First establish the interior convergence lemma for a power series. If converges at some , its terms are bounded: for a finite . For every ,
Comparison with a convergent geometric series proves absolute convergence at .
Now let
The set contains zero, since at only the constant term remains. If , the definition of supremum gives a point of convergence with . The lemma proves absolute convergence at . If , convergence would place in the set defining , a contradiction. This proves the existence of a radius of convergence, including and :
The proof leaves behavior on open; it need not be uniform around that circle. The two examples illustrate opposite possibilities.
For , the ratio of successive absolute terms is
The ratio test therefore gives absolute convergence for and divergence for , while at zero the series is trivially convergent. Hence . On the boundary, , and the convergent p-series proves
For , the inequality for gives
The first comes from ; both initial terms have exponent one. Thus the factorial-gap power series converges absolutely inside the unit circle. For , its terms have magnitudes , so the term test for divergence proves divergence and .
At every point with , each term has magnitude one, so the terms cannot tend to zero, regardless of their phases. Hence
The large gaps between exponents make this a lacunary power series; cancellation cannot overcome the necessary condition that the terms of a convergent series tend to zero.

Articles by others on the same topic (0)

There are currently no matching articles.