A power series centered at a is sum from n=0 to infinity of c_n * (x - a)^n, where c_n are the coefficients and x is the variable. It is a "polynomial of infinite degree" that defines a function of x on whatever interval it converges. Within its interval of convergence, a power series can be differentiated and integrated term by term. Power series are the bridge between series and functions, culminating in Taylor series representations.
Start with the geometric series 1/(1 - x) = sum of x^n for |x| < 1 as the prototype power series. Manipulate it (substitute, differentiate, integrate) to generate new power series. Introduce the concept of radius of convergence. Emphasize that the power series defines a function whose domain is determined by convergence.
A power series is best understood as a polynomial that never stops: Σ c_n (x - a)^n = c₀ + c₁(x-a) + c₂(x-a)² + .... Like a polynomial, it defines a function of x. Unlike a polynomial, it may only converge for certain values of x — specifically, within a radius R of the center a. Outside that radius, the series diverges and the formula gives no meaningful value.
The prototype power series is the geometric series: 1/(1-x) = 1 + x + x² + x³ + ... for |x| < 1. You've already seen this; now recognize it as the simplest power series, centered at 0 with radius of convergence 1. Every concept about power series can be illustrated with this example first. Substituting -x for x gives 1/(1+x) = Σ(-1)^n x^n; substituting x² gives 1/(1-x²) = Σx^(2n). Manipulation is almost always faster than computing coefficients from scratch.
Within its interval of convergence, a power series is extraordinarily well-behaved — it can be differentiated and integrated term by term, just like a finite polynomial. The resulting series has the same radius of convergence. This is the bridge to Taylor series: if f(x) has a power series representation, you can recover the coefficients by differentiating. The connection between functions and series — allowing you to compute things like sin(0.1) to arbitrary precision, or integrate functions with no closed-form antiderivative — rests entirely on this property.
The key discipline is always respecting the interval of convergence. Outside it, the series is meaningless, and manipulations like term-by-term differentiation are not valid. Endpoint behavior (at x = a ± R) requires separate analysis and is one of the subtler aspects of series theory. As you move toward Taylor series and applications like solving differential equations, keeping the interval of convergence in mind will prevent errors that look algebraically reasonable but are analytically invalid.