Sequences Convergence Flowchart Joseph Breen Background First, some context. Everything you’ve studied in calculus up until now has dealt with functions and integration of these functions. In comparison, sequences and series may initially seem very different. Because of this, this whole topic can seem kind of intimidating. However, it turns out that everything is the same! We’re really not doing anything all that different. Here’s why: The main object of study in math so far has been that of a function. In integral calculus, we take functions and integrate them. One big question we always ask about a function f is the following: what happens as x → ∞? Similarly, given an integral of a function (in particular, an improper integral where one of the bounds goes off to infinity), the question we always ask is the following: does the integral converge or diverge? This is summarized below: Functions Integrals R∞ f (x) 1 R∞ limx→∞ f (x) = ? 1 f (x) dx f (x) dx < ∞ ? Now, suppose I give you a function f , and only evaluate it at x = 1, 2, 3, . . . . This is a sequence! Let’s call this sequence an . Explicitly, {an } = {f (1), f (2), f (3), . . . }. Remember the main question about functions from above? We’re going to ask the same question about our sequence: what happens as n → ∞? In other words, does the sequence converge? Next, one (informal) way to think about integrating a function is that it is the process of adding up all the values of the function. If we apply this to our sequence (which again, is just a function), we get a series! Just as before, we ask the same question about a series as we do with an improper integral: does it converge or diverge? This is summarized here: Sequences Series P∞ an 1 P∞ limn→∞ an = ? 1 an an < ∞ ? So there is a very strong correspondence between functions and sequences, and very strong correspondence between integrals and series. In fact, it’s more than just a correspondence: a sequence is a 1 special type of function, and a series is a special type of integral. So you can apply all the ideas you know about functions and integrals to sequences and series. This connection makes things like the integral test seem very obvious, and brings to light why we have very similar tests for both improper integrals and series (direct comparison, p-test, etc.). Testing for Convergence Unfortunately (or fortunately), there are a lot of other tests for series that we can use that we didn’t use with integrals, so it can be difficult to figure out which one to use. Describing a step-by-step process is difficult to do in general, but here’s my attempt at describing the process my mind goes through when confronted P with a series. Given a series an , I proceed (roughly) in this order: 1. Divergence Test: If the sequence an doesn’t approach 0, then the series will definitely diverge. If the sequence does approach 0, we can’t conclude anything, so move on. 2. Special Series: Check to see if the series is in a special form: • Geometric Series: Looks (roughly) like P arn ; you can sometimes take series like P 2n+1 31−n term and morph them into geometric series. If |r| < 1, converges to first1−r P • Telescoping Series: Looks (roughly) like an − bn , because then things cancel out. Find the formula for the kth partial sum and let k → ∞. P 1 • P-Series: Looks like np . If p > 1 converges. If p ≤ 1, diverges. P P • Alternating Series: Looks like (−1)n bn . If bn → 0 and bn is decreasing, (−1)n bn converges. 3. Limit Ratio Test: If the series has a factorial, i.e., something like n!, chances are you should use the P | < 1, an converges. limit ratio test1 . If limn→∞ |a|an+1 n| 4. Comparison: If it isn’t a special form and doesn’t have a factorial, maybe we can compare it to something: P P n| • Limit Comparison: Take an and compare it to a simpler series bn . If 0 < limn→∞ |a |bn | < P P ∞, an and bn either both converge or both diverge. P P P P • Direct Comparison: If bn converges and an ≤ bn , then an converges. Alternatively, P P P P if bn diverges and an ≥ bn , then an diverges. 5. Integral Test: If nothing else works, try using the correspondence from above and look at the integral version of the series. Given that the function is decreasing, continuous, and positive, the series and integral will behave the same. 1 The reason for this is that (n+1)! n! = n + 1, so the terms have a habit of canceling out nicely. 2