Calculus II: Power Series and Taylor Series

When a function is too complex for standard algebra or integration, how can you work with it? Engineers and scientists routinely approximate intricate physical behaviors—from heat dissipation to signal oscillations—using infinite polynomial series. Mastering power series and Taylor series transforms your ability to analyze, approximate, and compute solutions to otherwise intractable problems by representing functions as sums of infinitely many simple terms.

The Foundation: Power Series and Convergence

A power series is an infinite series of the form $$\sum _{n=0}^{\infty }{c}_n(x-a{)}^n=c_0+c_1(x-a)+c_2(x-a{)}^2+\cdots .$$ Here, $c_n$ are the coefficients, $x$ is the variable, and $a$ is the center of the series. When $a=0$, the series simplifies to $\sum c_n{x}^n$, often called a power series centered at zero.

Not every $x$ value yields a finite sum. The radius of convergence, $R$, is a non-negative number (or infinity) such that the series converges absolutely if $|x-a|<R$ and diverges if $|x-a|>R$. You typically find $R$ using the Ratio Test or Root Test on the absolute values of the terms. For example, applying the Ratio Test to the series ${\sum }_{n=0}^{\infty }\frac{x^n}{n!}$ involves examining the limit: $$L=\underset{n\to \infty }{\lim }\left|\frac{x^{n+1}}{(n+1)!}\cdot \frac{n!}{x^n}\right|=\underset{n\to \infty }{\lim }\frac{|x|}{n+1}=0.$$ Since $L=0<1$ for all $x$, the radius of convergence is $R=\infty$.

Finding the interval of convergence requires checking the endpoints $x=a\pm R$ separately. A series may converge absolutely or conditionally at an endpoint, or it may diverge. The interval is the set of all $x$ for which the series converges, which can be $(a-R,a+R)$, $[a-R,a+R]$, or variations including one endpoint.

Constructing Taylor and Maclaurin Series

A Taylor series is a specific, immensely powerful type of power series. If a function $f$ has derivatives of all orders at $a$, its Taylor series centered at $a$ is: $$\sum _{n=0}^{\infty }\frac{f^{(n)}(a)}{n!}(x-a{)}^n=f(a)+f^{\prime }(a)(x-a)+\frac{f^{\prime \prime }(a)}{2!}(x-a{)}^2+\cdots .$$ A Maclaurin series is simply a Taylor series centered at $a=0$. The derivation comes from a bold idea: can we build a polynomial whose derivatives at $a$ match the function's derivatives exactly? The coefficients $\frac{f^{(n)}(a)}{n!}$ are the unique solution to this matching condition.

Crucially, the existence of the Taylor series does not guarantee it converges to $f(x)$. When the series does converge to $f(x)$ on an interval, we say $f$ is analytic at $a$. For engineering purposes, most elementary functions are analytic wherever they are defined.

Common Series Expansions and Operations

You should memorize these fundamental Maclaurin series expansions, as they serve as building blocks:

• Exponential: $e^x={\sum }_{n=0}^{\infty }\frac{x^n}{n!}=1+x+\frac{x^2}{2!}+\frac{x^3}{3!}+\cdots$ for all $x$.
• Sine: $\sin x={\sum }_{n=0}^{\infty }\frac{(-1{)}^n{x}^{2n+1}}{(2n+1)!}=x-\frac{x^3}{3!}+\frac{x^5}{5!}-\cdots$ for all $x$.
• Cosine: $\cos x={\sum }_{n=0}^{\infty }\frac{(-1{)}^n{x}^{2n}}{(2n)!}=1-\frac{x^2}{2!}+\frac{x^4}{4!}-\cdots$ for all $x$.
• Geometric: $\frac{1}{1-x}={\sum }_{n=0}^{\infty }{x}^n=1+x+x^2+x^3+\cdots$ for $|x|<1$.

You can manipulate these series to create new ones using operations that work within the interval of convergence:

• Differentiation & Integration: Term-by-term differentiation or integration yields a new power series for the derivative or integral. The radius of convergence remains the same (though endpoint behavior may change).
• Addition/Subtraction & Multiplication: Combine series by adding/subtracting corresponding coefficients or by performing Cauchy product (discrete convolution) for multiplication.
• Composition: Substitute one series into another, often truncating to a useful polynomial. For example, to find the series for $e^{-x^2}$, substitute $u=-x^2$ into the $e^u$ series: $e^{-x^2}=1-x^2+\frac{x^4}{2!}-\frac{x^6}{3!}+\cdots$.

Estimating Error with the Taylor Remainder Theorem

Using a finite number of terms from an infinite series creates an approximation. The Taylor remainder theorem (or Taylor's inequality) quantifies the error. If you approximate $f(x)$ by its $n$th-degree Taylor polynomial $T_n(x)$ centered at $a$, the remainder (error) is $R_n(x)=f(x)-T_n(x)$.

Taylor's Theorem states that if $|f^{(n+1)}(u)|\le M$ for all $u$ between $a$ and $x$, then the error is bounded: $$|R_n(x)|\le \frac{M}{(n+1)!}|x-a{|}^{n+1}.$$ In practice, you find a suitable bound $M$ for the $(n+1)$th derivative on the interval. For instance, to approximate $\sin (0.1)$ using its third-degree polynomial $T_3(0.1)=0.1-(0.1{)}^3/6$, you'd note that $|f^{(4)}(u)|=|\sin u|\le 1$. Therefore, the error is at most $\frac{1}{4!}|0.1{|}^4\approx 4.17\times 10^{-6}$, confirming the approximation's high accuracy.

This error estimation is critical for engineering design, ensuring numerical solutions from series approximations meet required tolerances.

Applications to Approximation and Analysis

These series are not just theoretical constructs; they are fundamental computational tools.

• Approximating Definite Integrals: Integrals like $\int e^{-x^2}dx$ (the error function) have no elementary antiderivative. Instead, integrate the power series term-by-term: ${\int }_0^t{e}^{-x^2}dx={\int }_0^t(1-x^2+\frac{x^4}{2!}-\cdots )dx=t-\frac{t^3}{3}+\frac{t^5}{5\cdot 2!}-\cdots$.
• Solving Differential Equations: The power series method assumes a solution of the form $y=\sum c_n{x}^n$, substitutes into the differential equation, and solves for coefficients recursively. This is a primary technique for equations with variable coefficients (e.g., Bessel's equation).
• Evaluating Limits: Limits yielding indeterminate forms like $0/0$ are often elegantly resolved using series expansions. For example, ${\lim }_{x\to 0}\frac{\cos x-1}{x^2}$ is found by substituting the cosine series: ${\lim }_{x\to 0}\frac{(1-\frac{x^2}{2!}+\frac{x^4}{4!}\cdots )-1}{x^2}={\lim }_{x\to 0}(-\frac{1}{2}+\frac{x^2}{4!}\cdots )=-\frac{1}{2}$.

Common Pitfalls

1. Confusing the Series with the Function: The biggest error is assuming the Taylor series equals $f(x)$ everywhere. Always check the interval of convergence and, for approximation, use the Remainder Theorem to ensure the error is acceptable for your $x$.
2. Misapplying Operations Beyond the Radius: Adding, differentiating, or composing series is only valid within the intersection of their intervals of convergence. Multiplying the series for $\frac{1}{1-x}$ (convergent for $|x|<1$) by itself yields a series for $\frac{1}{(1-x{)}^2}$ valid only on the same interval $|x|<1$.
3. Incorrect Error Bound Calculation: When using Taylor's inequality, the bound $M$ must be a number that holds for all values $u$ between the center $a$ and the target $x$. Picking an $M$ that is too small (e.g., a local minimum of the derivative) invalidates the error guarantee.
4. Endpoint Neglect: Finding a radius $R=3$ does not mean the interval is $(-3,3)$. You must test $x=a\pm 3$ separately. A series may converge conditionally at one endpoint (like the alternating harmonic series) and diverge at the other.

Summary

• A power series $\sum c_n(x-a{)}^n$ represents a function as an infinite polynomial with a specific radius and interval of convergence determined by tests like the Ratio Test.
• The Taylor series is the unique power series whose derivatives match the function's at the center $a$. A Maclaurin series is a Taylor series centered at zero.
• Common expansions (for $e^x$, $\sin x$, $\cos x$, $\frac{1}{1-x}$) serve as essential building blocks that can be manipulated via term-by-term differentiation, integration, and algebraic operations.
• The Taylor Remainder Theorem provides a crucial bound $|R_n(x)|\le \frac{M}{(n+1)!}|x-a{|}^{n+1}$ for the error when using a finite polynomial approximation, enabling precision engineering.
• Applications are vast, including approximating non-elementary integrals, solving differential equations, and evaluating limits, making these series indispensable tools for analysis and computation.