Taylor series: local power-series expansions and practical approximations

Definition and structure

Given a function f that is sufficiently differentiable at a point a, its Taylor series about a is built from the derivatives f(a), f'(a), f''(a), ... and is typically written as the infinite sum f(a) + f'(a)(x−a) + f''(a)(x−a)^2/2! + f'''(a)(x−a)^3/3! + ···. Each term involves a derivative at the centre a, divided by the factorial of the order and multiplied by a power of (x−a). The derivatives used to build the series are often described simply as the function's derivatives evaluated at the expansion point.

Convergence and analytic vs smooth functions

Not every infinitely differentiable function equals its Taylor series in a neighborhood of the expansion point. When a function does equal its Taylor series on some interval around the centre it is called analytic. The series has a radius of convergence, outside of which the series diverges. Taylor's theorem provides formulas for the remainder after truncating the series, with common remainder estimates named after Lagrange and Cauchy; these results quantify how good a finite polynomial approximation is. In complex analysis the radius of convergence equals the distance from the centre to the nearest singularity of the function.

There are classical counterexamples showing the distinction between smooth and analytic: a standard smooth function that is not analytic at the origin has all derivatives zero at that point while failing to equal its Taylor series; such examples demonstrate the necessity of checking convergence rather than relying solely on differentiability.

Elementary examples

• The exponential function has the Maclaurin series e^x = 1 + x + x^2/2! + x^3/3! + …, which converges for all real x.
• Sine and cosine have series that converge everywhere: sin x = x − x^3/3! + x^5/5! − … and cos x = 1 − x^2/2! + x^4/4! − …
• The geometric series 1/(1−x) = 1 + x + x^2 + x^3 + … converges only for |x| < 1 and provides a simple power-series expansion about zero.
• The natural logarithm has a Maclaurin-type expansion ln(1+x) = x − x^2/2 + x^3/3 − … valid for |x| < 1 (with x ≠ −1 at the boundary); other logarithmic expansions are taken about different centers.

Applications

• Numerical approximation: Truncated Taylor polynomials are used to estimate function values, to derive numerical methods for integration and differential equations, and to implement elementary functions in software libraries.
• Modeling and linearization: In engineering and physics, local linear or quadratic approximations simplify analysis near an operating point and underpin control design and stability analysis.
• Series solutions: When closed-form solutions are unavailable, power-series methods give formal solutions to differential equations and recurrence problems.
• Asymptotics and perturbation: Taylor expansions about a small parameter often provide the leading-order behaviour of a system and guide perturbation techniques and approximation hierarchies.

History and formal results

The expansion technique is named after Brook Taylor, whose work in the early 18th century introduced the theorem now known as Taylor's theorem; special expansions about zero are associated with Colin Maclaurin. Taylor's theorem and its remainder estimates formalize the relationship between derivatives and local polynomial approximations; see Taylor's theorem for precise statements and proof sketches. For introductory visual demonstrations of how adding terms improves the fit to the graph of a function, many calculus texts and interactive tools illustrate successive Taylor polynomials.

Distinctions, cautions and extensions

Important distinctions include the difference between a Taylor series and a general power series: a Taylor series is centered on a specific point and its coefficients are derivatives of a particular function, whereas a power series need not arise from any known function's derivatives. Users must check the radius of convergence and remainder bounds to ensure the truncated series gives acceptable accuracy; without this the polynomial approximation may be misleading. In applications, the finite Taylor polynomial provides an approximation to local behaviour, but the infinite series recovers the original function only when convergence conditions are met.

Related concepts and extensions include Fourier series (which expand functions in trigonometric bases rather than powers), Laurent series (which allow negative powers and are used near singularities), and asymptotic series (which approximate behaviour in limits but need not converge). For further reading on how Taylor series are used in computations and theoretical analysis consult sources on functions, numerical methods and complex analysis; introductory materials on functions and their expansions give practical recipes and examples.

For computational practitioners, Taylor expansions remain a fundamental bridge between local differential information and algebraic polynomial approximations, giving both conceptual insight and practical tools in science and engineering.