Sequences, Series, and Convergence
Sequences, Series, and Convergence
Definition: A sequence is an ordered list of numbers indexed by position (formally, a function from the positive integers to the real numbers); a series is the sum of a sequence’s terms, and it converges only when that running total settles on a single finite value as more terms are added.
How It Works
- A sequence is an ordered, infinite list of numbers a₁, a₂, a₃, …; formally, it is a function from the positive integers to the real numbers, where plugging in n returns the nth term, aₙ.
- An arithmetic sequence has a constant common difference d between consecutive terms — each term minus the one before it always equals d — which gives the explicit formula aₙ = a₁ + (n-1)d.
- A geometric sequence has a constant common ratio r between consecutive terms — each term divided by the one before it always equals r — which gives the explicit formula aₙ = a₁ · r^(n-1).
- A series is what results from adding a sequence’s terms together: a₁ + a₂ + a₃ + … A sequence and its associated series are closely related, but they are not the same object, and they can behave very differently.
- The nth partial sum, written Sₙ, is the sum of only the first n terms of a series: Sₙ = a₁ + a₂ + … + aₙ. The partial sums themselves form a brand-new sequence, S₁, S₂, S₃, …
- A sequence converges when its individual terms aₙ approach one specific finite number as n grows without bound; that number is the sequence’s limit.
- A series converges when its sequence of partial sums Sₙ approaches one specific finite number as n grows without bound; that number is called the sum of the series. This is a genuinely different question from whether the underlying terms converge, and conflating the two is the single most common mistake in this topic.
- A geometric series with |r| < 1 converges, because each new term is smaller than the last by a fixed factor, so the amount being added on shrinks toward nothing; its infinite sum is exactly S = a₁ / (1 - r).
- A geometric series with |r| ≥ 1 diverges: the terms never shrink toward zero (they stay constant size or keep growing), so the running total can never settle on a finite value.
- Shrinking terms are not enough to guarantee convergence. The harmonic series, 1 + 1/2 + 1/3 + 1/4 + …, is the classic counterexample: its terms march steadily toward 0, yet its partial sums grow without bound, so the series diverges.
- Whether a series converges depends only on its infinite tail, not on any finite starting stretch; changing, removing, or adding a finite number of terms at the beginning never changes whether the series converges, only what value it converges to.
Illustration
var pointsGroup = document.getElementById(‘series-points’); var limitLine = document.getElementById(‘series-limit-line’); var limitLabel = document.getElementById(‘series-limit-label’); var yMaxTxt = document.getElementById(‘series-y-max-txt’); var yMinTxt = document.getElementById(‘series-y-min-txt’);
var rTxt = document.getElementById(‘series-r-txt’); var nTxt = document.getElementById(‘series-n-txt’); var snTxt = document.getElementById(‘series-sn-txt’); var limitValTxt = document.getElementById(‘series-limitval-txt’);
function fmt(x) { return (Math.round(x * 100) / 100).toFixed(2).replace(/.00$/, ‘.0’); }
function xPix(i) { return plotLeft + i * ((plotRight - plotLeft) / 30); }
function update() { var r = parseFloat(rIn.value); var n = parseInt(nIn.value, 10); rOut.textContent = fmt(r); nOut.textContent = String(n);
// Genuine partial sums: add terms one at a time in a loop instead of
// jumping straight to the closed-form (1 - r^n) / (1 - r) shortcut.
var sums = [];
var term = a1;
var running = 0;
for (var i = 1; i <= n; i++) {
running += term;
sums.push(running);
term *= r;
}
var sn = sums[sums.length - 1];
var limit = a1 / (1 - r); // theoretical closed-form limit; slider caps |r| at 0.95 so this never divides by zero
var sMin = sums[0], sMax = sums[0];
for (var k = 1; k < sums.length; k++) {
if (sums[k] < sMin) sMin = sums[k];
if (sums[k] > sMax) sMax = sums[k];
}
var lo = Math.min(sMin, limit);
var hi = Math.max(sMax, limit);
if (hi - lo < 0.4) {
var mid = (hi + lo) / 2;
lo = mid - 0.4;
hi = mid + 0.4;
}
var pad = (hi - lo) * 0.1;
lo -= pad;
hi += pad;
function yPix(v) { return plotBottom - (v - lo) / (hi - lo) * (plotBottom - plotTop); }
while (pointsGroup.firstChild) pointsGroup.removeChild(pointsGroup.firstChild);
for (var j = 1; j <= n; j++) {
var c = document.createElementNS(svgNS, 'circle');
c.setAttribute('cx', xPix(j).toFixed(1));
c.setAttribute('cy', yPix(sums[j - 1]).toFixed(1));
if (j === n) {
c.setAttribute('r', '5');
c.setAttribute('fill', 'var(--brass-bright)');
c.setAttribute('stroke', 'var(--brass)');
c.setAttribute('stroke-width', '1.5');
} else {
c.setAttribute('r', '3.2');
c.setAttribute('fill', 'var(--accent)');
}
pointsGroup.appendChild(c);
}
var ly = yPix(limit);
limitLine.setAttribute('y1', ly.toFixed(1));
limitLine.setAttribute('y2', ly.toFixed(1));
limitLabel.setAttribute('y', (ly - 6).toFixed(1));
limitLabel.textContent = 'limit = ' + fmt(limit);
yMaxTxt.textContent = fmt(hi);
yMinTxt.textContent = fmt(lo);
rTxt.textContent = 'r = ' + fmt(r);
nTxt.textContent = 'n = ' + n;
snTxt.textContent = 'Sₙ = ' + fmt(sn);
limitValTxt.textContent = 'Limit = ' + fmt(limit);}
rIn.addEventListener(‘input’, update); nIn.addEventListener(‘input’, update); resetBtn.addEventListener(‘click’, function () { rIn.value = 0.5; nIn.value = 10; update(); });
update(); })();
Under the Hood
Arithmetic sequence, nth term: aₙ = a₁ + (n-1)d
Geometric sequence, nth term: aₙ = a₁ · r^(n-1)
Geometric series, finite sum (r ≠ 1): Sₙ = a₁(1 - r^n) / (1 - r)
Geometric series, infinite sum (|r| < 1): S = a₁ / (1 - r)The finite-sum formula comes from a telescoping trick: multiply Sₙ by r, subtract that from Sₙ, and every middle term cancels, leaving Sₙ(1 - r) = a₁ - a₁r^n. As n grows, if |r| < 1 then r^n shrinks toward 0, and the finite formula quietly turns into the infinite one.
Worked Example 1: Finding a term in an arithmetic sequence. Given: a₁ = 7, d = 4. Find the 12th term. Step 1: apply aₙ = a₁ + (n-1)d. Step 2: a₁₂ = 7 + (12-1)(4) = 7 + 44. Answer: a₁₂ = 51.
Worked Example 2: Summing a finite geometric series. Given: a₁ = 3, r = 2. Find the sum of the first 6 terms. Step 1: apply Sₙ = a₁(1 - r^n) / (1 - r). Step 2: S₆ = 3(1 - 2⁶) / (1 - 2) = 3(1 - 64) / (-1) = 3(-63) / (-1). Answer: S₆ = 189 (check by brute force: 3+6+12+24+48+96 = 189).
Worked Example 3: Summing an infinite geometric series. Given: a₁ = 5, r = 0.2. Step 1: confirm |r| = 0.2 < 1, so the infinite-sum formula applies and the series converges. Step 2: apply S = a₁ / (1 - r) = 5 / (1 - 0.2) = 5 / 0.8. Answer: S = 6.25.
History
- Zeno of Elea posed his paradoxes of motion around the 5th century BCE; “Achilles and the Tortoise,” in modern terms, is really a question about whether an infinite geometric series can sum to a finite distance, and it unsettled philosophers for centuries before the mathematics existed to resolve it cleanly.
- Archimedes, around the 3rd century BCE, used a geometric series with ratio 1/4 to compute the area of a parabolic segment, an early, informal use of an infinite series to get an exact finite answer, long before anyone had a formal theory of limits.
- The French scholar Nicole Oresme proved around 1350 that the harmonic series diverges, using an argument that groups its terms into blocks — (1/3+1/4) exceeds 1/4+1/4 = 1/2, (1/5+1/6+1/7+1/8) exceeds 4×(1/8) = 1/2, and so on — showing the sum keeps adding at least another 1/2 forever, even though the individual terms shrink to zero.
- Infinite series became a working tool of calculus in the 17th century, as Newton and Leibniz used them freely to represent functions, well before anyone had rigorously defined what “convergence” actually meant.
- Augustin-Louis Cauchy and Karl Weierstrass, in the 19th century, formalized limits and convergence with precise definitions, putting series manipulation that had been used somewhat loosely for 150 years onto rigorous logical footing.
- Cauchy also developed several of the convergence tests, including the ratio test, still taught today, giving mathematicians systematic ways to decide convergence instead of relying on intuition — intuition that the harmonic series had already shown could be misleading.
Why It Matters
- Compound interest and loan amortization schedules are geometric series in disguise: each period’s balance multiplies by a fixed factor, so the geometric series formula directly gives loan balances, mortgage payments, and the future value of regular savings contributions.
- Computer science leans on geometric series to analyze recursive algorithms: when a recursive algorithm does work at every level of recursion (like merge sort or a balanced divide-and-conquer search), summing that work across all levels is literally evaluating a geometric, or near-geometric, series.
- Zeno’s paradoxes stop being paradoxes once series convergence is understood: covering infinitely many shrinking distances can still take only a finite amount of time, because the corresponding infinite series of time intervals converges to a finite sum.
- Taylor and Maclaurin series represent complicated functions — sine, cosine, eˣ, logarithms — as infinite series of simple polynomial terms; engineers and physicists lean on this constantly, approximating a function with just its first few terms whenever the full function is too costly to compute directly.
- Signal processing depends on Fourier series, which represent a periodic signal as an infinite sum of sine and cosine waves; whether that sum converges, and how fast, determines how accurately audio, images, and radio signals can be filtered, compressed, or reconstructed.
- Physics uses infinite series routinely for approximation, from calculating orbital perturbations to reducing relativistic formulas back to their familiar Newtonian form when speeds are small compared to light.
Common Pitfalls
- Assuming that terms shrinking to zero guarantees a series converges. The harmonic series is the standard counterexample: its terms 1/n approach 0, but its partial sums grow without bound.
- Making sign errors with a negative common ratio, especially forgetting that r^(n-1) alternates sign when r is negative, which flips whether a given term adds to or subtracts from the running total.
- Plugging numbers into S = a₁/(1-r) when |r| ≥ 1. That formula is only valid for |r| < 1; for |r| ≥ 1 the series simply has no finite sum, so the formula does not misfire so much as not apply at all.
- Confusing a sequence’s limit with a series’ sum. A sequence’s terms can converge to 0 while the series formed by adding those same terms diverges, because “do the terms approach a limit” and “do the partial sums approach a limit” are two entirely different questions.
- Treating the finite-sum formula Sₙ = a₁(1-r^n)/(1-r) and the infinite-sum formula S = a₁/(1-r) as interchangeable. The infinite version only exists as a limiting case, once |r| < 1 lets r^n vanish; for a finite n, the r^n term still matters and cannot be dropped.
- Forgetting that a₁ names the first term actually being summed, not a fixed coefficient independent of where counting starts; reindexing a sequence to begin at n=0 instead of n=1 changes which term plays the role of a₁ in every formula.
Comparison
| Property | Arithmetic Sequence | Geometric Sequence |
|---|---|---|
| Defining relationship | Constant difference d between consecutive terms | Constant ratio r between consecutive terms |
| nth-term formula | aₙ = a₁ + (n-1)d | aₙ = a₁ · r^(n-1) |
| Infinite series (sum of all terms) | Never converges (grows without bound) unless every term is 0 | Converges only if |r| < 1, to S = a₁/(1-r); otherwise diverges |
FAQ
Does Zeno’s “Achilles and the Tortoise” paradox mean motion is actually impossible? No. The paradox quietly assumes that summing infinitely many time intervals must take infinite time, but that is false: if each interval shrinks by a constant factor, the infinite sum of intervals is itself finite, exactly because it is a converging geometric series, so Achilles catches the tortoise in a perfectly ordinary, finite amount of time.
If a series’ terms approach zero, why doesn’t the harmonic series converge? Convergence requires the terms to shrink fast enough that the partial sums stop growing. The harmonic series’ terms do shrink, but too slowly: grouping them as Oresme did shows each successive block still contributes at least 1/2 to the running total, forever, so the sum has no ceiling.
How do you know a series converges without adding up infinitely many terms? Mathematicians use convergence tests — the ratio test, comparison test, integral test, and others formalized in the 19th century — that examine how the terms behave as n grows, without ever needing to add up an actual infinite number of them.
Example
A ball dropped from 1 meter, that rebounds to exactly half its previous height on every bounce, travels a total distance given by a convergent geometric series: after the initial 1-meter drop, each bounce adds a rise and a fall of 2·(1/2)^k meters. Summing that series, with a₁ = 1 and r = 1/2 for the bounces plus the initial drop, gives a total distance of exactly 3 meters, even though the ball technically bounces infinitely many times before coming to rest — the same kind of resolution that dissolves Zeno’s paradox above.