On what interval is the sum of two power series guaranteed to be valid?

The sum is guaranteed to converge on the intersection of the individual intervals of convergence. This is the overlap rule. You must find I_A and I_B separately, then take the set of x-values where both converge. The intersection is the guaranteed interval, though the sum may converge on a larger set.

Does the product of two power series have the same guaranteed interval as the sum?

Yes. The Cauchy product also converges at least on the intersection of the individual intervals. The radius of convergence for the product is at least min(R_A, R_B), and the endpoint behavior follows the same rule: both series must converge at an endpoint for the product to be guaranteed there.

What happens if one power series has radius zero?

If a series converges only at its center (R = 0), the sum or product with any other series is guaranteed to converge only at that center. The intersection becomes the single point x = c, regardless of the other series' radius. This is a failure case: the combined series is essentially useless outside the center.

Can the sum or product converge on a larger interval than the intersection?

Yes, but the intersection is the guaranteed minimum. The combined series might converge on a larger interval due to cancellation or simplification, but you must prove that separately. Relying on the larger interval without proof is a mistake. The overlap rule gives you the safe region you can trust without additional analysis.

Do both series need the same center for the overlap rule to apply?

Yes. The sum and product definitions assume both series are centered at the same point c. If the centers differ, you must shift one series to match the other's center, which changes the coefficients. The overlap rule only applies once both series are expressed about the same c.

How do I test endpoints when using the overlap rule?

Endpoints are tested separately for each original series. Use the p-series test if the endpoint series simplifies to ∑ 1/n^p. Use the alternating series test if signs alternate. Use the comparison test or integral test for other forms. The nth-term test is your first check: if terms do not approach zero, the series diverges. Never apply the ratio test at endpoints, it is inconclusive there.

What is the most common error when applying the overlap rule?

Forgetting to test endpoints of the intersection. Students find the intersection of open intervals and stop, missing that one series may diverge at an endpoint while the other converges. The intersection of open intervals is not the same as the guaranteed interval for the sum or product. Always check endpoint convergence for both series before declaring the intersection.

The Overlap Rule for Adding Two Power Series Explained

The Overlap Rule for Adding Two Power Series

The most common mistake when adding two power series is assuming the sum converges everywhere both series converge individually. It does not. The sum or product of two power series is guaranteed to converge only on the intersection of their individual intervals of convergence. This is the overlap rule for adding two power series: you take the set of x-values where both series converge and use that as your guaranteed region. Everything else is speculation. The rule applies to both sums and Cauchy products, and it sets a floor, not a ceiling, the combined series might converge on a larger interval, but you cannot rely on that without separate proof.

Why the Method Works

Series A converges on interval I_A. Series B converges on interval I_B. The sum A+B and the product A*B converge at least on I_A ∩ I_B. That is the guaranteed region. The radius of convergence for the combined series is at least min(R_A, R_B), but the radius alone does not tell you which endpoints make the cut. You must test each endpoint of the intersection individually. If both series converge at x = c+R, then the sum converges there. If only one does, the sum diverges at that endpoint. The intersection inherits the stricter condition.

Why the Radius Is Not Enough

Two series can have the same radius but different endpoint behavior. Series A might converge at x = c+R while Series B diverges there. The intersection then excludes that endpoint. The radius gives you the open interval (c-R, c+R) as a starting point, but the final interval requires separate endpoint tests. The ratio test is inconclusive at endpoints, so you must apply p-series, alternating series, or comparison tests to each endpoint. A series convergence calculator will find the radius quickly, but you still have to verify the endpoints by hand, no calculator can decide conditional convergence for you.

A Worked Example

Take Series A: ∑ x^n from n=0 to ∞. This geometric series converges when |x| < 1, so I_A = (-1, 1). Series B: ∑ x^n/n from n=1 to ∞. Its radius is 1, but at x = -1 the series becomes the alternating harmonic series ∑ (-1)^n/n, which converges conditionally. At x = 1 it becomes the harmonic series ∑ 1/n, which diverges. So I_B = [-1, 1). The intersection I_A ∩ I_B is (-1, 1). The sum A+B = ∑ (1 + 1/n) x^n is guaranteed to converge on (-1, 1). The endpoint x = -1 is not in the intersection because Series A diverges there, the geometric series does not converge at x = -1. The product A*B = ∑ (∑_{k=1}^n 1/k) x^n is also guaranteed on (-1, 1).

How to Compute the Intersection Step by Step

First, find I_A using the ratio test or root test, then test each endpoint. Second, find I_B the same way. Third, take the set of x-values that belong to both intervals. Fourth, state that the sum or product converges at least on that set. Fifth, if you need the exact interval, test the sum or product series separately at each endpoint of the intersection. The intersection is the guaranteed minimum, not necessarily the full answer.

Calculation Steps

To perform the calculation, start with two power series centered at the same point c. Write each as ∑ a_n (x-c)^n and ∑ b_n (x-c)^n. Compute the radius R_A using the formula R = lim |a_n / a_{n+1}| or R = 1 / lim sup |a_n|^(1/n). Do the same for R_B. The radius of the sum or product is at least min(R_A, R_B). Next, test the endpoints x = c ± R_A for Series A and x = c ± R_B for Series B. Apply the p-series test, alternating series test, or comparison test as needed. The nth-term test is your first check, if the terms do not approach zero, the series diverges immediately.

A Concrete Calculation

Series A: ∑ (2^n)(x-1)^n / n!. The ratio test gives R_A = 1/2. At x = 1.5, the series becomes ∑ 1/n! which converges (p-series with p > 1? No, it is actually the exponential series, which converges absolutely). At x = 0.5, it becomes ∑ (-1)^n/n! which also converges by absolute convergence. So I_A = [0.5, 1.5]. Series B: ∑ (x-1)^n / n. Ratio test gives R_B = 1. At x = 2, the series becomes ∑ 1/n, a harmonic series that diverges. At x = 0, it becomes ∑ (-1)^n/n, an alternating harmonic series that converges conditionally. So I_B = [0, 2). The intersection I_A ∩ I_B is [0.5, 1.5) because at x = 1.5, Series B diverges (harmonic series at x = 2? Wait, x = 1.5 gives (0.5)^n/n, a convergent p-series with p = 1? No, the series is ∑ (0.5)^n/n, which converges by comparison to a geometric series, the ratio test works here, not p-series. At x = 1.5, Series B converges. At x = 0.5, Series B converges. So the intersection is actually [0.5, 1.5]. The sum converges guaranteed on [0.5, 1.5]. The failure case: if either series had R = 0, the intersection would be the single point x = c, regardless of the other series' radius.