Second order equations reducible by substitution

Second order equations reducible by substitution

Spec 5.2 · Edexcel International A Level Further Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • A given substitution gives a constant-coefficient equation that you already know how to solve.
  • With x=etx=e^t: xdydx=y˙x\frac{dy}{dx}=\dot y and x2d2ydx2=y¨−y˙x^2\frac{d^2y}{dx^2}=\ddot y-\dot y.
  • An Euler equation ax2y′′+bxy′+cy=f(x)ax^2y''+bxy'+cy=f(x) becomes ay¨+(b−a)y˙+cy=f(et)a\ddot y+(b-a)\dot y+cy=f(e^t).
  • With y=zf(x)y=zf(x), use the product rule and cancel the common factor.
  • Convert conditions carefully, and always return to xx and yy.

The idea of a substitution

The equation has non-constant coefficients, so a substitution changes it into one with constant coefficients.

A substitution replaces xx by a new variable tt (change of independent variable) or yy by a new function zz (change of dependent variable) so that the new equation has constant coefficients.

In this specification the substitution is always given. Differentiate first, substitute second.

  1. 1

    Differentiate

    Express each derivative of yy in the new variable (chain or product rule)

  2. 2

    Substitute

    Simplify to a constant-coefficient equation

  3. 3

    Solve

    Use the standard method

  4. 4

    Convert back

    Return to the original variables and apply any conditions

The steps are always the same

What is the purpose of the substitution?

Changing the independent variable

With x=etx=e^t, xdydx=y˙x\frac{dy}{dx}=\dot y and x2d2ydx2=y¨−y˙x^2\frac{d^2y}{dx^2}=\ddot y-\dot y, where a dot is ddt\frac{d}{dt}.

For x>0x>0 the standard substitution is x=etx=e^t (so t=ln⁡xt=\ln x). By the chain rule, since dtdx=1x=e−t\frac{dt}{dx}=\frac1x=e^{-t}, dydx=e−ty˙\frac{dy}{dx}=e^{-t}\dot y.

For the second derivative, use the product rule and then the chain rule again: d2ydx2=e−2t(y¨−y˙)\frac{d^2y}{dx^2}=e^{-2t}(\ddot y-\dot y). These turn an Euler equation ax2y′′+bxy′+cy=f(x)ax^2y''+bxy'+cy=f(x) into one with constant coefficients.

12345678910−2−1123xy(1, 0)(e, 1)t = ln x
The substitution x = eᵗ means t = ln x: this curve converts x into t
  • xdydxx\dfrac{dy}{dx}y˙\dot y
  • x2d2ydx2x^2\dfrac{d^2y}{dx^2}y¨−y˙\ddot y-\dot y
  • ax2y′′+bxy′+cyax^2y''+bxy'+cyay¨+(b−a)y˙+cya\ddot y+(b-a)\dot y+cy

Under x=etx=e^t, what does x2d2ydx2x^2\frac{d^2y}{dx^2} become?

Changing the dependent variable

Differentiate with the product rule and substitute; the common factor can be cancelled.

Sometimes the substitution replaces yy by a product, for example y=ze−x2y=ze^{-x^2}. Differentiate with the product rule and substitute. Every term contains the factor e−x2e^{-x^2}, which is never zero, so it can be cancelled.

Remember ddxe−x2=−2xe−x2\frac{d}{dx}e^{-x^2}=-2xe^{-x^2}, not −x2e−x2-x^2e^{-x^2}.

−3−2−11230.20.40.60.811.2xy(0, 1)y = e^(−x²)
The factor e^(−x²) is never zero, so it can be cancelled from every term
  • y=ze−x2y=ze^{-x^2}y′=(z′−2xz)e−x2y'=(z'-2xz)e^{-x^2}

Worked example

Use y=ze−x2y=ze^{-x^2} to simplify y′′+4xy′+(4x2+2)yy''+4xy'+(4x^2+2)y.

With y=ze−x2y=ze^{-x^2}, what is y′y'?

Solving the new equation

Use the auxiliary equation and a trial PI, then substitute back using t=ln⁡xt=\ln x.

Solve ay¨+by˙+cy=f(t)a\ddot y+b\dot y+c y=f(t) in the usual way: the auxiliary equation gives the complementary function, a particular integral matches f(t)f(t), and the general solution is their sum. If the PI form already appears in the CF, multiply by tt (or t2t^2 for a repeated root).

Finally substitute back using t=ln⁡xt=\ln x and et=xe^t=x: y=(A+Bt)e−2ty=(A+Bt)e^{-2t} becomes y=A+Bln⁡xx2y=\frac{A+B\ln x}{x^2}.

  1. 1

    Auxiliary equation

    find the roots

  2. 2

    CF

    real, repeated or complex

  3. 3

    PI

    match f(t)f(t); multiply by tt if needed

  4. 4

    Back-substitute

    t=ln⁡xt=\ln x and et=xe^t=x

Solving the transformed equation

y=(A+Bt)e−2ty=(A+Bt)e^{-2t} with t=ln⁡xt=\ln x. What is yy in terms of xx?

Conditions and converting back

Convert conditions carefully: at x=1x=1, t=0t=0 and y˙=dydx\dot y=\frac{dy}{dx} since y˙=xdydx\dot y=x\frac{dy}{dx}.

Conditions are usually given in terms of xx. A value y(x0)y(x_0) becomes yy at t0=ln⁡x0t_0=\ln x_0 (at x=1x=1, t=0t=0). A condition on dydx\frac{dy}{dx} becomes one on y˙\dot y through y˙=xdydx\dot y=x\frac{dy}{dx}; at x=1x=1, y˙=dydx\dot y=\frac{dy}{dx}.

Check the answer by differentiating it, or by testing the original equation at one value of xx.

0.511.522.533.544.555.56−2−1123xy(1, 1)y = (1 + 2 ln x)/x²
The solution y = (1 + 2 ln x)/x² has a maximum at (1, 1), where y′ = 0.

Worked example

Solve x2y′′+5xy′+4y=0x^2y''+5xy'+4y=0 with y(1)=1y(1)=1 and y′(1)=0y'(1)=0, using x=etx=e^t.

At x=1x=1, dydx=0\frac{dy}{dx}=0. What is y˙\dot y at t=0t=0?

Try an exam question

Use the substitution x=etx=e^t to find the general solution of x2d2ydx2+3xdydx+y=0x^2\frac{d^2y}{dx^2}+3x\frac{dy}{dx}+y=0 for x>0x>0.

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Which expression is equal to dydx\frac{dy}{dx}?

1 mark, marked against the mark scheme

Revision notes