How To Find The Sum Of A Geometric Series

A geometric sequence multiplies by the same factor at every step: 3, 1.5, 0.75, 0.375, and onward after halving each time. The corresponding series is that list added up. The defining test is not "does it shrink" but "is the ratio constant" — take any term divided by its predecessor and compare across several pairs. If those quotients agree, the series is geometric regardless of whether it is growing, shrinking, or alternating in sign.

The infinite case is where the idea earns its reputation. Adding infinitely many positive numbers normally gives infinity, but if the terms shrink fast enough the running total approaches a ceiling without ever crossing it. Halving from 3 gives partial sums of 3, 4.5, 5.25, 5.625, and each stage recovers half the remaining distance to 6. After ten terms you are within 0.005859375 of the limit, after twenty within a rounding error, and you will never arrive — which is exactly what a limit means.

The condition for that ceiling is the single most important fact here: it exists when the absolute value of the ratio is strictly below one. "Strictly" matters. A ratio of exactly 1 produces terms that never shrink, so the total grows without bound; a ratio of -1 oscillates between two values and never settles. Anything further from zero than 1 — including -1.5, which alternates and grows — diverges. Checking |r| before quoting an infinite sum is not pedantry, it is the whole question.

Alternating series are the case that catches people out because the result looks wrong at first glance. Starting at 1 with ratio -0.5 gives 1, -0.5, 0.25, -0.125, and so on. The terms alternate sign but shrink, so the partial sums converge: they approach 0.6667, which is exactly 1 / (1 - (-0.5)). This matters in practice because alternating behaviour appears naturally in approximations, error analyses, and any model of a quantity that overshoots and corrects.

Why the finite formula has that particular shape is worth knowing rather than memorising. Multiply the whole series by r and subtract it from the original; every middle term cancels a partner shifted by one position, leaving only the first term and the last. The result is S - rS = a1 - a1 r^n, and dividing by (1 - r) gives the formula. Once you have seen the cancellation, you can rebuild the expression from scratch rather than trusting recall.

The ratio r = 1 is a genuine hole in that formula, because it puts zero in the denominator. It is not a failure of mathematics but a sign that you chose the wrong expression: when the ratio is exactly one, every term equals a1, so the sum is simply n times a1. Recognising this degenerate case is worth doing explicitly, because it is the one input that will otherwise produce a divide-by-zero message.

Geometric series sit underneath a surprising amount of practical arithmetic. Compound growth with reinvested returns is geometric, which is why future-value and annuity formulas contain (1 + r)^n. Repeated percentage decay — depreciation by a fixed percent, radioactive half-life, a bouncing ball losing a fixed fraction of its height each bounce — is geometric with a ratio below one. Any time someone says "grows by 8 percent a year", they have handed you r = 1.08.

Comparing the finite and infinite answers is a diagnostic rather than a curiosity. Starting at 3 with ratio 0.5, ten terms give 5.994140625 against a ceiling of 6, telling you almost all the total has already accumulated. Starting at 1 with ratio 2, ten terms give 1023 and there is no ceiling to compare against. The size of that gap is how you decide whether adding more terms is worth doing at all.

Finally, watch the difference between a ratio and a percentage. A quantity that "grows 8 percent per year" has ratio 1.08, not 0.08. Using 0.08 instead produces a series that collapses toward zero rather than growing, and because both answers are positive numbers nothing announces the swap. Writing r = 1 + rate at the top of every compounding calculation is a cheap habit that prevents the most expensive version of this mistake.