Approximating with Polynomials
Complicated functions like , , or can be approximated near a point by polynomials built from their derivatives. Since polynomials use only addition and multiplication, this is how such functions are actually computed, and it is a central tool of analysis.
The simplest case is the linear (tangent-line) approximation: the value plus the slope times the displacement — literally the tangent line at . It is excellent for near and is the basis of Newton's method and differentials.
The Taylor polynomial improves accuracy by adding higher-derivative terms: Each term corrects the fit a little more: the constant matches the value, the linear term the slope, the quadratic the curvature, and so on. The special case (a Maclaurin series) gives clean expansions like .
The key practical facts:
- More terms generally improve accuracy and widen the region where the approximation is good.
- The approximation is local — most accurate near the expansion point , degrading as grows.
- The remainder (error) can be bounded, telling you how many terms you need for a desired accuracy.
Taylor approximation replaces a hard function by an easy polynomial that matches it locally — the workhorse behind numerical computation, physics approximations, and much of applied analysis.
Common pitfall: expecting a truncated Taylor polynomial to be accurate far from the expansion point. The approximation matches best near the center , and the error typically grows as increases — a few terms good near can be wildly wrong far away. More terms widen the accurate region but a finite polynomial never matches everywhere; the approximation is fundamentally local.
A smooth curve with successive Taylor polynomials (degree 1, 2, 3) overlaid, each accent curve hugging the function over a wider region near the center point as the degree increases.