Power Series To Solve Differential Equations

Introduction

Solving differential equations is a cornerstone of applied mathematics, physics, and engineering. So this is where the power series method becomes indispensable. Now, it provides a systematic, algorithmic approach to finding solutions in the form of infinite series, effectively transforming a difficult differential equation into a manageable algebraic problem of determining coefficients. While many standard equations yield to elementary methods like separation of variables or integrating factors, a vast number of real-world problems—ranging from quantum mechanics to fluid dynamics—involve equations with variable coefficients that resist closed-form solutions using standard functions. This article offers a thorough look to understanding, applying, and mastering the power series technique for solving ordinary differential equations (ODEs).

Detailed Explanation

What is a Power Series Solution?

A power series is an infinite sum of the form $\sum_{n=0}^{\infty} a_n (x - x_0)^n$, where $a_n$ are constants (coefficients) and $x_0$ is the center of the series. When we apply this to differential equations, we assume the unknown function $y(x)$ can be represented by such a series. The fundamental strategy relies on the uniqueness of power series expansions: if two power series are equal for all $x$ in an interval, their corresponding coefficients must be identical.

This method is particularly powerful for linear homogeneous ODEs with variable coefficients, typically written in standard form: $y'' + P(x)y' + Q(x)y = 0$ If $P(x)$ and $Q(x)$ are analytic at a point $x_0$ (meaning they can be represented by convergent power series near $x_0$), then $x_0$ is called an ordinary point. At an ordinary point, the theorem of existence and uniqueness guarantees that two linearly independent power series solutions exist, converging at least within the radius of convergence of $P(x)$ and $Q(x)$.

Ordinary Points vs. Regular Singular Points

The behavior of the coefficients $P(x)$ and $Q(x)$ dictates the solution strategy. Day to day, we use a standard Taylor series form $y = \sum a_n (x-x_0)^n$. Because of that, * Regular Singular Points: $P(x)$ or $Q(x)$ are not analytic, but $(x-x_0)P(x)$ and $(x-x_0)^2Q(x)$ are analytic. * Ordinary Points: Both $P(x)$ and $Q(x)$ are analytic. Here, the standard power series fails, and we must use the Method of Frobenius, assuming a solution of the form $y = (x-x_0)^r \sum a_n (x-x_0)^n$, where $r$ is determined by the indicial equation.

Understanding this classification is the first critical step before attempting any calculation.

Step-by-Step Concept Breakdown

Solving an ODE using power series at an ordinary point follows a rigid, repeatable workflow. Mastering these steps ensures accuracy even for complex equations.

Step 1: Assume the Series Form and Compute Derivatives

Assume the solution $y(x) = \sum_{n=0}^{\infty} a_n x^n$ (centering at $x_0=0$ for simplicity). Compute the derivatives term-by-term: $y'(x) = \sum_{n=1}^{\infty} n a_n x^{n-1}$ $y''(x) = \sum_{n=2}^{\infty} n(n-1) a_n x^{n-2}$ Note: The starting index shifts because the derivative of the constant term $a_0$ is zero.

Step 2: Substitute into the Differential Equation

Plug $y, y', y''$ into the ODE. This results in a sum of several power series involving $x$ raised to various powers.

Step 3: Align the Powers of $x$ (Index Shifting)

This is the most technically demanding step. To combine the series into a single summation, all series must have the same generic term $x^k$ (or $x^n$) It's one of those things that adds up..

• Identify the current power in each series (e.g., $x^{n-2}, x^{n-1}, x^n$).
• Perform index shifts: change the summation index $n$ so that the exponent matches the target power (usually $x^n$).
• Adjust the starting index ($n=0, 1, 2...$) accordingly.
• Example: $\sum_{n=2}^{\infty} n(n-1)a_n x^{n-2} \rightarrow$ let $k=n-2 \rightarrow n=k+2 \rightarrow \sum_{k=0}^{\infty} (k+2)(k+1)a_{k+2} x^k$.

Step 4: Combine Series and Set Coefficients to Zero

Once all series are expressed as $\sum (\dots) x^n$, combine them into a single summation: $\sum_{n=0}^{\infty} [\text{Expression involving } a_n, a_{n+1}, a_{n+2} \dots] x^n = 0$ Because the power series is identically zero, the coefficient of every power $x^n$ must be zero. This yields the Recurrence Relation Surprisingly effective..

Step 5: Solve the Recurrence Relation

The recurrence relation expresses higher coefficients ($a_{n+2}$) in terms of lower ones ($a_n, a_0, a_1$).

• Calculate $a_2, a_3, a_4 \dots$ in terms of $a_0$ and $a_1$.
• $a_0$ and $a_1$ remain arbitrary constants (determined by initial conditions).
• Identify patterns for even ($n=2k$) and odd ($n=2k+1$) coefficients separately.

Step 6: Write the General Solution

Substitute the coefficient formulas back into the assumed series $y = \sum a_n x^n$. Separate the series into the part multiplied by $a_0$ and the part multiplied by $a_1$: $y(x) = a_0 y_1(x) + a_1 y_2(x)$ $y_1$ and $y_2$ are the two linearly independent fundamental solutions.

Real Examples

Example 1: Airy’s Equation ($y'' - xy = 0$)

This classic equation appears in diffraction theory and quantum mechanics (particle in a linear potential). It has an ordinary point at $x=0$ Simple, but easy to overlook..

1. Assume: $y = \sum_{n=0}^\infty a_n x^n$, $y'' = \sum_{n=2}^\infty n(n-1)a_n x^{n-2}$.
2. Substitute: $\sum_{n=2}^\infty n(n-1)a_n x^{n-2} - x \sum_{n=0}^\infty a_n x^n = 0$.
3. Align Powers:
   • First sum: shift $n \to n+2 \rightarrow \sum_{n=0}^\infty (n+2)(n+1)a_{n+2} x^n$.
   • Second sum: $x \sum a_n x^n = \sum a_n x^{n+1}$. Shift $n \to n-1 \rightarrow \sum_{n=1}^\infty a_{n-1} x^n$.
4. Combine: Write $n=0$ term separately for the first sum: $2a_2 + \sum_{n=1}^\infty [(n+2)(n+1)a_{n+2} - a_{n-1}] x^n = 0$
5. Recurrence:
   • $n=0: 2a_2 = 0

Solving the Recurrence for Airy’s Equation

From the previous step we obtained the condition for the lowest power:

[ 2a_{2}=0\qquad\Longrightarrow\qquad a_{2}=0 . ]

Now we turn to the generic term that appears for (n\ge 1):

[ (n+2)(n+1)a_{n+2}-a_{,n-1}=0 \quad\Longrightarrow\quad a_{n+2}= \frac{a_{,n-1}}{(n+2)(n+1)} . ]

Because the index on the right‑hand side is three steps behind, the recurrence naturally separates the coefficients into three inter‑laced subsequences:

• Even‑indexed coefficients ((n=0,3,6,\dots)) are generated from (a_{0}).
• Odd‑indexed coefficients ((n=1,4,7,\dots)) are generated from (a_{1}).
• The remaining coefficients ((n=2,5,8,\dots)) are all zero, as already seen from (a_{2}=0).

Carrying out the first few substitutions gives a clear pattern:

[ \begin{aligned} a_{3}&=\frac{a_{0}}{3\cdot2},\[4pt] a_{4}&=\frac{a_{1}}{4\cdot3},\[4pt] a_{5}&=\frac{a_{2}}{5\cdot4}=0,\[4pt] a_{6}&=\frac{a_{3}}{6\cdot5}= \frac{a_{0}}{6\cdot5\cdot3\cdot2} =\frac{a_{0}}{6!},2!,\[4pt] a_{7}&=\frac{a_{4}}{7\cdot6}= \frac{a_{1}}{7\cdot6\cdot4\cdot3} =\frac{a_{1}}{7!},3!

In general one finds

[ a_{3k}= \frac{a_{0}}{(3k)!},k!,,\qquad a_{3k+1}= \frac{a_{1}}{(3k+1)!},k!,,\qquad a_{3k+2}=0\qquad(k=0,1,2,\dots). ]

Substituting these expressions back into the assumed series

[ y(x)=\sum_{n=0}^{\infty}a_{n}x^{n} ]

splits naturally into an even‑powered part multiplied by (a_{0}) and an odd‑powered part multiplied by (a_{1}):

[ \begin{aligned} y(x)&=a_{0}\Bigl[1+\frac{x^{3}}{3!}+\frac{x^{6}}{6!},2!+\frac{x^{9}}{9!},2!,3!+\cdots\Bigr] \ &\quad +a_{1}\Bigl[x+\frac{x^{4}}{4!}+\frac{x^{7}}{7!},3!+\frac{x^{10}}{10!},3!,4!+\cdots\Bigr]. \end{aligned} ]

These two linearly independent series are precisely the Airy functions (\operatorname{Ai}(x)) and (\operatorname{Bi}(x)), up to fixed normalization constants. In compact notation one may write

[ y(x)=C_{1},\operatorname{Ai}(x)+C_{2},\operatorname{Bi}(x), ]

where (C_{1}=a_{0}) and (C_{2}=a_{1}) after absorbing the factorial‑ratio constants into the definition of the special functions. The series converge for every real (or complex) (x); the radius of convergence is infinite because the coefficients decrease faster than any geometric progression The details matter here..

A Second Illustration: Bessel’s Equation of Order Zero

To see how the same machinery works for a different class of problems, consider

[ x^{2}y''+xy'+x^{2}y=0, ]

the Bessel equation of order zero. It possesses an ordinary point at (x=0).

1. Series ansatz
   [ y=\sum_{n=0}^{\infty}a_{n}x^{n},\qquad y'=\sum_{n=1}^{\infty}na_{n}x^{n-1},\qquad y''=\sum_{n=2}^{\infty}n(n-1)a_{n}x^{n-2}. ]

2. Substitution
   [ x^{2}y''+xy'+x^{2}y =\sum_{n=2}^{\infty}n(n-1)a_{n}x^{n

[ +\sum_{n=1}^{\infty}na_{n}x^{n}+\sum_{n=0}^{\infty}a_{n}x^{n+2}=0. ]

3. Index alignment
   Shift the summation index in the third series by letting (m=n+2) (so (n=m-2)): [ \sum_{n=0}^{\infty}a_{n}x^{n+2}=\sum_{m=2}^{\infty}a_{m-2}x^{m}. ] The first two series already involve (x^{n}) with (n\ge 2) and (n\ge 1) respectively. Writing everything as a single sum over (x^{n}) starting at (n=2) (and treating the (n=0,1) terms separately) gives [ \bigl[1\cdot a_{1}x\bigr] + \sum_{n=2}^{\infty}\Bigl[n(n-1)a_{n}+na_{n}+a_{n-2}\Bigr]x^{n}=0. ]

4. Indicial equation and recurrence
   Equating coefficients of like powers to zero yields: [ \begin{aligned} n=1:&\quad a_{1}=0,\ n\ge 2:&\quad \bigl[n(n-1)+n\bigr]a_{n}+a_{n-2}=n^{2}a_{n}+a_{n-2}=0. \end{aligned} ] Hence the recurrence relation is [ a_{n}=-\frac{a_{n-2}}{n^{2}},\qquad n\ge 2. ]

5. Structure of the coefficients
   Because (a_{1}=0), the recurrence forces all odd coefficients to vanish: [ a_{3}=a_{5}=a_{7}=\cdots=0. ] The even coefficients are determined by (a_{0}) (which remains arbitrary): [ \begin{aligned} a_{2}&=-\frac{a_{0}}{2^{2}},\ a_{4}&=-\frac{a_{2}}{4^{2}}=\frac{a_{0}}{2^{2}4^{2}},\ a_{6}&=-\frac{a_{4}}{6^{2}}=-\frac{a_{0}}{2^{2}4^{2}6^{2}},\ &\ \vdots\ a_{2k}&=(-1)^{k}\frac{a_{0}}{(2^{2})(4^{2})\cdots(2k)^{2}} =(-1)^{k}\frac{a_{0}}{2^{2k}(k!)^{2}}. \end{aligned} ]

6. The series solution
   Choosing the conventional normalization (a_{0}=1) yields the Bessel function of the first kind of order zero: [ J_{0}(x)=\sum_{k=0}^{\infty}\frac{(-1)^{k}}{(k!)^{2}}\left(\frac{x}{2}\right)^{2k} =1-\frac{x^{2}}{2^{2}}+\frac{x^{4}}{2^{2}4^{2}}-\frac{x^{6}}{2^{2}4^{2}6^{2}}+\cdots. ] This series converges for all finite (x) (infinite radius of convergence), as can be verified by the ratio test.

   A second linearly independent solution, the Bessel function of the second kind (Neumann function) (Y_{0}(x)), cannot be expressed as a pure power series about (x=0) because it contains a logarithmic singularity. Even so, it is obtained via the Frobenius method (or by taking a limit of (J_{\nu}(x)) as (\nu\to 0)) and has the asymptotic form [ Y_{0}(x)\sim \frac{2}{\pi}\left(\ln\frac{x}{2}+\gamma\right)J_{0}(x)+\cdots, ] where (\gamma) is the Euler–Mascheroni constant. The general solution of Bessel’s equation of order zero is therefore [ y(x)=C_{1}J_{0}(x)+C_{2}Y_{0}(x).

Conclusion

The two examples worked out above—Airy’s equation and Bessel’s equation of order zero—illustrate the core workflow of the power series method for linear ordinary differential equations with analytic coefficients:

1. Assume a power series (y=\sum

$a_n x^n) as a candidate solution. 2. Substitute the series and its derivatives into the original differential equation. 3. That said, Shift indices so that all terms are expressed as a single summation over a common power of $x$. Think about it: 4. Determine the recurrence relation by equating the coefficient of each power of $x$ to zero. So naturally, 5. Solve for the coefficients to construct the general solution, identifying any singularities that might necessitate the use of the Frobenius method.

Not obvious, but once you see it — you'll see it everywhere And that's really what it comes down to..

Through this systematic approach, complex differential equations that do not possess solutions in terms of elementary functions (like polynomials, exponentials, or trigonometric functions) can be solved using transcendental functions, providing a foundation for solving more advanced problems in physics and engineering.