the radius of convergence reaching the nearest singularity

Here is one of the most satisfying facts in the subject, the bridge between analysis and geometry. Take a holomorphic function and expand its Taylor series around a point z_0. The radius of convergence R is not some arbitrary technical number — it equals the distance from z_0 to the nearest point where the function fails to be holomorphic, its closest singularity. The series reaches out as far as it can and stops exactly when it bumps into trouble.

Think of it as the centre z_0 growing a disk outward. The disk can keep expanding as long as the function stays holomorphic; the moment a singularity — a pole, a branch point, an essential singularity — would fall inside, the disk has to stop. The boundary circle of the disk of convergence passes precisely through the nearest singularity (or through the nearest singularities if several tie for closest). One direction is intuitive: a power series cannot represent a function past a point where the function blows up, so R cannot exceed the distance to the nearest singularity. The other direction is the content of Taylor's theorem: as long as f is holomorphic on a disk, the Taylor series converges on all of it, so R is at least that distance. Together, R equals the distance, exactly.

This explains a real-variable mystery. The function 1/(1 + x^2) on the line is perfectly smooth everywhere, yet its Maclaurin series 1 - x^2 + x^4 - ... only converges for |x| < 1, and no real-axis reasoning reveals why. The complex view does: 1/(1 + z^2) has singularities at z = +i and z = -i, each at distance 1 from the origin, and the series feels them across the plane. So whenever a real Taylor series has a puzzlingly small radius, the explanation is almost always a pair of invisible complex singularities off the real axis.