core29 min studyLesson 174 in Mathematics

Rearrangements and infinite products

Identify which infinite operations preserve a value and translate nonzero product convergence into logarithmic sums.

Before you begin

On this page

The question for today

What you’ll understand

Explain why an absolutely convergent scalar sum is invariant under permutations.
Construct the mechanism by which conditional rearrangements alter a sum.
Analyze positive infinite products using logarithms and a telescoping example.

Finite addition and multiplication allow us to change the order of operations freely. Infinite expressions are limits of finite operations, so the order in which information enters can become part of their definition. A permutation changes which finite subsets occur first; a product compounds many small relative changes. This lesson makes both issues explicit. Instead of applying finite algebra to an already completed infinity, track partial sums or partial products and justify the new limiting procedure.

A finite tail budget makes permutations safe

Suppose a scalar series is absolutely convergent. Choose a finite initial set whose omitted magnitudes sum to less than a tolerance. Any permutation eventually includes that entire set. The additional included terms and the terms still omitted all lie in the same small absolute tail, so the permuted partial sums approach the original sum. This proof needs a bijection of the positive integers: every term must appear exactly once, with none silently removed or duplicated.

Conditional real series behave differently. Their positive subseries diverges to positive infinity and their negative subseries to negative infinity; otherwise convergence of their difference would force finite total magnitude. To target a chosen real number, add unused positive terms until passing above it, then unused negative terms until passing below it, and repeat. Because individual terms approach zero, the overshoots shrink. The resulting rearrangement converges to the chosen target despite using each original term once.

\sum_{n=1}^{\infty}|a_n|<\infty,\quad\pi:\mathbb N\to\mathbb N\ \text{bijective}\quad\Longrightarrow\quad\sum_{n=1}^{\infty}a_{\pi(n)}=\sum_{n=1}^{\infty}a_n

Absolute convergence guarantees permutation invariance for real or complex scalar series. A conditional real series does not have this guarantee.↔ Scroll wide equations horizontally.

Regrouping is not the same operation as rearranging

Grouping consecutive terms into finite blocks preserves the limit of a convergent series, because the new partial sums form a subsequence of the old ones. Removing brackets from a convergent grouped expression is another matter: intermediate partial sums may misbehave. The grouped expression built from pairs one minus one has zero block sums, but its ungrouped partial sums alternate between one and zero. A convergent subsequence of partial sums does not establish convergence of all partial sums.

Likewise, finite algebra can suggest an identity involving a conditional series without establishing it. One must specify exactly how each truncated expression maps to another and show that boundary leftovers vanish. Absolute convergence often supplies the needed uniform bound, but conditional expressions may require a specially coordinated summation rule. This distinction is useful in Fourier manipulations and perturbative calculations, where formal rearrangement can accidentally substitute a new summation prescription for an old one.

P_N=\prod_{n=1}^{N}(1+u_n)>0,\qquad\log P_N=\sum_{n=1}^{N}\log(1+u_n)

Here all factors are positive. Finite logarithmic algebra is exact before any infinite-limit claim is made.↔ Scroll wide equations horizontally.

Products converge through accumulated relative change

For positive factors, convergence of the logarithm series to a finite real number is equivalent to partial products approaching a finite nonzero number. A product approaching zero is excluded by the usual nonzero definition of a convergent infinite product; stating that convention avoids ambiguity. Factors tending to one are necessary but insufficient. The factors one plus one over n approach one while their products grow like n, because small relative changes still accumulate.

If the absolute sum of uₙ is finite and every factor one plus uₙ is positive, eventually |uₙ|≤1/2 and the absolute logarithm is at most twice |uₙ|. Comparison proves logarithmic convergence and hence a positive product limit. The optimization lab illustrates a different multiplicative recurrence with one repeated factor. Predict growth or decay as that factor crosses unit magnitude; the lab does not display variable factors or rearrange a conditional product.

Work it through

A product with an exact boundary cancellation

For N≥2, calculate $\prod_{n=2}^{N}(1-1/n^2)$ and its limit.

Factor each positive term as $(n-1)(n+1)/n^2$ . Split the finite product into the product of (n−1)/n and the product of (n+1)/n.
The first finite product telescopes to 1/N. The second telescopes to (N+1)/2, because every intermediate numerator cancels the next denominator.
Multiply the remaining boundary factors to obtain (N+1)/(2N). This approaches 1/2, a finite nonzero value, so the infinite product converges under the stated convention.

The result

The partial product equals (N+1)/(2N), and the infinite product is 1/2. All cancellations were performed at finite N before taking a limit.

The living notebook

Explore the model

Interactive lab

Interactive lab · optimization

Repeated factors versus variable infinite products

For the quadratic update, xₙ=(1−η)ⁿx₀. Predict the magnitude change when |1−η| is below, equal to, or above one, then adjust η and step through. This is a repeated-factor example, not a variable-factor product convergence test or a rearrangement simulator.

ObjectiveDiscrete iterate

Position x0: 2
Objective x²/2: 2
Iteration behavior: Stable step

Step size η0.3Initial position x₀2

n = 0 · t = 0 s

Gradient descent on f(x) = x²/2: xₙ = (1 − η)ⁿx₀. Playback advances two discrete iterations per second. Values outside |x| ≤ 4 are explicitly reported. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forMultiplicative accumulation depends on the whole sequence of factors, not merely on their proximity to one.

Pause & reason

Check your understanding

Interactive checks need JavaScript. You can still read the explanation.

Read the checkpoint explanation

For uₙ=1/n, the factors approach one but the finite product from n=1 to N is N+1. The accumulated logarithmic changes diverge.

Your turn

Try it yourself

Prove that $\prod_{n=1}^{\infty}(1+1/n^2)$ has a finite nonzero limit without evaluating it.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Take logarithms and compare with a positive p-series.

Compare with the solution

Every factor is positive. For t≥0, the inequality $0\le\log(1+t)\le t$ follows by integrating 1/(1+s) from zero to t. Thus the logarithm partial sums are increasing and bounded by the convergent sum of 1/n². They approach a finite real L. Continuity of the exponential gives a product limit eᴸ, which is finite and at least one, hence nonzero.

Keep the idea

Take it with you

Absolute summability protects a scalar sum against permutations.
Consecutive grouping and arbitrary rearrangement are distinct limiting operations.
Positive product convergence to a nonzero limit is equivalent to finite logarithmic-sum convergence.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.