and cosu−sinu becomes a function of z:
cosu−sinu=2cos(u+π/4).
From Step 1: tan(u/2+π/8)=Bz2 where B depends on C1. The connection from u to z is implicit.
This is getting complicated. Let me try the symmetric approach.
Step 2''' — Use dx−dy and dy/(sinu)
Try cosu−sinudx−dy=zdz — this gives a relation between d(x−y) and dz.
But we also have u=x+y, and from Step 1, u relates to z via tan(u/2+π/8)= const ⋅z2.
So cosu−sinu is a function of z. Let me denote F(z)=cosu(z)−sinu(z), where u(z) is the function defined by the Step 1 first integral tan(u/2+π/8)=Cz2.
Then d(x−y)=F(z)zdz, and integrating gives x−y=G(z)+C2 for some G. The second integral is x−y−G(z)=C2.
This works in principle but the explicit G(z) is messy.
Simpler approach: parametric characteristics
Maybe the cleanest answer is to leave the general solution in terms of two arbitrary integrals:
Φ(z[tan(2x+y+8π)]1/2,(second integral))=0,
with the second integral implicitly defined by the relation derived above.
A different formulation: pcosu+qsinu=z where u=x+y
Since the coefficient is a function of u=x+y only, substitute u=x+y, v=x−y as new independent variables. Then ∂/∂x=∂/∂u+∂/∂v, ∂/∂y=∂/∂u−∂/∂v.
p=zx=zu+zv, q=zy=zu−zv.
PDE: (zu+zv)cosu+(zu−zv)sinu=z, i.e.
zu(cosu+sinu)+zv(cosu−sinu)=z.
This is still a first-order linear PDE in (u,v) but with u-dependent coefficients.
Characteristics: cosu+sinudu=cosu−sinudv=zdz.
First integral (from du,dz): as in Step 1, z[tan(u/2+π/8)]1/2=C1.
Second integral (from du,dv): dudv=cosu+sinucosu−sinu. RHS depends only on u, so integrable:
cosu+sinucosu−sinu=dudln∣cosu+sinu∣=sinu+cosu−sinu+cosu ✓. So