Ordinary Differential Equations | The Ultimate Crash Course for PLEM Majors

Math Monthly · Lesson Library

Volume 10 · Ordinary Differential Equations

Ordinary Differential Equations Crash Course

The Ultimate Crash Course for PLEM Majors

Study ordinary differential equations with Author Jonathan David. Work through first- and higher-order equations, learn how the form of an equation guides your choice of method, and verify that your solution satisfies the original equation and any given initial conditions.

Choose a lesson below.
Select its title to expand the lesson. Select it again to close it.

What You Need to Know from a Differential Equations Book Before Starting the Course

Watch on YouTube

AI Summary of the Video

Author Jonathan David compares differential equations textbooks and shows how to recognize an equation’s form, find the matching method in the assigned text, and set up a solution.

  • Classify the equation first. Identify its order, whether it is linear, and whether its coefficients are constant or variable. For a linear equation, check whether a forcing term is present. An autonomous equation has no explicit dependence on the independent variable.
  • See what each method requires. For reduction of order, the example starts with a known solution and puts the equation in standard form. For a Cauchy–Euler equation, the powers of the independent variable match the derivative orders, suggesting a trial solution of the form y = xk.
  • Review the skills you’ll use in the calculation. The walkthrough highlights algebraic simplification, derivatives, antiderivatives, and trigonometric identities. Check a proposed solution by substituting it and its derivatives back into the original equation, then verify any initial conditions.
  • Use lecture examples to study systems. If your course combines linear algebra and differential equations, compare how worked examples handle different eigenvalue cases, including complex eigenvalues and cases where the matrix lacks a full set of eigenvectors.

“By identifying the differential equation, you can then pinpoint what solution method to use.”

— Author Jonathan David
6.1 Solutions About Ordinary Points

Watch the original lesson on YouTube

Differential Equations: Series solution about ordinary point

[24] Zill, D. G. (2009). A first course in differential equations with modeling applications (9th ed.). Brooks/Cole, Cengage Learning.

QUESTION 6.1.17

Find two power series solutions of the given differential equation about the ordinary point $x=0$.

$$y”-xy=0$$

ANSWER 411 ANS-8

$$y_1(x)=c_0\left[1+\frac{1}{3\cdot2}x^3+\frac{1}{6\cdot5\cdot3\cdot2}x^6+\frac{1}{9\cdot8\cdot6\cdot5\cdot3\cdot2}x^9+\cdots\right]$$ $$y_2(x)=c_1\left[x+\frac{1}{4\cdot3}x^4+\frac{1}{7\cdot6\cdot4\cdot3}x^7+\frac{1}{10\cdot9\cdot7\cdot6\cdot4\cdot3}x^{10}+\cdots\right]$$

$\mathcal{F}_1$ [Formulae]

$$a_2(x)y”+a_1(x)y’+a_0(x)y=0\qquad (5),$$ $$y”+\frac{a_1(x)}{a_2(x)}y’+\frac{a_0(x)}{a_2(x)}y =y”+P(x)y’+Q(x)y=0\qquad (6).$$
DEFINITION 6.1.1   Ordinary and Singular Points

A point $x_0$ is said to be an ordinary point of the differential equation (5) if both $P(x)$ and $Q(x)$ in the standard form (6) are analytic at $x_0$. A point that is not an ordinary point is said to be a singular point of the equation.

Image from Pg. 223 for Educational Purposes
THEOREM 6.1.1   Existence of Power Series Solutions

If $x=x_0$ is an ordinary point of the differential equation (5), we can always find two linearly independent solutions in the form of a power series centered at $x_0$, that is, $y=\sum_{n=0}^{\infty}c_n(x-x_0)^n$. A series solution converges at least on some interval defined by $|x-x_0|<R$, where $R$ is the distance from $x_0$ to the closest singular point.

Image from Pg. 223 for Educational Purposes

Power Series Centered at $x=a$

$$y=\sum_{n=0}^{\infty}c_n(x-a)^n$$

$\mathcal{F}_2$ [Formulate] Start with the series and go from there.

$$y=\sum_{n=0}^{\infty}c_n(x-a)^n =\sum_{n=0}^{\infty}c_n(x-0)^n =\sum_{n=0}^{\infty}c_nx^n \Rightarrow y=\sum_{n=0}^{\infty}c_nx^n\Rightarrow$$ $$\begin{aligned} y’&=\sum_{n=0}^{\infty}c_nnx^{n-1} =c_0(0)x^{0-1}+\sum_{n=1}^{\infty}c_nnx^{n-1} =\sum_{n=1}^{\infty}c_nnx^{n-1}\\ &=\sum_{n=1-1}^{\infty}c_{n+1}(n+1)x^{n-1+1} =\sum_{n=0}^{\infty}c_{n+1}(n+1)x^n\Rightarrow \end{aligned}$$ $$y”=\sum_{n=0}^{\infty}c_nn(n-1)x^{n-2} =\sum_{n=2}^{\infty}c_nn(n-1)x^{n-2} =\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n.$$

Now, plug the series into the provided ODE (2nd order-linear-homogeneous)

MISTAKE STARTS HERE I put $y’$ in for $y$.

$$y”-xy=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n\right] -x\left[\sum_{n=0}^{\infty}c_{n+1}(n+1)x^n\right] \quad\leftarrow\text{ MISTAKE}$$

[I put $y’$ in for $y$. The math is correct but $y\ne y’$. We had to start over at the end]

$$\begin{aligned} &=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n\right] -\left[\sum_{n=0}^{\infty}c_{n+1}(n+1)x^{n+1}\right]\\ &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=0}^{\infty}c_{n+1}(n+1)x^{n+1} \end{aligned}$$

Before going forward, we need to understand that the lower limit $n$ and the variable $x$ need match the starting integer and the exponent. I could add 1 to the first term but then I would have to subtract 1 from the subscript but we are bound by $\mathbb{Z}^{+}\cup\{0\}$, so we cannot go negative. That is not entirely true but we will have issues here if we do that.

$$\begin{aligned} &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=0+1}^{\infty}c_{(n-1)+1}\bigl((n-1)+1\bigr)x^{n+1-1}\\ &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=1}^{\infty}c_nnx^n \end{aligned}$$

Now, the exponents match but the limits do not. Simple solution, take the series out 1 entry.

$$\begin{aligned} &=c_{0+2}(0+2)(0+1)x^0+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=c_2(2)(1)x^{1-1}+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=2c_2x^1x^{-1}+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=2c_2(x)\left(\frac{1}{x}\right)+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=2c_2\frac{x}{x}+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=2c_2(1)+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n\\ &=2c_2+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n-\sum_{n=1}^{\infty}c_nnx^n \end{aligned}$$

Now, we can use summation rules to un-distribute the summation and polynomial.

$$\begin{aligned} &=2c_2+\sum_{n=1}^{\infty}\left[c_{n+2}(n+2)(n+1)x^n-c_nnx^n\right]\\ &=2c_2+\sum_{n=1}^{\infty}\left[c_{n+2}(n+2)(n+1)-c_nn\right]x^n=0. \end{aligned}$$

Now, in order for this to be equal to zero, $2c_2=0$ and $c_{n+2}(n+2)(n+1)-c_nn=0$.

This provides our recursive relation. Clearly, $c_2=0$.

$$\begin{aligned} c_{n+2}(n+2)(n+1)-c_nn=0 &\Rightarrow c_{n+2}(n+2)(n+1)=c_nn\\ &\Rightarrow \frac{c_{n+2}}{c_n}=\frac{n}{(n+2)(n+1)} \Rightarrow c_{n+2}=\frac{c_nn}{(n+2)(n+1)}. \end{aligned}$$

We now let $c_0=0$, $c_1=1$ and $c_0=1$, $c_1=0$ and find our constants.

NOTE The question does not specify to go out to four terms, but we will match the answer.

$$\begin{aligned} c_0=0,\quad n=0\qquad &c_{0+2}=\frac{(0)(0)}{(0+2)(0+1)}=0=c_2\\ c_1=1,\quad n=1\qquad &c_{1+2}=\frac{c_1(1)}{(1+2)(1+1)} =\frac{(1)(1)}{(1+2)(1+1)}=\frac1{3\cdot2}=c_3\\ c_2=0,\quad n=2\qquad &c_4=\frac{(0)n}{(n+2)(n+1)}=0\\ c_3=\frac1{3\cdot2},\quad n=3\qquad &c_5=\frac{\left(\frac1{3\cdot2}\right)(3)}{(3+2)(3+1)} =\frac3{5\cdot4\cdot3\cdot2}=\frac1{5\cdot4\cdot2}=c_5\\ c_4=0,\quad n=4\qquad &c_6=0\\ c_5=\frac1{5\cdot4\cdot2},\quad n=5\qquad &c_7=\frac{\left(\frac1{5\cdot4\cdot2}\right)5}{(5+2)(5+1)} =\frac1{7\cdot6\cdot4\cdot2} \end{aligned}$$

ABOVE ↑ IS NOT MATCHING

Going back to the book’s method, the technique I chose to use is not the book’s technique and is thus not matching. That doesn’t mean that it is wrong as this is an infinite approximation essentially but since it does not match, it is definitely wrong! The book chooses to start with the index and get everything in terms of $c_0$ and $c_1$.

Going backward, we start with $c_2=0$ and

NOTE The author also chose a slightly different method for simplifying the ODE with the power series in terms of $k$ after a substitution.

$$c_{n+2}=\frac{c_nn}{(n+2)(n+1)}.$$ $$\begin{aligned} n=0\qquad &c_2=\frac{c_nn}{(n+2)(n+1)}=0\\ n=1\qquad &c_3=\frac{c_1(1)}{(1+2)(1+1)}=\frac{c_1}{3\cdot2}\\ n=2\qquad &c_4=\frac{c_2n}{(n+2)(n+1)}=0\\ n=3\qquad &c_5=\frac{c_3(3)}{(3+2)(3+1)} =\frac{\left(\frac{c_1}{3\cdot2}\right)(3)}{5\cdot4} =\frac{c_1}{5\cdot4\cdot2} \end{aligned}$$

NOTICE that we are getting the same answer with two different methods. Neither of which are matching the books answer.

If I put the values into $y$, we will see

$$y_1(x)=\sum_{n=0}^{\infty}c_nx^n=c_0x^0+c_1x^1+c_2x^2+c_3x^3+c_4x^4+c_5x^5+\cdots$$

Let $c_0=0$

$$\begin{aligned} &=(0)x^0+c_1x^1+(0)x^2+\frac{c_1}{3\cdot2}x^3+(0)x^4+\frac{c_1}{5\cdot4\cdot2}x^5+\cdots\\ &=c_1x+\frac{c_1}{3\cdot2}x^3+\frac{c_1}{5\cdot4\cdot2}x^5+\cdots =c_1\left[x+\frac1{3\cdot2}x^3+\frac1{5\cdot4\cdot2}x^5+\cdots\right] \end{aligned}$$

Write now, we have a conflicting situation with the book’s answer. It is similar but not the same.

The answer is looking very similar, but it is not matching the book’s answer.

[AI CHALLENGE] We ask AI to provide the answer but not to provide any solution. We are not looking for any assistance in writing the algorithm. We are investigating whether or not the book’s answer is correct. I will ask ChatGPT to give me an answer to four terms for this series solution. This prevents cheating and plagiarism while assisting with the potential typos. Before I do that, I want to review the textbook literature just to make sure. The author uses a substitution but that shouldn’t make a difference.

ChatGPT is in concurrence with the textbook’s answer. This leads me to believe that I made a typo. Since it is late at night, I must sleep. I will post this now at MathMonthly.com for the research students to type in LaTeX code via MathMonthly.com to find the typo.

$$y_1(x)=1+\frac{x^3}{6}+\frac{x^6}{180}+\frac{x^9}{12960}+\cdots$$ $$y_2(x)=x+\frac{x^4}{12}+\frac{x^7}{504}+\frac{x^{10}}{45360}+\cdots$$

Since ChatGPT agrees with the book, it is safe to say we likely made a mistake. I will come to back to this tomorrow and let the Math Monthly students look it over.

OKAY!!! I just realized my mistake. Had I had any students with a keen eye, they would have seen this right away.

THE MISTAKE: I used $y’$ in place of $y$. We start over now,

$$y”-xy=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n\right] -x\left[\sum_{n=0}^{\infty}c_nx^n\right]$$

MISTAKE I forgot to distribute the $x$.

$$=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)-c_n\right]x^n=0$$ $$\Rightarrow c_{n+2}(n+2)(n+1)-c_n=0 \Rightarrow c_{n+2}=\frac{c_n}{(n+2)(n+1)}$$ $$\begin{aligned} n=0\qquad &c_2=\frac{c_0}{(0+2)(0+1)}=\frac{c_0}{2}\\ n=1\qquad &c_3=\frac{c_1}{3\cdot2}\\ n=2\qquad &c_4=\frac{c_2}{4\cdot3}\\ n=3\qquad &c_5=\frac{c_3}{5\cdot4} =\frac{\frac{c_1}{3\cdot2}}{5\cdot4} =\frac{c_1}{5\cdot4\cdot3\cdot2} \end{aligned}$$

It is late, I have to come back in the morning with some coffee and find the error.

FIXING THE PROBLEM OF STUPIDITY!!!

$$\begin{aligned} y”-xy &=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n\right] -x\left[\sum_{n=0}^{\infty}c_nx^n\right]\\ &=\left[\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n\right] -\left[\sum_{n=0}^{\infty}c_nx^{n+1}\right]\\ &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=0}^{\infty}c_nx^{n+1}\\ &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=0+1}^{\infty}c_{(n-1)}x^{(n-1)+1}\\ &=\sum_{n=0}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=1}^{\infty}c_{(n-1)}x^n\\ &=c_{0+2}(0+2)(0+1)x^0+ \sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=1}^{\infty}c_{(n-1)}x^n\\ &=2c_2+\sum_{n=1}^{\infty}c_{n+2}(n+2)(n+1)x^n -\sum_{n=1}^{\infty}c_{(n-1)}x^n\\ &=2c_2+\sum_{n=1}^{\infty}\left[c_{n+2}(n+2)(n+1)-c_{n-1}\right]x^n=0 \end{aligned}$$ $$\Leftrightarrow c_2=0\text{ and }c_{n+2}(n+2)(n+1)-c_{n-1}=0.$$

MISTAKE AGAIN $n+2$ … Typos…

$$\therefore c_2=0,\qquad c_{n+1}=\frac{c_{n-1}}{(n+2)(n+1)}.$$

We start at $n=1$ because we cannot have a negative index.

$$\begin{aligned} n=1\qquad &c_{1+1}=\frac{c_{1-1}}{(1+2)(1+1)} =\frac{c_0}{3\cdot2}=0\Rightarrow c_0=0\\ n=2\qquad &c_{2+1}=c_3=\frac{c_1}{(2+2)(2+1)}=\frac{c_1}{4\cdot3}\\ n=3\qquad &c_4=\frac{c_2}{(3+2)(3+1)}=0\\ n=4\qquad &c_5=\frac{c_3}{(4+2)(4+1)} =\frac{\frac{c_1}{4\cdot3}}{6\cdot5} =\frac{c_1}{6\cdot5\cdot4\cdot3}\\ n=5\qquad &c_6=0\\ n=6\qquad &c_7=\frac{c_5}{8\cdot7} =\frac{\frac{c_1}{6\cdot5\cdot4\cdot3}}{8\cdot7} \end{aligned}$$

CORRECTION I was up since 4 AM and it is now 10 PM and I have to deal with stupid illiterates interrupting me in chat! We have completed the question now.

$$\therefore c_2=0,\qquad c_{n+2}=\frac{c_{n-1}}{(n+2)(n+1)}.$$ $$\begin{aligned} n=1\qquad &c_{1+2}=c_3=\frac{c_{1-1}}{(1+2)(1+1)}=\frac{c_0}{3\cdot2}\\ n=2\qquad &c_4=\frac{c_{2-1}}{(2+2)(2+1)}=\frac{c_1}{4\cdot3}\\ n=3\qquad &c_5=\frac{c_2}{(3+2)(3+1)}=0\\ n=4\qquad &c_6=\frac{c_3}{(4+2)(4+1)} =\frac{\frac{c_0}{3\cdot2}}{6\cdot5} =\frac{c_0}{6\cdot5\cdot3\cdot2}\\ n=5\qquad &c_7=\frac{c_{5-1}}{(5+2)(5+1)} =\frac{\frac{c_1}{4\cdot3}}{7\cdot6} =\frac{c_1}{7\cdot6\cdot4\cdot3}\\ n=6\qquad &c_8=0\\ n=7\qquad &c_9=\frac{c_6}{(9)(8)} =\frac{c_0}{9\cdot8\cdot6\cdot5\cdot3\cdot2} \end{aligned}$$

$\mathcal{F}_3$ [Finalize]

For this question, we see now that we have found the pattern. One simply needs to isolate the $c_0$ and $c_1$ to finalize the following answer.

Just rearrange and un-distribute the constants.

$$\begin{aligned} y(x)&=y_1(x)+y_2(x)\\ &=c_0+c_1x+\frac{c_0}{3\cdot2}x^3+\frac{c_1}{4\cdot3}x^4 +\frac{c_0}{6\cdot5\cdot3\cdot2}x^6+\frac{c_1}{7\cdot6\cdot4\cdot3}x^7\\ &\quad+\frac{c_0}{9\cdot8\cdot6\cdot5\cdot3\cdot2}x^9 +\frac{c_1}{10\cdot9\cdot7\cdot6\cdot4\cdot3}x^{10}+\cdots \end{aligned}$$ $$\Rightarrow$$ $$y_1(x)=c_0\left[1+\frac{1}{3\cdot2}x^3+\frac{1}{6\cdot5\cdot3\cdot2}x^6+\frac{1}{9\cdot8\cdot6\cdot5\cdot3\cdot2}x^9+\cdots\right],$$ $$y_2(x)=c_1\left[x+\frac{1}{4\cdot3}x^4+\frac{1}{7\cdot6\cdot4\cdot3}x^7+\frac{1}{10\cdot9\cdot7\cdot6\cdot4\cdot3}x^{10}+\cdots\right].$$

[FULL CIRCLE] Now, you as a reader should prove or show that this answer does satisfy the original question.

[CHALLENGE] Use the other method that I was doing first letting $c_0=0$, $c_1=1$ and then $c_0=1$, $c_1=0$ to show that the answer is the same. I.e., there are multiple approaches to solving this problem for the recursive relation.

[CHALLENGE HINT] For the challenge, go back through the question and solve it with the $c_0=0,1$, $c_1=1,0$ as we did halfway through the problem.

When you find $c_0$, $c_1$, $c_2$, $c_3$, … . you plug them into the series to find $y_1$ and $y_2$ as a set $\{y_1,y_2\}$. You then add them together with the constant but use a different letter than $c$. E.g. (for example), $y=a_1y_1+a_2y_2$, where $a_1$ and $a_2$ are arbitrary constants just as $c_0$ and $c_1$ are. Meaning $a_1\equiv c_0$ and $a_2\equiv c_1$ for the comparison of the two solution methods.

[CHALLENGE 2] Can you prove that this satisfies the original question under the full circle of math?

The Math Monthly Journal · November 2026 | Volume 1 · Professor Jonathan D. Tullis · MathMonthly.com

AI Summary of the Lesson

  • Start at the ordinary point. In standard form, the coefficients $P(x)=0$ and $Q(x)=-x$ are analytic at $x=0$. Write the solution as a power series centered there and differentiate term by term.
  • Track the expression being substituted. The walkthrough first catches a substitution of $y’$ for $y$. It then shows how multiplying by $x$ raises every exponent in the series for $y$.
  • Match powers before equating coefficients. The constant term gives $c_2=0$. Aligning the remaining powers gives $c_{n+2}=c_{n-1}/((n+2)(n+1))$ for $n\ge1$.
  • Compare the two approaches. Keep $c_0$ and $c_1$ arbitrary and group their terms at the end, or use the pairs $(1,0)$ and $(0,1)$ separately. The recap asks you to try both and verify the resulting series in the original equation.

Video Recap | Power Series About an Ordinary Point

Watch the recap on YouTube