Practice / Calculus

Arithmetic and geometric sequences and sums

Ten problems on arithmetic and geometric sequences and their sums: a sequence from two of its terms, the sum of the multiples of 7 between 100 and 500, the first sum past 1000, a geometric sequence with a negative ratio, the derivation of the geometric sum formula, a repeating decimal as a fraction, sums in sigma notation with shifted indices, the weights of an exponential moving average and its bias correction, the sum of k xᵏ and the mean of a geometric distribution, and the monthly payment on a loan, with worked solutions and the mistakes of counting terms as upper minus lower, summing a geometric series with |r| ≥ 1, putting the wrong digits over 999, bias-correcting with 1 − β instead of 1 − βᵗ, and rounding the number of terms down.

Before you start

An arithmetic sequence adds the same number at every step and a geometric sequence multiplies by the same number, and between them they describe straight-line growth, compound interest, repeating decimals, the decay of an exponential moving average and the first sums anyone meets in calculus. These ten problems find a sequence from two of its terms, add up long stretches of each kind, derive the geometric sum formula rather than quote it, and then use it where it matters: a repeating decimal, the bias correction in the Adam optimizer, the mean of a geometric distribution and the payment on a loan. The five mistakes at the end are the ones that give a confident wrong number: a sum in sigma notation with one term missing, a divergent series given a finite sum, the wrong digits over 999999, a bias correction with the wrong power, and a number of terms rounded the wrong way.

  • Arithmetic. With first term a1a_1 and common difference dd, the nn-th term is an=a1+(n−1)da_n = a_1 + (n - 1)d, and the sum of the first nn terms is Sn=n(a1+an)2S_n = \dfrac{n(a_1 + a_n)}{2}: the number of terms times the average of the first and last.
  • Geometric. With first term a1a_1 and common ratio rr, an=a1rn−1a_n = a_1r^{n - 1}, and for r≠1r \ne 1, Sn=a1(1−rn)1−rS_n = \dfrac{a_1(1 - r^n)}{1 - r}. For r=1r = 1, Sn=na1S_n = na_1.
  • Infinite geometric series. For ∣r∣<1\lvert r\rvert < 1, rn→0r^n \to 0 and Sn→a11−rS_n \to \dfrac{a_1}{1 - r}. For ∣r∣≥1\lvert r\rvert \ge 1 the terms do not shrink to 00 and the series diverges (with a1≠0a_1 \ne 0).
  • Counting. The n−1n - 1 in ana_n counts steps, not terms: from a1a_1 to ana_n is n−1n - 1 steps. A sum ∑k=mM\sum_{k=m}^{M} has M−m+1M - m + 1 terms.
  • Sigma notation. ∑k=mMck\sum_{k=m}^{M} c_k means cm+cm+1+⋯+cMc_m + c_{m+1} + \dots + c_M. Substituting j=k−mj = k - m shifts the index to start at 00 without changing the sum.

Problems

  1. ·

    An arithmetic sequence has a5=23a_5 = 23 and a12=51a_{12} = 51. Find dd, a1a_1, a formula for ana_n, and a40a_{40}.

  2. ·

    Find the sum of all the multiples of 77 between 100100 and 500500.

  3. ··

    For the arithmetic sequence 3,7,11,…3, 7, 11, \dots, find SnS_n as a function of nn, and the smallest nn with Sn>1000S_n > 1000.

  4. ·

    A geometric sequence has a3=12a_3 = 12 and a6=−96a_6 = -96. Find rr, a1a_1 and S8S_8.

  5. ··

    Derive the formula Sn=a1(1−rn)1−rS_n = \dfrac{a_1(1 - r^n)}{1 - r} for r≠1r \ne 1. Use it to find 1+2+4+⋯+2631 + 2 + 4 + \dots + 2^{63}, the grains of rice on a chessboard that doubles from one grain on the first square.

  6. ··

    Write 2.145‾=2.1454545…2.1\overline{45} = 2.1454545\ldots as a fraction in lowest terms by summing a geometric series.

  7. ··

    Evaluate both sums, and write the second with its index starting at 00:

    ∑k=415(2k−5)and∑k=273⋅2k\sum_{k=4}^{15}(2k - 5) \qquad\text{and}\qquad \sum_{k=2}^{7} 3\cdot 2^k
  8. ···

    An exponential moving average starts at v0=0v_0 = 0 and updates vt=βvt−1+(1−β)gtv_t = \beta v_{t-1} + (1 - \beta)g_t. Show that vt=(1−β)∑k=1tβt−kgkv_t = (1 - \beta)\sum_{k=1}^{t}\beta^{t-k}g_k and that the weights add up to 1−βt1 - \beta^t. With β=0.9\beta = 0.9 and every gk=gg_k = g, find v10v_{10} and the bias-corrected v^10=v10/(1−β10)\hat v_{10} = v_{10}/(1 - \beta^{10}). How many steps until the weights add up to more than 0.990.99?

  9. ···

    Show that ∑k=1∞kxk=x(1−x)2\sum_{k=1}^{\infty} kx^k = \dfrac{x}{(1 - x)^2} for ∣x∣<1\lvert x\rvert < 1 by subtracting xTxT from TT. Then find ∑k=1∞k/2k\sum_{k=1}^\infty k/2^k, and the mean number of trials ∑k=1∞kp(1−p)k−1\sum_{k=1}^\infty kp(1 - p)^{k-1} until the first success when each succeeds with probability p=0.2p = 0.2.

  10. ···

    A loan of 20 00020\,000 dollars charges 0.5%0.5\% interest per month and is repaid in 6060 equal monthly payments, each made at the end of a month. Find the payment and the total interest.

Worked solutions

Problem 1

An arithmetic sequence has a5=23a_5 = 23 and a12=51a_{12} = 51. Find dd, a1a_1, a formula for ana_n, and a40a_{40}.

  1. 7d=51−23=287d = 51 - 23 = 28, so d=4d = 4From term 55 to term 1212 is 12−5=712 - 5 = 7 steps.
  2. a1=23−4⋅4=7a_1 = 23 - 4\cdot 4 = 7Walk back four steps from a5a_5.
  3. an=7+4(n−1)=4n+3a_n = 7 + 4(n - 1) = 4n + 3The nn-th term formula, simplified.
  4. a40=4⋅40+3=163a_{40} = 4\cdot 40 + 3 = 163Check the formula: a12=48+3=51a_{12} = 48 + 3 = 51.
  5. d=4d = 4, a1=7a_1 = 7, an=4n+3a_n = 4n + 3, a40=163a_{40} = 163An arithmetic sequence is a straight line sampled at whole numbers: an=4n+3a_n = 4n + 3 has slope d=4d = 4, and any two terms fix the line.

Problem 2

Find the sum of all the multiples of 77 between 100100 and 500500.

  1. First: 7⋅15=1057\cdot 15 = 105; last: 7⋅71=4977\cdot 71 = 497100/7≈14.3100/7 \approx 14.3 and 500/7≈71.4500/7 \approx 71.4, so the multiples run from 7⋅157\cdot 15 to 7⋅717\cdot 71.
  2. n=71−15+1=57n = 71 - 15 + 1 = 57 termsOne term for each of 15,16,…,7115, 16, \dots, 71.
  3. S57=57(105+497)2S_{57} = \dfrac{57(105 + 497)}{2}Arithmetic with d=7d = 7: number of terms times the average of first and last.
  4. =57⋅301= 57\cdot 301(105+497)/2=602/2=301(105 + 497)/2 = 602/2 = 301.
  5. 17 15717\,157This is Gauss's pairing: 105+497105 + 497, 112+490112 + 490, and so on all equal 602602. With an odd number of terms the middle one, 301=7⋅43301 = 7\cdot 43, pairs with itself, and the formula still holds.

Problem 3

For the arithmetic sequence 3,7,11,…3, 7, 11, \dots, find SnS_n as a function of nn, and the smallest nn with Sn>1000S_n > 1000.

  1. a1=3a_1 = 3, d=4d = 4, an=3+4(n−1)=4n−1a_n = 3 + 4(n - 1) = 4n - 1The common difference is 7−3=47 - 3 = 4.
  2. Sn=n(3+4n−1)2=n(2n+1)S_n = \dfrac{n(3 + 4n - 1)}{2} = n(2n + 1)First plus last is 4n+24n + 2; halve it.
  3. 2n2+n−1000=02n^2 + n - 1000 = 0 at n=−1+80014n = \dfrac{-1 + \sqrt{8001}}{4}The quadratic formula, positive root.
  4. n≈22.11n \approx 22.118001≈89.45\sqrt{8001} \approx 89.45.
  5. S22=22⋅45=990S_{22} = 22\cdot 45 = 990; S23=23⋅47=1081S_{23} = 23\cdot 47 = 1081SnS_n increases with nn, so check the whole numbers on each side of the root.
  6. Sn=n(2n+1)=2n2+nS_n = n(2n + 1) = 2n^2 + n; the smallest nn is 2323The sum of an arithmetic sequence is quadratic in nn, as the area under a line is quadratic in its length. The root 22.1122.11 must be rounded up: 2222 terms are not enough (Mistake 5).

Problem 4

A geometric sequence has a3=12a_3 = 12 and a6=−96a_6 = -96. Find rr, a1a_1 and S8S_8.

  1. r3=a6a3=−8r^3 = \dfrac{a_6}{a_3} = -8, so r=−2r = -2Three steps of the ratio separate term 33 from term 66; a cube root keeps the sign.
  2. a1=12(−2)2=3a_1 = \dfrac{12}{(-2)^2} = 3a3=a1r2a_3 = a_1r^2.
  3. S8=3(1−(−2)8)1−(−2)S_8 = \dfrac{3(1 - (-2)^8)}{1 - (-2)}The sum formula with n=8n = 8.
  4. =3(1−256)3=−255= \dfrac{3(1 - 256)}{3} = -255(−2)8=256(-2)^8 = 256, and 1−(−2)=31 - (-2) = 3.
  5. r=−2r = -2, a1=3a_1 = 3, S8=−255S_8 = -255Check by adding: 3−6+12−24+48−96+192−384=−2553 - 6 + 12 - 24 + 48 - 96 + 192 - 384 = -255. With an even number of terms and r<−1r < -1, the last, negative term is the largest in size, so the sum is negative.

Problem 5

Derive the formula Sn=a1(1−rn)1−rS_n = \dfrac{a_1(1 - r^n)}{1 - r} for r≠1r \ne 1. Use it to find 1+2+4+⋯+2631 + 2 + 4 + \dots + 2^{63}, the grains of rice on a chessboard that doubles from one grain on the first square.

  1. Sn=a1+a1r+⋯+a1rn−1S_n = a_1 + a_1r + \dots + a_1r^{n-1}The nn terms, from r0r^0 to rn−1r^{n-1}.
  2. rSn=a1r+a1r2+⋯+a1rnrS_n = a_1r + a_1r^2 + \dots + a_1r^nMultiply every term by rr: the list shifts one place.
  3. Sn−rSn=a1−a1rnS_n - rS_n = a_1 - a_1r^nEvery term but the first of SnS_n and the last of rSnrS_n cancels.
  4. Sn=a1(1−rn)1−rS_n = \dfrac{a_1(1 - r^n)}{1 - r}Factor out SnS_n and divide by 1−r1 - r, which is allowed because r≠1r \ne 1.
  5. Chessboard: a1=1a_1 = 1, r=2r = 2, n=64n = 64, so S64=1−2641−2=264−1S_{64} = \dfrac{1 - 2^{64}}{1 - 2} = 2^{64} - 16464 squares hold 202^0 to 2632^{63}.
  6. Sn=a1(1−rn)1−rS_n = \dfrac{a_1(1 - r^n)}{1 - r}; the board holds 264−1≈1.845×10192^{64} - 1 \approx 1.845\times 10^{19} grainsThe same sum is the largest unsigned 6464-bit integer, 18 446 744 073 709 551 61518\,446\,744\,073\,709\,551\,615: in binary it is sixty-four 11s. For r>1r > 1 the form a1(rn−1)/(r−1)a_1(r^n - 1)/(r - 1) keeps both parts positive.

Problem 6

Write 2.145‾=2.1454545…2.1\overline{45} = 2.1454545\ldots as a fraction in lowest terms by summing a geometric series.

  1. 2.145‾=2.1+0.045+0.00045+…2.1\overline{45} = 2.1 + 0.045 + 0.00045 + \dotsSplit off the part that does not repeat; each repeat of 4545 is two places further right.
  2. The tail is geometric with a1=0.045a_1 = 0.045 and r=0.01r = 0.01Each term is 1100\tfrac{1}{100} of the one before.
  3. Tail =0.0451−0.01=0.0450.99= \dfrac{0.045}{1 - 0.01} = \dfrac{0.045}{0.99}∣r∣<1\lvert r\rvert < 1, so the infinite sum is a1/(1−r)a_1/(1 - r).
  4. =45990=122= \dfrac{45}{990} = \dfrac{1}{22}Multiply top and bottom by 10001000, then cancel 4545.
  5. 2110+122=231+5110=236110\dfrac{21}{10} + \dfrac{1}{22} = \dfrac{231 + 5}{110} = \dfrac{236}{110}Common denominator 110110.
  6. 11855\dfrac{118}{55}Check: 118/55=2.14545…118/55 = 2.14545\ldots. Every repeating decimal is a geometric series with ratio 10−p10^{-p} for a block of pp digits, which is why every repeating decimal is a fraction.

Problem 7

Evaluate both sums, and write the second with its index starting at 00:

∑k=415(2k−5)and∑k=273⋅2k\sum_{k=4}^{15}(2k - 5) \qquad\text{and}\qquad \sum_{k=2}^{7} 3\cdot 2^k
  1. First sum: 15−4+1=1215 - 4 + 1 = 12 terms, from 2⋅4−5=32\cdot 4 - 5 = 3 to 2⋅15−5=252\cdot 15 - 5 = 25The terms go up by 22 each time: arithmetic with d=2d = 2.
  2. ∑k=415(2k−5)=12(3+25)2=168\sum_{k=4}^{15}(2k - 5) = \dfrac{12(3 + 25)}{2} = 168Number of terms times the average of first and last.
  3. Second sum: j=k−2j = k - 2 runs from 00 to 55, and 3⋅2k=3⋅2j+2=12⋅2j3\cdot 2^k = 3\cdot 2^{j+2} = 12\cdot 2^jSubstituting k=j+2k = j + 2 moves the shift into the coefficient.
  4. ∑j=0512⋅2j=12(26−1)2−1\sum_{j=0}^{5} 12\cdot 2^j = \dfrac{12(2^6 - 1)}{2 - 1}Geometric with a1=12a_1 = 12, r=2r = 2, and 66 terms.
  5. =12⋅63=756= 12\cdot 63 = 756Check: 12+24+48+96+192+384=75612 + 24 + 48 + 96 + 192 + 384 = 756.
  6. 168168, and ∑j=0512⋅2j=756\sum_{j=0}^{5} 12\cdot 2^j = 756Before applying a formula, read off three things from the sigma: the first term, the ratio or difference, and the number of terms, upper minus lower plus one.

Problem 8

An exponential moving average starts at v0=0v_0 = 0 and updates vt=βvt−1+(1−β)gtv_t = \beta v_{t-1} + (1 - \beta)g_t. Show that vt=(1−β)∑k=1tβt−kgkv_t = (1 - \beta)\sum_{k=1}^{t}\beta^{t-k}g_k and that the weights add up to 1−βt1 - \beta^t. With β=0.9\beta = 0.9 and every gk=gg_k = g, find v10v_{10} and the bias-corrected v^10=v10/(1−β10)\hat v_{10} = v_{10}/(1 - \beta^{10}). How many steps until the weights add up to more than 0.990.99?

  1. v1=(1−β)g1v_1 = (1 - \beta)g_1, v2=(1−β)(βg1+g2)v_2 = (1 - \beta)(\beta g_1 + g_2), and so onUnroll the recursion from v0=0v_0 = 0: each step multiplies the old terms by β\beta and adds a new one.
  2. vt=(1−β)∑k=1tβt−kgkv_t = (1 - \beta)\sum_{k=1}^{t}\beta^{t-k}g_kBy induction: βvt−1\beta v_{t-1} raises every power by one, and (1−β)gt(1 - \beta)g_t is the k=tk = t term with β0\beta^0.
  3. Weights: (1−β)(1+β+⋯+βt−1)(1 - \beta)(1 + \beta + \dots + \beta^{t-1})The powers βt−k\beta^{t-k} for k=t,t−1,…,1k = t, t-1, \dots, 1, a geometric sum of tt terms.
  4. =(1−β)⋅1−βt1−β=1−βt= (1 - \beta)\cdot\dfrac{1 - \beta^t}{1 - \beta} = 1 - \beta^tProblem 5's formula with a1=1a_1 = 1 and r=βr = \beta.
  5. v10=(1−0.910)g≈0.6513 gv_{10} = (1 - 0.9^{10})g \approx 0.6513\,gWith a constant input, vtv_t is the total weight times gg; 0.910≈0.34870.9^{10} \approx 0.3487.
  6. v^10=0.6513 g0.6513=g\hat v_{10} = \dfrac{0.6513\,g}{0.6513} = gDividing by the total weight makes the weights add up to 11.
  7. 0.9t<0.010.9^t < 0.01 when t>ln⁡0.01ln⁡0.9≈43.7t > \dfrac{\ln 0.01}{\ln 0.9} \approx 43.7Take logs; ln⁡0.9<0\ln 0.9 < 0 flips the inequality.
  8. v10≈0.651 gv_{10} \approx 0.651\,g, v^10=g\hat v_{10} = g; the weights pass 0.990.99 at t=44t = 44Starting from 00 biases the average toward 00 by the factor 1−βt1 - \beta^t, and Adam's bias correction divides it out. By t=44t = 44 the factor is 1−0.944≈0.99031 - 0.9^{44} \approx 0.9903 and the correction barely matters; with β=0.999\beta = 0.999, as for Adam's second moment, it takes about 46004600 steps.

Problem 9

Show that ∑k=1∞kxk=x(1−x)2\sum_{k=1}^{\infty} kx^k = \dfrac{x}{(1 - x)^2} for ∣x∣<1\lvert x\rvert < 1 by subtracting xTxT from TT. Then find ∑k=1∞k/2k\sum_{k=1}^\infty k/2^k, and the mean number of trials ∑k=1∞kp(1−p)k−1\sum_{k=1}^\infty kp(1 - p)^{k-1} until the first success when each succeeds with probability p=0.2p = 0.2.

  1. T=x+2x2+3x3+…T = x + 2x^2 + 3x^3 + \dots and xT=x2+2x3+…xT = x^2 + 2x^3 + \dotsMultiply by xx: each coefficient moves one power up.
  2. T−xT=x+x2+x3+…T - xT = x + x^2 + x^3 + \dotsCoefficients kk and k−1k - 1 meet at each power xkx^k, leaving 11.
  3. =x1−x= \dfrac{x}{1 - x}A geometric series with a1=xa_1 = x, r=xr = x and ∣x∣<1\lvert x\rvert < 1.
  4. T=x(1−x)2T = \dfrac{x}{(1 - x)^2}Divide by 1−x1 - x. The series converges for ∣x∣<1\lvert x\rvert < 1, since the ratio of consecutive terms, k+1k∣x∣\tfrac{k+1}{k}\lvert x\rvert, tends to ∣x∣<1\lvert x\rvert < 1; so TT is a finite number and the subtraction is allowed.
  5. x=12x = \tfrac12: T=1/21/4=2T = \dfrac{1/2}{1/4} = 2Partial sums: 0.50.5, 11, 1.3751.375, 1.625,…1.625, \dots
  6. Mean: p1−p∑kk(1−p)k=p1−p⋅1−pp2=1p\dfrac{p}{1 - p}\sum_{k} k(1 - p)^k = \dfrac{p}{1 - p}\cdot\dfrac{1 - p}{p^2} = \dfrac1pTake one factor 1−p1 - p outside, then step 4 with x=1−px = 1 - p, 1−x=p1 - x = p.
  7. ∑kxk=x(1−x)2\sum kx^k = \dfrac{x}{(1 - x)^2}; ∑k/2k=2\sum k/2^k = 2; the mean is 1/p=51/p = 5 trialsA sum whose terms are an arithmetic sequence times a geometric one is called arithmetico-geometric, and the subtraction that proves the geometric formula reduces it to a geometric sum. The same answer comes from differentiating ∑xk=1/(1−x)\sum x^k = 1/(1 - x) and multiplying by xx.

Problem 10

A loan of 20 00020\,000 dollars charges 0.5%0.5\% interest per month and is repaid in 6060 equal monthly payments, each made at the end of a month. Find the payment and the total interest.

  1. After nn months the debt has grown to L(1+i)nL(1 + i)^n with L=20 000L = 20\,000, i=0.005i = 0.005Each month multiplies the balance by 1+i1 + i.
  2. Payment kk has grown to P(1+i)60−kP(1 + i)^{60 - k} by month 6060It is made at the end of month kk and earns interest for the remaining months.
  3. P∑j=059(1+i)j=P⋅(1+i)60−1iP\sum_{j=0}^{59}(1 + i)^j = P\cdot\dfrac{(1 + i)^{60} - 1}{i}A geometric sum with a1=1a_1 = 1, r=1+ir = 1 + i and 6060 terms.
  4. L(1+i)60=P⋅(1+i)60−1iL(1 + i)^{60} = P\cdot\dfrac{(1 + i)^{60} - 1}{i}The loan is repaid when the payments, grown to month 6060, equal the grown debt.
  5. P=Li1−(1+i)−60=1001−1.005−60P = \dfrac{Li}{1 - (1 + i)^{-60}} = \dfrac{100}{1 - 1.005^{-60}}Divide both sides by (1+i)60(1 + i)^{60}; Li=100Li = 100.
  6. 1.00560≈1.348851.005^{60} \approx 1.34885, so 1−1.005−60≈0.258631 - 1.005^{-60} \approx 0.25863One power, then its reciprocal.
  7. Payment ≈386.66\approx 386.66 dollars; total paid ≈23 199.36\approx 23\,199.36, so interest ≈3 199.36\approx 3\,199.36 dollars60×386.656≈23 199.3660\times 386.656 \approx 23\,199.36. Simple interest on the full 20 00020\,000 for five years would be 6 0006\,000; it is about half that because the balance falls as it is repaid.

Where this goes wrong

1. Counting the terms as upper minus lower

A sum ∑k=415\sum_{k=4}^{15} looks like it covers a stretch of length 15−415 - 4.

  1. ∑k=415(2k−5)\sum_{k=4}^{15}(2k - 5) runs from 33 to 2525Right so far: Problem 7, step 1.
  2. “There are 15−4=1115 - 4 = 11 terms.”The subtraction counts the steps between k=4k = 4 and k=15k = 15, and there is one more term than there are steps.
  3. S=11(3+25)2=154S = \dfrac{11(3 + 25)}{2} = 154The sum has 15−4+1=1215 - 4 + 1 = 12 terms and equals 168168. A quick check is to count a short case by hand: ∑k=45\sum_{k=4}^{5} has two terms, not one.

2. Summing a geometric series with |r| ≥ 1

The formula a1/(1−r)a_1/(1 - r) gives a number for any r≠1r \ne 1, and nothing in the arithmetic stops at the condition.

  1. ∑k=0∞(32)k\sum_{k=0}^{\infty}\big(\tfrac32\big)^k is geometric with a1=1a_1 = 1 and r=32r = \tfrac32Right so far.
  2. “The infinite sum is first term over 1−r1 - r.”That holds only when ∣r∣<1\lvert r\rvert < 1, because the formula is the limit of a1(1−rn)/(1−r)a_1(1 - r^n)/(1 - r) and needs rn→0r^n \to 0.
  3. ∑k=0∞(32)k=11−3/2=−2\sum_{k=0}^{\infty}\big(\tfrac32\big)^k = \dfrac{1}{1 - 3/2} = -2A sum of positive terms cannot be negative. Here rnr^n grows without bound, the partial sums 1,2.5,4.75,…1, 2.5, 4.75, \dots do too, and the series diverges.

3. Putting the wrong digits over 999

The rule that a repeating block of pp digits goes over pp nines is remembered without the condition that the block starts right after the decimal point.

  1. 2.145‾=2.1454545…2.1\overline{45} = 2.1454545\ldotsRight so far: Problem 6.
  2. “The digits after the point are 145145, three of them, so the fraction is 145/999145/999.”Only 4545 repeats; the 11 is written once. Putting it inside the block repeats it too.
  3. 2.145‾=2+1459992.1\overline{45} = 2 + \dfrac{145}{999}That is 2.145145…2.145145\ldots, not 2.14545…2.14545\ldots. Split off the part that does not repeat first: 2.1+45990=118552.1 + \dfrac{45}{990} = \dfrac{118}{55}.

4. Bias correction with 1 − β instead of 1 − βᵗ

The update has a 1−β1 - \beta in it, and dividing by the same factor looks like the way to undo it.

  1. v10=(1−0.910)g≈0.6513 gv_{10} = (1 - 0.9^{10})g \approx 0.6513\,gRight so far: Problem 8, step 5.
  2. “The average is scaled by 1−β1 - \beta, so divide by 1−β1 - \beta.”The 1−β1 - \beta in the update is already balanced by the sum of the powers of β\beta; what is missing is the weight on the steps before t=1t = 1, which is βt\beta^t.
  3. v^10=v101−0.9≈6.51 g\hat v_{10} = \dfrac{v_{10}}{1 - 0.9} \approx 6.51\,gThat is more than six times the true average. The total weight is 1−βt1 - \beta^t (Problem 8, step 4), and dividing by it gives v^10=g\hat v_{10} = g. With 1−β1 - \beta the estimate tends to 10g10g instead of gg as tt grows.

5. Rounding the number of terms down

The quadratic formula gives a decimal, and rounding to the nearest whole number is the usual reflex.

  1. 2n2+n=10002n^2 + n = 1000 at n≈22.11n \approx 22.11Right so far: Problem 3, step 4.
  2. “Round 22.1122.11 to 2222.”The question asks for Sn>1000S_n > 1000, and SnS_n increases, so nn must be at least the root.
  3. n=22n = 22S22=990S_{22} = 990, still short of 10001000. The smallest nn is 2323, with S23=1081S_{23} = 1081. When the answer must exceed a threshold, round up and check both neighbors.

Print this set: arithmetic-and-geometric-sequences.pdf (problems, answers, and worked solutions on separate pages).

Read the guide: Arithmetic and geometric sequences: the nth term, the sum, and the off-by-one in n