Stolz–CesΓ ro Theorem

When the full ratio is stubborn, this scene detours through the ratio of differences. It is the sequence version of l'HΓ΄pital's rule.

Compare the growth of single steps, and that ratio settles first. The full ratio follows slowly from behind. The denominator must increase strictly and grow without bound. Both ratios share the same destination. The headline example's limit is the second formula below.

For example, on a πŸš— car trip the speed over the last minute settles at 60 km/h. Then the average speed over the whole trip gets dragged toward 60 km/h as well.

lim⁑an+1βˆ’anbn+1βˆ’bn=Lβ€…β€ŠβŸΉβ€…β€Šlim⁑anbn=L\lim \frac{a_{n+1}-a_n}{b_{n+1}-b_n} = L \implies \lim \frac{a_n}{b_n} = L
lim⁑nβ†’βˆž12+β‹―+n2n3=13\lim_{n\to\infty}\frac{1^2+\cdots+n^2}{n^3} = \frac{1}{3}

Examples & Exercises

If anβ†’La_n \to L, prove with the Stolz–CesΓ ro theorem that the CesΓ ro mean a1+β‹―+ann\frac{a_1+\cdots+a_n}{n} also converges to LL (take bn=nb_n = n).

lim⁑nβ†’βˆža1+β‹―+ann=L\lim_{n\to\infty}\frac{a_1+\cdots+a_n}{n} = L
n=100n=100: in this scene's example the difference ratio r100β‰ˆ0.3367r_{100} \approx 0.3367 settles at 13\frac{1}{3} first and the full ratio s100β‰ˆ0.3384s_{100} \approx 0.3384 follows. The CesΓ ro mean is the special case bn=nb_n = n, where the difference ratio is an+1a_{n+1} itself.
Show proof
  1. Let cn=a1+β‹―+anc_n = a_1+\cdots+a_n and bn=nb_n = n. The sequence (bn)(b_n) is strictly increasing and unbounded, so it satisfies the hypotheses of the Stolz–CesΓ ro theorem.

  2. The difference ratio is cn+1βˆ’cnbn+1βˆ’bn=an+11=an+1\frac{c_{n+1}-c_n}{b_{n+1}-b_n} = \frac{a_{n+1}}{1} = a_{n+1}. A shift of a convergent sequence converges to the same limit: if anβ†’La_n \to L, then an NN for a given Ξ΅>0\varepsilon > 0 also works for an+1a_{n+1} (since nβ‰₯Nβ‡’n+1β‰₯Nn \ge N \Rightarrow n+1 \ge N, giving ∣an+1βˆ’L∣<Ξ΅|a_{n+1} - L| < \varepsilon).

    Hence an+1β†’La_{n+1} \to L, and the Stolz–CesΓ ro theorem gives cnbn=a1+β‹―+annβ†’L\frac{c_n}{b_n} = \frac{a_1+\cdots+a_n}{n} \to L.

Use the Stolz–CesΓ ro theorem to prove lim⁑nβ†’βˆž12+22+β‹―+n2n3=13\displaystyle \lim_{n\to\infty}\frac{1^2+2^2+\cdots+n^2}{n^3} = \frac{1}{3}.

lim⁑nβ†’βˆž12+22+β‹―+n2n3=13\lim_{n\to\infty}\frac{1^2+2^2+\cdots+n^2}{n^3} = \frac{1}{3}
n=200n=200: r200β‰ˆ0.3350r_{200} \approx 0.3350, s200β‰ˆ0.3358s_{200} \approx 0.3358 β€” both ratios have nearly reached 13\frac{1}{3}.
Show proof
  1. Let an=12+β‹―+n2a_n = 1^2+\cdots+n^2 and bn=n3b_n = n^3. The sequence (bn)(b_n) is strictly increasing for nβ‰₯1n \ge 1 and unbounded.

  2. Since an+1βˆ’an=(n+1)2a_{n+1}-a_n = (n+1)^2 and bn+1βˆ’bn=(n+1)3βˆ’n3=3n2+3n+1b_{n+1}-b_n = (n+1)^3-n^3 = 3n^2+3n+1, the difference ratio is (n+1)23n2+3n+1=1+2/n+1/n23+3/n+1/n2β†’13\frac{(n+1)^2}{3n^2+3n+1} = \frac{1+2/n+1/n^2}{3+3/n+1/n^2} \to \frac{1}{3}.

  3. By the Stolz–CesΓ ro theorem, anbn=12+β‹―+n2n3β†’13\frac{a_n}{b_n} = \frac{1^2+\cdots+n^2}{n^3} \to \frac{1}{3}.

1+3+5+β‹―+(2nβˆ’1)1+3+5+\cdots+(2n-1) is the sum of the first nn odd numbers.

Use the Stolz–CesΓ ro theorem to prove lim⁑nβ†’βˆž1+3+β‹―+(2nβˆ’1)n2=1\displaystyle \lim_{n\to\infty}\frac{1+3+\cdots+(2n-1)}{n^2} = 1.

βˆ‘k=1n(2kβˆ’1)=n2,lim⁑nβ†’βˆž1+3+β‹―+(2nβˆ’1)n2=1\sum_{k=1}^{n}(2k-1) = n^2, \qquad \lim_{n\to\infty}\frac{1+3+\cdots+(2n-1)}{n^2} = 1
n=60n=60: in this scene's example r60β‰ˆ0.3389r_{60} \approx 0.3389, s60β‰ˆ0.3417s_{60} \approx 0.3417 β€” the odd-sum problem runs on the same machinery (bn=n2b_n = n^2, difference ratio exactly 11).
Show proof
  1. Let an=1+3+β‹―+(2nβˆ’1)a_n = 1+3+\cdots+(2n-1); by induction an=n2a_n = n^2: for n=1n=1, 1=121 = 1^2, and if an=n2a_n = n^2 then an+1=n2+(2n+1)=(n+1)2a_{n+1} = n^2+(2n+1) = (n+1)^2. The sequence bn=n2b_n = n^2 is strictly increasing and unbounded.

  2. Since an+1βˆ’an=2n+1a_{n+1}-a_n = 2n+1 and bn+1βˆ’bn=(n+1)2βˆ’n2=2n+1b_{n+1}-b_n = (n+1)^2-n^2 = 2n+1, the difference ratio is 2n+12n+1=1\frac{2n+1}{2n+1} = 1 for every nn, hence its limit is 11.

  3. By the Stolz–CesΓ ro theorem, anbnβ†’1\frac{a_n}{b_n} \to 1 β€” in fact an=bna_n = b_n, so the ratio is exactly 11 for every nn.

3. Sequences