# Read infinite series in Book Isle: alternating partial sums approach a limit, but vanishing terms alone do not prove convergence

Language: en
Canonical: [Read infinite series in Book Isle: alternating partial sums approach a limit, but vanishing terms alone do not prove convergence](<https://allluckly.cn/en/guides/nasbook-infinite-geometric-series-negative-ratio-partial-sum-remainder/>)
中文：[中文 HTML](<https://allluckly.cn/guides/nasbook-infinite-geometric-series-negative-ratio-partial-sum-remainder/>)
English: [English HTML](<https://allluckly.cn/en/guides/nasbook-infinite-geometric-series-negative-ratio-partial-sum-remainder/>)
Published: 2026-11-19T00:00:00+08:00
Modified: 2026-11-19T10:10:36+08:00
Author: Bison

Check exact remainders for ratio negative one-half, with boundary-ratio and harmonic-series controls.

English translation of the [Chinese original](<https://allluckly.cn/guides/nasbook-infinite-geometric-series-negative-ratio-partial-sum-remainder/>), translated on 2026-11-19. Original publication dates, attribution, code and version notes are retained.

Shrinking terms alone do not guarantee a finite limiting sum. Prepare partial-sum cards and distinguish terms, first-N-term sums and infinite sum in Book Isle, then check alternating errors. Derivations remain external; the reader is not an automatic convergence tester.

## Fix zero-based terms and the first N terms

Set a\_k=(−1/2)^k for k=0, 1,… and S\_N=sum over k=0,…, N−1, with positive integer N. Put four terms/partial sums on P1, limit/remainder on P2 and divergent controls on P3; export PDF and retain source work. Check the negative sign inside the powered ratio and bookmark P2. The first term is one, and N terms do not run through k=N. Preserve indexing before deriving the error exponent.

## Keep individual terms separate from accumulated sums

The first terms are 1, −1/2, 1/4, −1/8; partial sums are 1, 1/2, 3/4, 5/8. Negative terms are not missing data and sums need not increase monotonically. S\_3 is 3/4, not the third term 1/4. Align the two ledgers and retain signs. A cropped positive-looking bar pattern must not silently replace the original alternating sequence.

Use an index card: S\_1 contains the k=0 term one, not negative one-half. S\_0 would be an empty sum zero if introduced, but the error task takes positive integer N. Separate first term from exponent one to avoid reversing every sign. Starting at k=1 defines another series, not a relabelled copy with unchanged values.

## Derive the finite formula before taking a limit

Shifting rS\_N with r=−1/2 gives S\_N=(1−r^N)/(1−r)=(2/3)\[1−(−1/2)^N\]. The denominator is 3/2, not 1/2. Since |r|\<1, r^N→0; the limit of partial sums defines S=2/3. There is no finite “last term” set to zero, and every finite N still has a nonzero remainder. Viewing several cards is not completing infinitely many additions.

```text
a_k=(-1/2)^k, k=0,1,...
S_N=sum k=0..N-1; S_N=(2/3)*(1-(-1/2)^N)
S_1=1; S_2=1/2; S_3=3/4; S_4=5/8
S=2/3; R_N=S-S_N=(2/3)*(-1/2)^N
S_6=21/32; abs(R_6)=1/96
S_7=43/64; abs(R_7)=1/192
```

## Separate signed remainder from absolute error

R\_N=S−S\_N=(2/3)(−1/2)^N, so |R\_N|=2/(3·2^N). Even N gives positive remainder and an underestimate; odd N gives an overestimate. N=6 yields 21/32 with remainder 1/96; N=7 gives 43/64 with remainder −1/192. Neither equals 2/3. An overestimate need not be farther away; compare absolute errors rather than signs alone.

Six terms give 0.65625, seven give 0.671875 and the limit is approximately 0.666666…. One lies below and the other above, neither equal to the sum. R\_N here means limit minus approximation; reversing subtraction preserves magnitude but reverses sign. Record the adopted definition before checking parity rather than assume opposite sign conventions are an arithmetic contradiction.

## Use a strict inequality for the one-hundredth target

For |R\_N|\<1/100, six terms fail because 1/96 is larger; seven pass with 1/192. Halving each successive error makes seven minimal. At two decimal places six terms display 0.66, seven display 0.67, and the limit also displays 0.67. Matching displays for seven terms and the limit do not make the remainder zero; compare both approximation errors exactly. Use exact fractions for the threshold and decimals only for reading. Changed first term, ratio or tolerance requires a new count; seven is not a universal geometric-series requirement.

## Do not extend the infinite formula past boundary ratios

With first term one, r=1 gives S\_N=N; r=−1 alternates partial sums 1, 0 with no limit. For r=2 positive partial sums grow, so calling 1/(1−2)=−1 their infinite sum violates convergence conditions. The finite identity remains available when its denominator 1−r is nonzero; taking an infinite limit is a separate step. Restore the |r|\<1 scope in notes, without altering values or making claims about other generalized summation conventions.

## Group a counterexample to terms-tending-to-zero sufficiency

The harmonic series 1+1/2+1/3+… has terms tending to zero but diverges. After 1, group 1/2, then 1/3 through 1/4, then 1/5 through 1/8, and so on. Each block doubles its count while every term is at least its last term, making every block sum at least 1/2. Arbitrarily many blocks give unbounded partial sums. This refutes sufficiency of vanishing terms, not the proved alternating geometric limit. Distinguish term and partial-sum limits.

Check harmonic block boundaries: the two-term block is 1/3+1/4, not a block omitting 1/3; the four-term block runs from 1/5 to 1/8. Block m has 2^(m−1) terms, each at least 1/2^m, for lower bound 1/2. Arbitrarily many such blocks exceed any finite bound; slow growth in a short plot is not convergence evidence.

## Reopen with partial-sum definitions and remainder controls

At reopened P2 recover the N-term range, finite formula and limit 2/3, then explain six failing and seven passing. Explain r=−1 and the harmonic counterexample through partial sums. Deliver term/sum ledgers, signed remainders and three boundary cards, marking unperformed import, highlighting and reopening pending. This is manual mathematics reading, not App divergence detection, an infinite computational loop or a proof of convergence for every series.

## Related tutorials and sources

[Read finite geometric sums in Book Isle: separate the fifth term from the five-term total](<https://allluckly.cn/en/guides/nasbook-finite-geometric-sum-last-term-limit/>); [Read sequence limits in Book Isle: choose an index guaranteeing error strictly below one hundredth](<https://allluckly.cn/en/guides/nasbook-sequence-limit-epsilon-strict-bound-index-threshold/>).

Method references: [OpenStax infinite series and partial sums](<https://openstax.org/books/calculus-volume-2/pages/5-2-infinite-series>). Diagrams, report data, counterexamples and ledgers are original exercises.

Version checked on 19 November 2026: the [Apple public lookup](<https://itunes.apple.com/lookup?id=6766198898&country=cn&entity=software&timestamp=20261119>) lists Book Isle 1.1.4, requiring iOS / iPadOS 15.6 or later; see the [official App Store](<https://apps.apple.com/cn/app/id6766198898>) for published capabilities. Use self-made PDFs, EPUBs or clear text materials; bookmarks, highlights and notes depend on format and visible controls, without automatic solving or literary analysis. External preparation and paper predictions do not constitute device observations. Leave unperformed checks blank.

