Many continuous functions (such as or cannot be calculated using simple algebraic operations. Power series, and specifically Taylor series, allow us to represent these transcendental functions as “infinite polynomials,” which are easy to compute, differentiate, and integrate.
Power Series and Convergence
Before defining Taylor series, we must understand the behavior of power series.
A power series centered at is an infinite series of the form:
where is a variable and are the coefficients.
A power series centered at is an infinite series of the form:
where is a variable and are the coefficients.
Every power series converges at its center The set of all for which the series converges is an interval, governed by the radius of convergence.
For any power series exactly one of the following holds:
- The series converges only at (Radius of convergence
- The series converges absolutely for all (Radius of convergence
- There exists a real number such that the series converges absolutely for and diverges for
The value is the radius of convergence, and can be calculated using the Root or Ratio test:
For any power series exactly one of the following holds:
- The series converges only at (Radius of convergence
- The series converges absolutely for all (Radius of convergence
- There exists a real number such that the series converges absolutely for and diverges for
The value is the radius of convergence, and can be calculated using the Root or Ratio test:
Taylor and Maclaurin Series
If a function can be represented by a power series, the coefficients are uniquely determined by the function’s derivatives.
Let be a function with derivatives of all orders on an interval containing The Taylor series of centered at is:
If the series is called the Maclaurin series of
Let be a function with derivatives of all orders on an interval containing The Taylor series of centered at is:
If the series is called the Maclaurin series of
Taylor’s Theorem with Remainder
In practice, we can only compute finite sums. Taylor’s Theorem provides a formula for the error (remainder) when approximating a function with a Taylor polynomial of degree
Let have derivatives on an open interval containing For any there exists a number between and such that:
where is the -th degree Taylor polynomial:
and is the remainder (error) term, expressed in the Lagrange form:
Let have derivatives on an open interval containing For any there exists a number between and such that:
where is the -th degree Taylor polynomial:
and is the remainder (error) term, expressed in the Lagrange form:
Pause & Predict: Does guarantee convergence to Suppose a function has continuous derivatives of all orders and its Taylor series converges for all (radius of convergence Does the sum of the series necessarily equal for all Commit your intuition before opening the analysis below.
Show hintExpandCollapse
Recall the distinction between smooth () and real-analytic () functions. Does vanishing derivatives at zero imply the function itself is zero everywhere?
Reveal analysisExpandCollapse
Analysis: No! Having infinitely many derivatives does not imply that the Taylor series reconstructs the original function away from the expansion point. A function is called real-analytic only if its Taylor series actually converges to in a neighborhood of There exist infinitely differentiable functions that are not analytic: their remainder term fails to vanish as We analyze this profound counterexample below.
ProofExpand proofCollapse
We fix Let be the unique real number satisfying:
We want to prove that for some between and Define an auxiliary function for in the interval:
- Since has derivatives, is continuous and differentiable.
- Evaluate at
- By definition of we also have
- Evaluate at
- by our choice of
Since by Rolle’s Theorem, there exists between and such that Now, since and by Rolle’s Theorem again, there exists between and such that Repeating this process times, we find that there exists between and such that:
Differentiating exactly times:
- The -th derivative of is
- The -th derivative of the polynomial of degree is
- The -th derivative of is
Thus:
Evaluating at
Substituting back into our original equation yields the Lagrange remainder.
Let us write a Python script to plot the Maclaurin polynomials of and analyze how adding terms expands the interval where the approximation remains accurate.
Interactive Lab
Worked Trace: Rigorous Lagrange Error Bound for
Suppose we approximate using the Maclaurin polynomial and we require the approximation error to be strictly less than
-
Formulate the Lagrange Remainder: By Taylor’s Theorem, for the -th derivative is The remainder at is:
-
Establish a Rigorous Upper Bound: Because is strictly increasing, (using the elementary bound Therefore:
-
Determine the Required Degree : We require:
-
Factorial Growth Trace:
- For (insufficient).
- For (condition satisfied).
-
Conclusion: A 9th-degree Taylor polynomial guarantees an error strictly bounded by without needing to know the true value of in advance.
Counterexample & Failure Mode: Cauchy’s Smooth Non-Analytic Function
Does imply that the Taylor series converges to Consider Augustin-Louis Cauchy’s classical function:
-
Derivatives at the Origin: For the -th derivative has the form where is a polynomial. Using the substitution as By induction and the definition of the derivative, for every
-
Taylor Series Collapse: The Maclaurin series of around is: This power series converges for every (radius of convergence with sum identically zero.
-
Breakdown Analysis: For any Thus, the Taylor series agrees with only at the single point ! The function is infinitely differentiable on all of yet fails to be analytic at the origin because the remainder term does not converge to zero as
Euler’s Formula via Maclaurin Series
By substituting into the Maclaurin series of we can derive the fundamental connection between complex exponentiation and trigonometry.
Let The Maclaurin series for is:
Grouping real and imaginary parts:
Recognizing the real part as the Maclaurin series for and the imaginary part as the Maclaurin series for we establish Euler’s Formula:
Worked example. At the identity gives squaring gives Separating the even and odd terms is valid because the exponential power series converges absolutely for every complex so its terms may be regrouped. This is an extension of the real-variable series, not a substitute for defining the complex exponential.
Deep Practice
Determine radii and intervals of convergence, construct Taylor polynomials, and establish rigorous truncation error bounds via the Lagrange remainder.
Exercise 3.1.1: Radius and Interval of Convergence with Endpoint Analysis
Determine the radius of convergence and the exact interval of convergence (including endpoint behavior) for the power series (Theorem 3.1.2
Verification Check
Consider the power series . Evaluate each statement regarding its radius and interval of convergence:
Classify each option as correct or incorrect, or leave uncertain options unanswered (“I don't know”) and verify at any time.
HintExpandCollapse
After finding via the Ratio Test, test (alternating harmonic series) and (harmonic series) separately.
Worked solutionExpandCollapse
- Ratio test: Let Then Convergence holds for so and the open interval is
- Left endpoint The series becomes which converges conditionally by the Alternating Series Test.
- Right endpoint The series becomes the harmonic series, which diverges.
- Exact interval of convergence:
Verification criteria. Requires separate convergence checks at and and the half-closed interval
Exercise 3.1.2: Second-Degree Taylor Polynomial and Remainder Bound
Construct the quadratic Taylor polynomial of centered at use it to approximate and bound the error using Theorem 3.1.4
HintExpandCollapse
Compute for , then bound on using the worst-case endpoint .
Worked solutionExpandCollapse
- Derivatives at
- Quadratic Taylor polynomial (Definition 3.1.3): AtDefinition 3.1.3Taylor and Maclaurin SeriesLet be a function with derivatives of all orders on an interval containing . The **Taylor series** of centered at is:
- Lagrange remainder bound (Theorem 3.1.4): For some Since so:Theorem 3.1.4TaylorLet have derivatives on an open interval containing . For any , there exists a number between and such that:
Verification criteria. Requires numerical evaluation and bounding at the left endpoint
Exercise 3.1.3: Uniform Error Bound for the Cubic Maclaurin Approximation of Sine
Prove using Theorem 3.1.4
HintExpandCollapse
Note that the even-order derivative , so . Apply the degree- Lagrange remainder with .
Worked solutionExpandCollapse
- Exploit vanishing 4th term: For we have so the 3rd- and 4th-degree Maclaurin polynomials coincide:
- Apply Theorem 3.1.4of order There exists between and such that:Theorem 3.1.4TaylorLet have derivatives on an open interval containing . For any , there exists a number between and such that:
- Bound on Since for all and
Verification criteria. Requires recognizing writing and bounding and
Exercise 3.1.4: Implementing the Maclaurin Series of the Exponential Function
Complete the Python implementation of the Maclaurin partial sum for
Code Exercise
Complete the Maclaurin series function for e^x
def exp_maclaurin(x, terms): val = 0.0 import math for n in range(terms): # General term is x^n / n! val += () / math.factorial(n) return val
References & Further Reading
- Spivak, M. (2008). Calculus (4th ed.), Chapter 20 (“Approximation by Polynomial Functions”, including Lagrange and integral remainder forms) and Chapters 25–27 (complex power series and why singularities in govern real radius of convergence). Publish or Perish.
- Rudin, W. (1976). Principles of Mathematical Analysis (3rd ed.), Chapter 5 (Taylor’s Theorem, Theorem 5.15) & Chapter 8 (“Some Special Functions”, power series and Fourier series). McGraw-Hill.
- Apostol, T. M. (1967). Calculus, Volume 1: One-Variable Calculus, with an Introduction to Linear Algebra (2nd ed.), Chapter 7 (“Polynomial Approximations to Functions”) and Chapter 11 (“Infinite Series and Improper Integrals”). Wiley.