Reduction of Order

Reduction of order finds a second solution of a second-order linear homogeneous equation when you already know one solution y₁. Substitute y₂ = v(x)y₁(x). The terms with v itself cancel, leaving a first-order equation for w = v′. Solve for w, integrate to get v, and then y₂ = vy₁.

Before this lesson: Second-Order Homogeneous Linear Equations, The Wronskian and Separable Differential Equations

The idea: multiply the known solution by a function

If y₁ solves the equation, so does Cy₁ for any constant C. Those are not new, though. To get a different solution, let the constant become a function: try y = v(x)y₁(x).

When you substitute, every term that still contains v (not v′ or v″) collects into v times the original equation applied to y₁. That is 0, because y₁ is a solution. Only v′ and v″ remain, so setting w = v′ lowers the order from 2 to 1. That is where the name comes from.

Any nonconstant v works, so you can drop constants of integration along the way. The result y₂ = vy₁ is not a constant multiple of y₁, so {y₁, y₂} is a fundamental set.

Graph for t from 0 to 3 of three curves for the spring equation x″ + 6x′ + 9x = 0. The known solution e^(−3t) starts at 1 and decays. The second solution from reduction of order, te^(−3t), starts at 0, rises to a peak of about 0.12 at t = 1/3 and decays. The motion x(t) = (0.5 + 1.5t)e^(−3t) starts at 0.5 with a flat start and decays to 0 without crossing it.
Reduction of order turns the known solution e^(−3t) into a second one, te^(−3t). Together they describe every motion, such as x(t) = (0.5 + 1.5t)e^(−3t).

Reduction of order formula

For y″ + p(x)y′ + q(x)y = 0 with a known solution y₁, substituting y = vy₁ gives

y_1v'' + \left(2y_1' + p\,y_1\right)v' = 0 \quad\Longrightarrow\quad v' = \dfrac{e^{-\int p(x)\,dx}}{y_1^2}
y_2 = y_1\displaystyle\int \dfrac{e^{-\int p(x)\,dx}}{y_1^2}\,dx

The equation must be in standard form, with the coefficient of y″ equal to 1, before you read off p(x).

How to use reduction of order

  1. Write the equation in standard form y″ + p(x)y′ + q(x)y = 0 and identify p.
  2. Compute e^(−∫p dx), dropping the constant.
  3. Divide by y₁² to get v′, then integrate to get v (again dropping the constant).
  4. Multiply: y₂ = vy₁. The general solution is y = C₁y₁ + C₂y₂.

You can also substitute y = vy₁ directly, as in Examples 1 and 2. That is longer but shows why the method works.

Reduction of order examples

Example 1: A repeated root, by direct substitution

The equation y″ − 4y′ + 4y = 0 has the solution y₁ = e^(2x). Find a second solution.

y = ve^{2x}

Let the constant multiplier become a function v(x).

y' = v'e^{2x} + 2ve^{2x}

Differentiate with the product rule.

y'' = v''e^{2x} + 4v'e^{2x} + 4ve^{2x}

Differentiate again and combine the two v′e^(2x) terms.

v''e^{2x} + 4v'e^{2x} + 4ve^{2x} - 4v'e^{2x} - 8ve^{2x} + 4ve^{2x} = 0

Substitute into y″ − 4y′ + 4y = 0 and distribute.

v''e^{2x} = 0

Combine like terms: the v′ terms and the v terms both cancel.

v'' = 0

Divide both sides by e^(2x), which is never 0.

v = x

Integrate twice, keeping the simplest nonconstant choice.

y_2 = xe^{2x}

Multiply v by y₁.

Answer: y₂ = xe^(2x), and the general solution is y = (C₁ + C₂x)e^(2x). The characteristic equation r² − 4r + 4 = 0 has the repeated root 2, which is why the usual method gives only one exponential.

Example 2: Variable coefficients, by direct substitution

For positive x, y₁ = x² solves x²y″ − 3xy′ + 4y = 0. Find a second solution.

y = vx^2,\quad y' = v'x^2 + 2xv

Substitute y = vx² and differentiate with the product rule.

y'' = v''x^2 + 4xv' + 2v

Differentiate again and combine like terms.

x^4v'' + 4x^3v' + 2x^2v - 3x^3v' - 6x^2v + 4x^2v = 0

Substitute into the equation and distribute.

x^4v'' + x^3v' = 0

Combine like terms: the v terms cancel.

xw' + w = 0

Let w = v′ and divide both sides by x³.

\dfrac{dw}{w} = -\dfrac{dx}{x}

Separate the variables.

\ln w = -\ln x \Rightarrow w = \dfrac{1}{x}

Integrate both sides and exponentiate, dropping the constant (see properties of logarithms).

v = \ln x

Integrate w = 1/x (see integrals that give natural logarithms).

y_2 = x^2\ln x

Multiply v by y₁.

Answer: y₂ = x² ln x, so y = C₁x² + C₂x² ln x for positive x.

Example 3: Using the formula

For positive x, y₁ = x solves x²y″ + xy′ − y = 0. Use the formula to find a second solution.

y'' + \dfrac{1}{x}y' - \dfrac{1}{x^2}y = 0

Divide by x² to get standard form, so p(x) = 1/x.

e^{-\int p\,dx} = e^{-\ln x} = \dfrac{1}{x}

Integrate p and use e^(−ln x) = 1/x.

v' = \dfrac{1/x}{x^2} = x^{-3}

Divide by y₁² = x².

v = \dfrac{x^{-2}}{-2} = -\dfrac{1}{2x^2}

Integrate with the power rule.

y_2 = x\left(-\dfrac{1}{2x^2}\right) = -\dfrac{1}{2x}

Multiply v by y₁.

y_2 = \dfrac{1}{x}

Drop the constant factor −1/2; any multiple of a solution is a solution.

Answer: y₂ = 1/x. Check: y′ = −1/x² and y″ = 2/x³, so x²(2/x³) + x(−1/x²) − 1/x = 2/x − 1/x − 1/x = 0.

Example 4: A formula integral that needs integration by parts

The function y₁ = e^x solves xy″ − (x + 1)y′ + y = 0 for positive x. Find a second solution.

y'' - \left(1 + \dfrac{1}{x}\right)y' + \dfrac{1}{x}y = 0

Divide by x, so p(x) = −(1 + 1/x).

e^{-\int p\,dx} = e^{x + \ln x} = xe^{x}

Integrate 1 + 1/x to get x + ln x, then use e^(ln x) = x.

v' = \dfrac{xe^{x}}{e^{2x}} = xe^{-x}

Divide by y₁² = e^(2x).

v = -xe^{-x} - e^{-x}

Integrate by parts with u = x, dv = e^(−x) dx (see integration by parts).

y_2 = e^{x}\left(-xe^{-x} - e^{-x}\right) = -x - 1

Multiply v by y₁.

y_2 = x + 1

Drop the factor −1.

Answer: y₂ = x + 1. Check: y′ = 1 and y″ = 0, so x(0) − (x + 1)(1) + (x + 1) = 0.

Example 5: A critically damped spring (word problem)

A spring’s position x (in meters) satisfies x″ + 6x′ + 9x = 0, with t in seconds, and one solution is x₁ = e^(−3t). Find a second solution, then the motion with x(0) = 0.5 and x′(0) = 0.

e^{-\int 6\,dt} = e^{-6t}

Standard form already holds, with p = 6.

v' = \dfrac{e^{-6t}}{\left(e^{-3t}\right)^2} = 1

Divide by x₁² = e^(−6t).

v = t,\quad x_2 = te^{-3t}

Integrate and multiply by x₁.

x = (C_1 + C_2t)e^{-3t}

Write the general solution.

x' = C_2e^{-3t} - 3(C_1 + C_2t)e^{-3t}

Differentiate with the product rule.

C_1 = 0.5

Substitute t = 0 into x.

C_2 - 3(0.5) = 0 \Rightarrow C_2 = 1.5

Substitute t = 0 into x′ and solve.

x = (0.5 + 1.5t)e^{-3t}

Write the motion.

x(1) = 2e^{-3} \approx 0.0996

For example, substitute t = 1.

Answer: x₂ = te^(−3t) and x(t) = (0.5 + 1.5t)e^(−3t). After 1 second the spring is about 0.1 m from rest position, and it returns without ever overshooting.

Common mistakes

MistakeExample of the errorFix
Reading p before dividing by the leading coefficientUsing p = −3x for x²y″ − 3xy′ + 4y = 0Divide by x² first: p = −3/x.
Forgetting the product ruleWriting (vy₁)′ = v′y₁′(vy₁)′ = v′y₁ + vy₁′.
Stopping at vGiving v = ln x as the second solutionThe second solution is y₂ = vy₁ = x² ln x.
Choosing a constant vTaking v = 5A constant v only gives 5y₁. Choose a nonconstant antiderivative.
Forgetting the minus sign in e^(−∫p dx)Using e^(∫p dx)The formula uses e^(−∫p dx), the same factor as Abel’s formula for the Wronskian.
Dividing by y₁ instead of y₁²v′ = e^(−∫p dx)/y₁Divide by y₁ squared.

Related ideas

Why the formula works

The Wronskian of y₁ and y₂ = vy₁ is W = y₁(v′y₁ + vy₁′) − vy₁y₁′ = v′y₁². By Abel’s formula from the Wronskian lesson, W = e^(−∫p dx) up to a constant. Setting them equal gives v′ = e^(−∫p dx)/y₁².

Repeated roots

Examples 1 and 5 show the pattern for a repeated root r of the characteristic equation: the second solution is xe^(rx). Repeated Roots of the Characteristic Equation uses this as a rule.

Equations with nonconstant coefficients

For constant coefficients, the characteristic equation is faster. Reduction of order earns its keep when the coefficients depend on x, as in Examples 2 to 4, and for Cauchy-Euler Equations.

Practice problems

  1. Given y₁ = e^(−x), find a second solution of y″ + 2y′ + y = 0.
  2. Given y₁ = e^(3x), find a second solution of y″ − 6y′ + 9y = 0.
  3. For positive x, y₁ = x solves x²y″ − xy′ + y = 0. Find a second solution.
  4. For positive x, y₁ = x³ solves x²y″ − 5xy′ + 9y = 0. Find a second solution.
  5. For positive x, y₁ = x solves x²y″ + 2xy′ − 2y = 0. Find a second solution.
  6. Given y₁ = e^x, use reduction of order to find a second solution of y″ − y = 0.
  7. On the interval from 0 to π, y₁ = sin x solves y″ + y = 0. Use reduction of order to find a second solution.
  8. Given y₁ = e^(x/2), find the general solution of 4y″ − 4y′ + y = 0.
  9. Show that substituting y = vy₁ into y″ + py′ + qy = 0 gives y₁v″ + (2y₁′ + py₁)v′ = 0.
  10. A spring satisfies x″ + 4x′ + 4x = 0 with x₁ = e^(−2t), x(0) = 1 and x′(0) = 3. Find x(t) and the time of the largest displacement.
Show answers

1.

v' = \dfrac{e^{-2x}}{e^{-2x}} = 1

p = 2, so e^(−∫p dx) = e^(−2x); divide by y₁² = e^(−2x).

v = x

Integrate.

Answer: y₂ = xe^(−x)

2.

v' = \dfrac{e^{6x}}{e^{6x}} = 1

p = −6, so e^(−∫p dx) = e^(6x); divide by y₁² = e^(6x).

v = x

Integrate.

Answer: y₂ = xe^(3x)

3.

p = -\dfrac{1}{x},\quad e^{-\int p\,dx} = e^{\ln x} = x

Divide by x² for standard form, then integrate p.

v' = \dfrac{x}{x^2} = \dfrac{1}{x}

Divide by y₁² = x².

v = \ln x

Integrate.

Answer: y₂ = x ln x

4.

p = -\dfrac{5}{x},\quad e^{-\int p\,dx} = e^{5\ln x} = x^5

Divide by x², integrate p, and use 5 ln x = ln x⁵.

v' = \dfrac{x^5}{x^6} = \dfrac{1}{x}

Divide by y₁² = x⁶.

v = \ln x

Integrate.

Answer: y₂ = x³ ln x

5.

p = \dfrac{2}{x},\quad e^{-\int p\,dx} = e^{-2\ln x} = x^{-2}

Divide by x² and integrate p.

v' = \dfrac{x^{-2}}{x^2} = x^{-4}

Divide by y₁² = x².

v = -\dfrac{x^{-3}}{3}

Integrate with the power rule.

y_2 = x\left(-\dfrac{1}{3x^3}\right) = -\dfrac{1}{3x^2}

Multiply by y₁.

Answer: y₂ = 1/x² (dropping the factor −1/3)

6.

v' = \dfrac{1}{e^{2x}} = e^{-2x}

p = 0, so e^(−∫p dx) = 1; divide by y₁² = e^(2x).

v = -\tfrac{1}{2}e^{-2x}

Integrate.

y_2 = e^{x}\left(-\tfrac{1}{2}e^{-2x}\right) = -\tfrac{1}{2}e^{-x}

Multiply by y₁.

Answer: y₂ = e^(−x) (dropping the factor −1/2)

7.

v' = \dfrac{1}{\sin^2 x} = \csc^2 x

p = 0; divide 1 by y₁² = sin² x.

v = -\cot x

Integrate: the derivative of cot x is −csc² x.

y_2 = \sin x(-\cot x) = -\cos x

Multiply by y₁ and use cot x = cos x/sin x.

Answer: y₂ = cos x (dropping the factor −1)

8.

y'' - y' + \tfrac{1}{4}y = 0

Divide by 4, so p = −1.

v' = \dfrac{e^{x}}{e^{x}} = 1

e^(−∫p dx) = e^x, and y₁² = e^x.

v = x

Integrate.

Answer: y = (C₁ + C₂x)e^(x/2)

9.

y' = v'y_1 + vy_1',\quad y'' = v''y_1 + 2v'y_1' + vy_1''

Differentiate y = vy₁ twice with the product rule.

y_1v'' + (2y_1' + py_1)v' + v(y_1'' + py_1' + qy_1) = 0

Substitute into y″ + py′ + qy = 0 and group by v″, v′ and v.

y_1v'' + (2y_1' + py_1)v' = 0

The last group is 0 because y₁ is a solution.

Answer: the v terms drop out, leaving y₁v″ + (2y₁′ + py₁)v′ = 0

10.

v' = \dfrac{e^{-4t}}{e^{-4t}} = 1 \Rightarrow x_2 = te^{-2t}

p = 4; divide e^(−4t) by x₁² = e^(−4t) and integrate.

x = (C_1 + C_2t)e^{-2t},\quad C_1 = 1

Write the general solution and substitute t = 0.

x' = C_2e^{-2t} - 2(C_1 + C_2t)e^{-2t}

Differentiate with the product rule.

C_2 - 2 = 3 \Rightarrow C_2 = 5

Substitute t = 0 and C₁ = 1 into x′.

x' = (5 - 2 - 10t)e^{-2t} = (3 - 10t)e^{-2t}

Substitute the constants into x′ and simplify.

t = 0.3,\quad x(0.3) = 2.5e^{-0.6} \approx 1.372

Set x′ = 0, then evaluate x at t = 0.3.

Answer: x(t) = (1 + 5t)e^(−2t); the largest displacement is about 1.372 at t = 0.3 s

Reduction of order FAQ

What is reduction of order?

It is a method for finding a second solution of a second-order linear homogeneous differential equation when one solution y₁ is known. You set y₂ = v(x)y₁ and solve a first-order equation for v′.

What is the reduction of order formula?

For y″ + p(x)y′ + q(x)y = 0, the second solution is y₂ = y₁∫(e^(−∫p dx)/y₁²) dx. The equation must be in standard form first.

Why is it called reduction of order?

After substituting y = vy₁, the terms with v cancel and only v′ and v″ remain. Letting w = v′ turns the second-order equation into a first-order one.

When do you use reduction of order?

Use it when you know one solution but not a second, especially for equations with variable coefficients and for repeated roots of a characteristic equation.

Summary

  • Reduction of order finds a second solution y₂ = vy₁ from a known solution y₁.
  • Substituting y = vy₁ cancels the v terms and leaves a first-order equation for w = v′.
  • Shortcut: v′ = e^(−∫p dx)/y₁², with the equation in standard form.
  • Drop constants of integration and constant factors; any nonconstant v works.
  • The general solution is y = C₁y₁ + C₂y₂.

Previous lesson: The Wronskian

Next lesson: Complex Roots of the Characteristic Equation

Back to the Differential Equations Index