Real Analysis 21 | Reordering for Series [dark version]

Reordering terms in an infinite series can change its value—sometimes dramatically—but absolute convergence is the safeguard that prevents this. For conditionally convergent series, shuffling terms can alter the set of partial-sum accumulation points, producing different limits or even multiple “end behaviors.” A classic alternating series example illustrates how a carefully chosen rearrangement can shift accumulation values from one pair to another, showing that convergence alone is not enough to guarantee stability under reordering.

The transcript first contrasts finite sums with infinite series. In finite sums, swapping terms never changes the result. In infinite sums, however, reordering can change the limit because the partial sums depend on the order in which terms are added. An example built from alternating +1 and −1 terms (a series known to be non-convergent) demonstrates this vividly: the partial sums oscillate between two accumulation values (0 and 1). After a specific rearrangement that groups terms differently, the series still fails to converge, but the accumulation values shift (now to 1 and 2). The takeaway is direct: infinite series require caution—reordering can change what “converging” even means.

Next comes a convergent alternating series where the original limit is unique, supported by standard convergence criteria (alternating with decreasing magnitude). Even with a single original limit, the transcript shows that a rearrangement can still change the limit when the series is only conditionally convergent. The method is constructive: take two positive terms, then one negative term, then again two positives, then one negative, continuing in a pattern that preserves all original terms exactly once. This rearrangement yields a new convergent series whose limit is larger than the original—specifically described as (3/2)·C rather than C—highlighting that conditional convergence allows limit-shifting.

The final section turns from examples to theory by fixing the definition of a reordering. A reordering is formalized via a bijection τ from natural numbers to natural numbers: the rearranged series is obtained by placing the original term a_{τ(k)} in the k-th position. With that definition in place, the key theorem is stated: if a series is absolutely convergent, then every reordering is also absolutely convergent, and—most importantly—every reordering has the same limit as the original series. In other words, the “strange thing” seen for conditional convergence cannot occur under absolute convergence.

The proof strategy uses the Cauchy criterion (and its absolute-value variant) to control tails of the series. Letting A denote the original limit, the argument bounds the difference between A and the rearranged partial sums by splitting it into two parts: one controlled directly by the Cauchy property of the original series, and another controlled by the bijection τ’s ability to eventually “hit” all indices needed to cover the finite initial segment. Triangle inequality and tail estimates show that the rearranged partial sums can be made arbitrarily close to A, establishing that the rearranged series converges to the same limit. The result sets up a powerful practical rule: absolute convergence makes series order-independent, while conditional convergence does not.