Method of variation of parameters
At a Glance
- Frequency: 11 sub-parts across 11 of 13 years (2013, 2014, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2024, 2025)
- Priority tier: T1
- Marks (count): 10 (4), 13 (1), 15 (4), 20 (1), 8 (1)
- Average solve time: ~13 min
- Difficulty mix: medium 7, hard 3, easy 1
- Section: B | Dominant type: computation
Why This Chapter Matters
Variation of parameters appears in 11 of the last 13 years — only two years missed, making it the most consistently examined technique in Paper 1 Section B. The method works on any second-order linear ODE regardless of the right-hand side: it handles sec, log, tan, polynomials, and combinations that undetermined coefficients cannot touch. In every question the four-step skeleton is the same; the skill is recognising which variant you need (standard CF, Cauchy-Euler, or reduction-of-order) before the VoP machinery kicks in.
Minimum Theory
The formula. Given y′′+P(x)y′+Q(x)y=g(x) with two complementary functions y1,y2, seek yp=u1y1+u2y2 where
u1′=−Wy2g,u2′=Wy1g,W=y1y2′−y2y1′.
This is the standard form — the ODE must have leading coefficient 1 before you read off g(x). For ay′′+…=R(x), divide everything by a first; then g=R/a.
Two constraints, two equations. The conditions u1′y1+u2′y2=0 and u1′y1′+u2′y2′=g uniquely determine u1′ and u2′ via Cramer’s rule — the formulas above.
Getting both CF solutions. The VoP formula needs two independent homogeneous solutions. Three sub-cases arise:
- Constant coefficients (most common): solve the auxiliary equation; the roots give exponential, trigonometric, or repeated-root CF.
- Cauchy-Euler (x2y′′−…): try y=xm to get the indicial equation; roots m1,m2 give y1=xm1, y2=xm2.
- One CF given: use reduction of order (y2=v(x)y1) or operator factorisation to find the second.
Question Archetypes
Three patterns cover all VoP questions.
variation-of-parameters (5 question(s); 2013, 2014, 2016, 2017, 2018)
Recognition Cues
- The ODE has constant coefficients (or is first-order linear).
- The right-hand side is sec(ax), logx, tan(ax), or a polynomial — functions that undetermined coefficients cannot handle, or the problem explicitly says “use variation of parameters.”
- CF is found by the standard auxiliary equation.
Solution Template
- Write the auxiliary equation and find the two CF solutions y1,y2.
- Compute W=y1y2′−y2y1′.
- Write u1′=−y2g/W and u2′=y1g/W where g is the RHS of the standard-form ODE.
- Integrate to find u1,u2. Use standard integrals: ∫tan(ax)dx=−a1ln∣cos(ax)∣; ∫sec(2x)dx=21ln∣sec2x+tan2x∣; ∫eaxsin(bx)dx=a2+b2eax(asinbx−bcosbx); ∫xlogxdx=2x2logx−4x2.
- Write yp=u1y1+u2y2 and simplify using trig identities if needed.
- General solution: y=yc+yp.
Worked Example(s)
2013 Paper 1, 2013-P1-Q6b (10 marks)
Solve y′′+a2y=secax by variation of parameters.
CF. Auxiliary equation r2+a2=0⇒y1=cosax,y2=sinax.
Wronskian. W=acos2ax+asin2ax=a.
Integrands (g=secax):
u1′=−asinax⋅secax=−atanax⇒u1=a2ln∣cosax∣.
u2′=acosax⋅secax=a1⇒u2=ax.
Particular integral:
yp=a2cosaxln∣cosax∣+axsinax.
y=C1cosax+C2sinax+a2cosaxln∣cosax∣+axsinax.
2016 Paper 1, 2016-P1-Q6b (15 marks)
Solve (D2+2D+1)y=e−xlogx by variation of parameters.
CF. (D+1)2=0 — repeated root D=−1: y1=e−x, y2=xe−x.
Wronskian. W=e−2x.
Integrands (g=e−xlogx): the exponentials cancel: y2g/W=xlogx and y1g/W=logx.
∫xlogxdx=2x2logx−4x2,∫logxdx=xlogx−x.
Assemble:
yp=−e−x(2x2logx−4x2)+xe−x(xlogx−x)=e−x(2x2logx−43x2).
y=(c1+c2x)e−x+e−x(2x2logx−43x2).
2018 Paper 1, 2018-P1-Q6c (13 marks)
Solve y′′+16y=32sec2x.
CF. r2+16=0⇒y1=cos4x,y2=sin4x; W=4.
Compute u1,u2 (g=32sec2x):
u1′=−4sin4x⋅32sec2x=−8⋅cos2xsin4x.
Use sin4x=2sin2xcos2x: u1′=−16sin2x⇒u1=8cos2x.
u2′=4cos4x⋅32sec2x=8⋅cos2xcos4x.
Use cos4x=2cos22x−1: u2′=16cos2x−8sec2x⇒u2=8sin2x−4ln∣sec2x+tan2x∣.
Combine: u1y1+u2y2 has elementary part 8(cos2xcos4x+sin2xsin4x)=8cos2x:
yp=8cos2x−4sin4xln∣sec2x+tan2x∣.
y=C1cos4x+C2sin4x+8cos2x−4sin4xln∣sec2x+tan2x∣.
2014 Paper 1, 2014-P1-Q6a (10 marks)
Solve y′−5y=sinx by variation of parameters.
First-order case. CF: y1=e5x. Try yp=u(x)e5x; substituting: u′e5x=sinx⇒u′=e−5xsinx.
Using ∫eaxsinbx=a2+b2eax(asinbx−bcosbx) with a=−5,b=1:
u=−26e−5x(5sinx+cosx)⇒yp=−265sinx+cosx.
y=Ce5x−265sinx+cosx.
2017 Paper 1, 2017-P1-Q7b-ii (8 marks)
Solve y′′−y′−2y=44−76x−48x2 by variation of parameters.
CF. m2−m−2=0⇒m=2,−1; y1=e−x, y2=e2x; W=3ex.
Using ∫exPdx=ex(P−P′+P′′−⋯) with P=44−76x−48x2:
u1y1+u2y2=24x2+14x−5(exponential factors cancel).
y=C1e−x+C2e2x+24x2+14x−5.
Common Traps
- The Wronskian for {cosax,sinax} is a, not 1. A dropped factor of a changes every subsequent result.
- For sec2x: the key identities are sin4x/cos2x=2sin2x and cos4x/cos2x=2cos2x−sec2x. Without them the integrals look intractable.
- Repeated root gives y2=xe−x — using e−x twice gives W=0, and the formula collapses.
- In the 2016 problem, the exponentials cancel in both integrands, leaving ∫xlogx and ∫logx — keep logx, do not differentiate it away.
- The formula has a minus sign on the first piece: yp=−y1∫Wy2g+y2∫Wy1g.
vop-cauchy-euler (2 question(s); 2019, 2024)
Recognition Cues
- The homogeneous ODE has the form x2y′′−2xy′+2y=0 or similar (powers of x matching the order of differentiation — Euler-Cauchy type).
- The full equation is x2y′′−…=xnsinx or similar polynomial-trig forcing.
- You must find y1,y2 first by the indicial method, then apply VoP to the normalised equation.
Solution Template
- Solve the homogeneous Cauchy-Euler ODE by trying y=xm; get the indicial equation.
- Write down y1=xm1 and y2=xm2.
- Divide the full ODE by x2 to get standard form; read off g(x) from the normalised RHS.
- Compute W=y1y2′−y2y1′.
- Apply VoP: u1′=−y2g/W, u2′=y1g/W; integrate; assemble yp.
Worked Example(s)
2019 Paper 1, 2019-P1-Q7a (15 marks) and 2024 Paper 1, 2024-P1-Q6c-ii (10 marks)
Find the independent CF solutions of x2y′′−2xy′+2y=x3sinx and find the general solution by variation of parameters.
Sources:,
Indicial equation. Try y=xm in x2y′′−2xy′+2y=0: m(m−1)−2m+2=m2−3m+2=(m−1)(m−2)=0, giving
y1=x,y2=x2.
Standard form. Divide by x2: y′′−(2/x)y′+(2/x2)y=xsinx, so g(x)=xsinx.
Wronskian. W=x⋅2x−x2⋅1=x2.
VoP integrals:
u1′=−x2x2⋅xsinx=−xsinx⇒u1=xcosx−sinx.
u2′=x2x⋅xsinx=sinx⇒u2=−cosx.
Particular integral:
yp=(xcosx−sinx)⋅x+(−cosx)⋅x2=x2cosx−xsinx−x2cosx=−xsinx.
y=C1x+C2x2−xsinx.
Common Traps
- The forcing in the VoP formulas is g=xsinx (the normalised RHS after dividing by x2), not the raw x3sinx. Using the raw RHS gives yp wrong by a factor of x2.
- The x2cosx terms cancel in yp — a nice simplification; do not leave them unsimplified.
- The Wronskian is W=x2 (from y1=x,y2=x2); it appears in every denominator.
vop-after-reduction (4 question(s); 2020, 2021, 2022, 2025)
Recognition Cues
- One CF solution y1 is given (e.g., y1=e−x, y1=x).
- Or the equation has variable coefficients and you must find y2 yourself.
- The second step is always: find y2 by reduction of order (try y2=v(x)y1) or operator factorisation.
Solution Template
- Accept the given y1 (or find it by inspection/trial).
- Reduction of order: substitute y2=v(x)y1 into the homogeneous ODE; reduce to a first-order ODE in u=v′; integrate twice to get v and hence y2.
- Divide the full ODE by its leading coefficient to get standard form; read off g(x).
- Compute W=y1y2′−y2y1′.
- Apply VoP: u1′=−y2g/W, u2′=y1g/W; integrate; assemble yp.
Worked Example(s)
2021 Paper 1, 2021-P1-Q8a-ii (10 marks)
Solve x2y′′−2x(1+x)y′+2(1+x)y=x3 by variation of parameters, given y1=x satisfies the homogeneous equation.
Find y2. Try y2=v(x)⋅x. Substituting into the homogeneous ODE and simplifying:
x3v′′−2x3v′=0⇒v′′=2v′⇒u=v′=e2x⇒v=e2x/2.
Take y2=xe2x.
Standard form. Divide by x2: g(x)=x3/x2=x.
Wronskian. W=x⋅(1+2x)e2x−xe2x⋅1=2x2e2x.
VoP: u2′=2x2e2xx⋅x=21e−2x⇒u2=−41e−2x. Then u1′=−21⇒u1=−x/2.
yp=(−x/2)(x)+(−e−2x/4)(xe2x)=−x2/2−x/4.
The −x/4 is absorbed into C1x:
y=C1x+C2xe2x−2x2.
2022 Paper 1, 2022-P1-Q6b (15 marks)
Solve (x2−1)y′′−2xy′+2y=(x2−1)2 by variation of parameters, given y1=x.
Find y2. Try y2=vx; substituting into the homogeneous equation:
x(x2−1)v′′−2v′=0⇒v′dv′=x(x2−1)2dx.
Partial fractions: x(x−1)(x+1)2=−x2+x−11+x+11. Integrating: v′=(x2−1)/x2=1−1/x2, so v=x+1/x and y2=x2+1.
Standard form. Divide by (x2−1): g=x2−1.
Wronskian. W=x⋅2x−(x2+1)⋅1=x2−1.
VoP: u2′=x⇒u2=x2/2; u1′=−(x2+1)⇒u1=−x3/3−x.
yp=(−x3/3−x)⋅x+(x2/2)⋅(x2+1)=6x4−2x2.
y=C1x+C2(x2+1)+6x4−2x2.
2020 Paper 1, 2020-P1-Q6a (20 marks)
Solve y′′+(1−cotx)y′−ycotx=sin2x by variation of parameters, given y1=e−x.
Find y2. Factor the operator: (D−cotx)(D+1)y=0. Letting v=(D+1)y: (D−cotx)v=0⇒v=sinx. Solve (D+1)y=sinx: integrating factor ex: y2=sinx−cosx.
Wronskian. W=2e−xsinx.
VoP:
u2′=2e−xsinxe−xsin2x=2sinx⇒u2=−2cosx.
u1′=−2e−xsinx(sinx−cosx)sin2x=−4ex(1−cos2x−sin2x)⇒u1=20ex(3sin2x−cos2x−5).
Assembling and simplifying (constant terms cancel):
yp=−101sin2x+51cos2x.
y=C1e−x+C2(sinx−cosx)−10sin2x+5cos2x.
2025 Paper 1, 2025-P1-Q8a (15 marks)
Solve (x+2)y′′−(2x+5)y′+2y=(1+x)ex by variation of parameters.
CF by inspection. Try y=emx: matching coefficients of x gives m=2, yielding y1=e2x. Try y=αx+β: gives y2=2x+5.
Standard form. Divide by (x+2): g=(1+x)ex/(x+2).
Wronskian. W=−4(x+2)e2x.
VoP: u2′=−4(x+2)2(x+1)ex; use dxdx+2ex=(x+2)2(x+1)ex, so u2=−4(x+2)ex.
After assembling, both terms combine to give yp=−ex.
y=C1e2x+C2(2x+5)−ex.
Common Traps
- Variable-coefficient ODEs require standard form first. Divide by the leading coefficient before reading off g(x) for the VoP integrands.
- The reduction-of-order substitution y2=vy1 always reduces to a first-order ODE in u=v′, separable or easily solvable — keep substituting until you see it.
- For the 2022 problem, the partial-fraction calculation x(x2−1)2=−x2+x−11+x+11 is the key step; skip it and the whole integral breaks.
- The constant term −x/4 from yp in the 2021 problem merges into C1x in the general solution — this is correct and saves rewriting.
- In the 2020 operator-factorisation: the factor (D−cotx) is the left factor; the second CF is found by solving a first-order equation for v=(D+1)y.
Marks-Aware Writing
8-mark questions (2017): CF + Wronskian (one line each), two VoP integrals (show clearly), assemble and simplify. Since forcing is polynomial, the exponentials in u1y1+u2y2 cancel — state this explicitly.
10-mark questions (2013, 2014, 2024): Write CF and Wronskian; show both integrands; integrate; write yp neatly; state the general solution. Four to five working lines suffice.
13-15-mark questions (2016, 2018, 2019, 2022, 2025): Full layout — CF with justification, Wronskian calculation (show the 2×2 determinant), both integrands and their integrals step by step, assembled yp, boxed final answer.
20-mark questions (2020): All of the above, plus the derivation of y2 (operator factorisation or reduction of order written out in full), Wronskian computation, all four integrals shown. Double-check the constant cancellations.
Practice Set
| Year | Paper/Q | Marks | Archetype | One-line hint |
|---|
| 2025 | P1-Q8a | 15 | vop-after-reduction | CF: e2x and 2x+5 by inspection; W=−4(x+2)e2x; yp=−ex telescopes neatly |
| 2024 | P1-Q6c-ii | 10 | vop-cauchy-euler | Same as 2019; indicial roots 1,2; g=xsinx; yp=−xsinx |
| 2022 | P1-Q6b | 15 | vop-after-reduction | Partial fractions to find v′=(x2−1)/x2; y2=x2+1; yp=x4/6−x2/2 |
| 2021 | P1-Q8a-ii | 10 | vop-after-reduction | y2=xe2x from v′′=2v′; standard form g=x; yp=−x2/2 |
| 2020 | P1-Q6a | 20 | vop-after-reduction | Operator factor (D−cotx)(D+1); y2=sinx−cosx; yp=−101sin2x+51cos2x |
| 2019 | P1-Q7a | 15 | vop-cauchy-euler | Indicial roots 1,2; g=xsinx after dividing by x2; yp=−xsinx |
| 2018 | P1-Q6c | 13 | variation-of-parameters | W=4; double-angle identities; $y_p=8\cos2x-4\sin4x\ln |
| 2017 | P1-Q7b-ii | 8 | variation-of-parameters | W=3ex; ∫exPdx shortcut; exponentials cancel; yp=24x2+14x−5 |
| 2016 | P1-Q6b | 15 | variation-of-parameters | Repeated root; W=e−2x; ∫xlogx and ∫logx; yp=e−x(2x2logx−43x2) |
| 2014 | P1-Q6a | 10 | variation-of-parameters | First-order VoP; ∫e−5xsinxdx; denominator 26; yp=−(5sinx+cosx)/26 |
| 2013 | P1-Q6b | 10 | variation-of-parameters | W=a; $\int\tan ax=-\tfrac{1}{a}\ln |