Higher-order linear equations · Definition

Variation of parameters

differential equationshigher order

Before you start you need to know

It solves the case that undetermined coefficients cannot reach: any continuous \(g\), in exchange for knowing two homogeneous solutions first.

Definition

A method that builds a particular solution \(y_p\) of \(y''+py'+qy=g(t)\) from two homogeneous solutions \(y_1,y_2\) and their Wronskian \(W\), by letting the constants vary.

\[y_p=-y_1\!\int\!\frac{y_2\,g}{W}\,dt+y_2\!\int\!\frac{y_1\,g}{W}\,dt\]

If \(y_1,y_2\) form a fundamental set of \(y''+py'+qy=0\), the idea is to replace the constants in \(c_1y_1+c_2y_2\) by functions: \(y_p=u_1y_1+u_2y_2\). Since there are two unknown functions, one also imposes \(u_1'y_1+u_2'y_2=0\), and substituting into the equation gives \(u_1'y_1'+u_2'y_2'=g\). This is a \(2\times2\) system whose determinant is the Wronskian \(W\neq0\); by Cramer, \(u_1'=-y_2g/W\) and \(u_2'=y_1g/W\), and integrating yields the formula above.

The equation must be in standard form, with coefficient \(1\) on \(y''\): in \(2y''+2y=8t\) divide first, so \(g=4t\), not \(8t\). The constants of integration are omitted, because they would only add homogeneous solutions. With \(y''+y=\sec t\), \(y_1=\cos t\), \(y_2=\sin t\) and \(W=1\), one gets \(u_1=\ln\cos t\) and \(u_2=t\), so \(y_p=\cos t\ln\cos t+t\sin t\); undetermined coefficients would not have worked here.

The costs are that \(y_1,y_2\) are needed beforehand (for example from the characteristic equation) and the integrals can be hard; when \(g\) is a polynomial, exponential or trigonometric function, undetermined coefficients is usually faster. The general solution is always \(y=c_1y_1+c_2y_2+y_p\). The method also extends to order \(n\) with an \(n\times n\) system whose determinant is the Wronskian of \(n\) functions.

Think about it

In \(y''+y=\tan t\), why does guessing \(y_p=A\tan t\) not work?

Because \(\tan t\) is not of a type closed under differentiation: its derivatives generate \(\sec^2t\), \(\sec^2t\tan t\), etc., so there is no finite set of coefficients to adjust. Variation of parameters is used instead.

Exercises 10

References

  • William E. Boyce, Richard C. DiPrima, Douglas B. Meade, Elementary Differential Equations and Boundary Value Problems, 12th edition, Wiley, 2021. §3.6
  • Dennis G. Zill, A First Course in Differential Equations with Modeling Applications, 11th edition, Cengage, 2018. §4.6