Skip to main content

Exercises 10.6 Mixed Review: Choosing Your Method

The last two chapters more than doubled your toolkit. On top of the first-order methods—direct integration, separation of variables, and the integrating factor—you can now solve any LHCC equation with the characteristic equation, and any LNCC equation whose forcing function fits the Method of Undetermined Coefficients. With more tools comes a new burden: an exam problem won’t tell you which one to pull out. This review practices making that call.
The decision comes down to three questions: What is the order? Is the equation linear with constant coefficients? What does the forcing function look like? The table below extends the “which method?” flowchart from the first-order mixed review to cover everything you know so far.
If the equation is … then …
first order
walk the first-order flowchart: direct integration \(\rightarrow\) separation of variables \(\rightarrow\) integrating factor.
linear, constant coefficients, forcing \(= 0\) (LHCC, any order)
solve the characteristic equation and build the general solution from its roots.
linear, constant coefficients, forcing \(f(x)\) built from polynomials, exponentials, sines, and cosines (LNCC)
build \(y = y_h + y_p\text{:}\) characteristic equation for \(y_h\text{,}\) then undetermined coefficients for \(y_p\text{.}\)
anything else: variable coefficients, a nonlinear term, or a forcing function outside that family
none of our formula methods apply yet—reach for qualitative or numerical tools, and watch for the Laplace transform in upcoming chapters.
One habit worth keeping: a first-order linear equation with constant coefficients, like \(y' + 3y = 6\text{,}\) belongs to both toolkits. The integrating factor method and \(y_h + y_p\) reasoning give the same answer—being able to see it both ways is a good sign you understand each method.

💡 Which Method? Quiz.

For each differential equation, choose the best plan of attack. Do not solve any of the equations—practice making the call the way you would on an exam.

1.

(a) Pick the Method.

\begin{equation*} y'' - 5y' + 6y = 0 \end{equation*}
  • Characteristic equation—this is an LHCC equation.
  • Correct! Linear, constant coefficients, zero forcing. The characteristic equation \(r^2 - 5r + 6 = 0\) hands you the general solution.
  • Undetermined coefficients, to find \(y_p\text{.}\)
  • The forcing function is \(0\text{,}\) so there is no \(y_p\) to find—the homogeneous solution is the whole story.
  • Integrating factor.
  • The integrating factor method is a first-order tool; this equation is second order.
  • Separation of variables.
  • Separation of variables is a first-order tool; this equation is second order.

(b) Pick the Method.

\begin{equation*} y'' - 5y' + 6y = e^{x} \end{equation*}
  • Characteristic equation for \(y_h\text{,}\) then undetermined coefficients with guess \(y_p = Ae^{x}\text{.}\)
  • Correct! This is an LNCC equation whose forcing function \(e^x\) is in the undetermined-coefficients family, and \(e^x\) does not overlap \(y_h = C_1 e^{2x} + C_2 e^{3x}\text{.}\)
  • Characteristic equation alone—the general solution comes entirely from its roots.
  • The roots only produce \(y_h\text{.}\) With a nonzero forcing function you also need a particular solution: \(y = y_h + y_p\text{.}\)
  • Integrating factor with \(\mu = e^{\int -5\, dx}\text{.}\)
  • The integrating factor method is a first-order tool; this equation is second order.
  • None of our methods apply.
  • Check the table: linear, constant coefficients, and \(e^x\) is squarely in the undetermined-coefficients family.

(c) Two Toolkits, One Equation.

Which statement is true about the equation below?
\begin{equation*} y' + 3y = 6 \end{equation*}
  • Both the integrating factor method and \(y_h + y_p\) reasoning (undetermined coefficients) apply.
  • Correct! It’s first-order linear, so \(\mu = e^{3x}\) works—and it’s also an LNCC equation with constant forcing, so \(y_h = Ce^{-3x}\) plus the constant guess \(y_p = A\) works too. Both give \(y = 2 + Ce^{-3x}\text{.}\)
  • Only the integrating factor method applies, because the equation is first order.
  • First order doesn’t disqualify the new machinery: this is a linear constant-coefficient equation with constant forcing, so \(y_h + y_p\) reasoning applies just as well.
  • Only undetermined coefficients applies, because the forcing function is a constant.
  • Undetermined coefficients works here, but so does the integrating factor method—the equation is first-order linear.
  • Neither method applies.
  • Quite the opposite—this equation belongs to both toolkits. Check the linearity and coefficient tests again.

(d) Know the Limits of the Guess.

\begin{equation*} y'' + 4y = \tan x \end{equation*}
  • The characteristic equation gives \(y_h\text{,}\) but none of our methods produce \(y_p\text{.}\)
  • Correct! The forcing function \(\tan x\) is not built from polynomials, exponentials, sines, and cosines, so undetermined coefficients has no form to guess. Equations like this need a method beyond our current toolkit (its name, for the curious: variation of parameters).
  • Undetermined coefficients with guess \(y_p = A\tan x + B\sec x\text{.}\)
  • Tempting, but derivatives of \(\tan x\) generate an endless stream of new function types—the guess never closes. Undetermined coefficients only works for forcing functions whose derivative families are finite: polynomials, exponentials, sines, and cosines.
  • Separation of variables.
  • Separation of variables is a first-order tool; this equation is second order.
  • No method applies, not even for \(y_h\text{.}\)
  • The homogeneous side is a perfectly good LHCC equation—\(y_h = C_1\cos 2x + C_2 \sin 2x\text{.}\) It’s only the particular solution that’s out of reach.

(e) Check the Fine Print.

\begin{equation*} y'' + y^2 = 0 \end{equation*}
  • None of our methods apply—the equation is nonlinear.
  • Correct! The \(y^2\) term makes this nonlinear, and the characteristic equation shortcut is built on linearity. Substituting \(y = e^{rx}\) here would give \(r^2 e^{rx} + e^{2rx} = 0\text{,}\) which no constant \(r\) can satisfy.
  • Characteristic equation: \(r^2 + r^2 = 0\text{.}\)
  • Careful—\(y^2\) is the square of the function, not a second derivative. The equation is nonlinear, and the characteristic equation only applies to linear constant-coefficient equations.
  • Undetermined coefficients.
  • Undetermined coefficients builds particular solutions for linear equations with forcing; here the equation is nonlinear with no forcing at all.
  • Direct integration, twice.
  • Integrating twice would need the right side free of \(y\text{;}\) the \(y^2\) term blocks that.

(f) Check the Fine Print.

\begin{equation*} x\, y'' + y' = 0 \end{equation*}
  • The characteristic-equation method does not apply—the coefficients are not constant.
  • Correct! The leading coefficient is \(x\text{,}\) not a constant, so the exponential-guess machinery breaks down. (If you’re curious: the substitution \(v = y'\) turns this into the separable first-order equation \(x v' + v = 0\text{,}\) leading to \(y = C\ln x + D\) for \(x \gt 0\text{.}\))
  • Characteristic equation: \(xr^2 + r = 0\text{.}\)
  • A characteristic equation must have constant coefficients—an \(x\) inside means the method was never valid to begin with.
  • Undetermined coefficients.
  • Undetermined coefficients assumes constant coefficients and a nonzero forcing function; this equation has neither.
  • Integrating factor, directly on \(y''\text{.}\)
  • The integrating factor method as we know it handles first-order linear equations. (There is a first-order equation hiding here, though—what happens if you rename \(y'\) as a new unknown?)

(g) Any Order Welcome.

\begin{equation*} y''' - y' = 0 \end{equation*}
  • Characteristic equation—it works for LHCC equations of any order.
  • Correct! \(r^3 - r = r(r-1)(r+1) = 0\) gives roots \(0, 1, -1\text{,}\) so \(y = C_1 + C_2 e^{x} + C_3 e^{-x}\text{.}\)
  • No method applies—the characteristic equation only handles second-order equations.
  • The characteristic equation scales to any order: an order-\(n\) LHCC equation gives a degree-\(n\) polynomial with \(n\) roots and \(n\) terms in the general solution.
  • Direct integration, three times.
  • After one integration you’d have \(y'' - y = C\text{,}\) which still involves \(y\)—direct integration can’t finish the job.
  • Undetermined coefficients.
  • The forcing function is \(0\text{,}\) so there is no particular solution to build—the characteristic equation alone gives the general solution.

✍🏻 Choose, Then Solve.

Find the general solution of each equation (or the particular solution where initial conditions are given). No method is announced—begin each solution by naming the method you chose and the feature of the equation that told you it applies.

2.

\(y'' - y' - 6y = 0\)
Answer.
Characteristic equation; \(y = C_1 e^{3x} + C_2 e^{-2x}\text{.}\)

3.

\(y'' + 4y' + 13y = 0\)
Answer.
Characteristic equation (complex roots); \(y = e^{-2x}\left(C_1 \cos 3x + C_2 \sin 3x\right)\text{.}\)

4.

\(y'' + 6y' + 9y = 0\)
Answer.
Characteristic equation (repeated root); \(y = \left(C_1 + C_2 x\right)e^{-3x}\text{.}\)

5.

\(y'' - 3y' + 2y = 4x\)
Answer.
Undetermined coefficients; \(y = C_1 e^{x} + C_2 e^{2x} + 2x + 3\text{.}\)

6.

\(y'' - 4y = 8e^{2x}\)
Answer.
Undetermined coefficients with an overlap adjustment; \(y = C_1 e^{2x} + C_2 e^{-2x} + 2xe^{2x}\text{.}\)

7.

\(y' - 2y = 4, \quad y(0) = 5\)
Answer.
Integrating factor (or first-order \(y_h + y_p\)); \(y = 7e^{2x} - 2\text{.}\)

8.

\(y'' + y = 5\sin 2x\)
Answer.
Undetermined coefficients; \(y = C_1 \cos x + C_2 \sin x - \dfrac{5}{3}\sin 2x\text{.}\)
You have attempted of activities on this page.