E078 Section 7 Notes – Power Series

Source: Condensed from the Section 7 Study Guide (zyBooks 7.1–7.4). Not official WGU material.

Big idea. A power series ∑cn(x−a)n\sum c_n (x-a)^n is an "infinite polynomial": inside its interval of convergence it is a real function you can add, multiply, differentiate, and integrate term by term. 7.1 asks where it converges (ratio test plus endpoints); 7.2 builds new series from 11−x\frac{1}{1-x}; 7.3 gives the Taylor formula cn=f(n)(a)n!c_n = \frac{f^{(n)}(a)}{n!} and the error bound; 7.4 adds the binomial series, the table of known series, and series solutions of differential equations and integrals.

7.1 Power Series and Functions

Big idea

  • Power series centered at aa: ∑n=0∞cn(x−a)n=c0+c1(x−a)+c2(x−a)2+⋯\sum_{n=0}^{\infty} c_n (x-a)^n = c_0 + c_1(x-a) + c_2(x-a)^2 + \cdots (convention: (x−a)0=1(x-a)^0 = 1 even at x=ax = a).

  • It always converges at its center x=ax = a (the value is c0c_0).

  • Exactly one happens: (i) converges only at x=ax = a (R=0R = 0); (ii) converges for all xx (R=∞R = \infty); (iii) some R>0R > 0 with convergence for ∣x−a∣<R|x-a| < R and divergence for ∣x−a∣>R|x-a| > R.

  • Interval of convergence = every xx where it converges; radius RR = half its length. At the endpoints x=a±Rx = a \pm R anything can happen: test each separately.

Formulas

  • Recipe for the interval: (1) ratio test ρ=lim⁡n→∞∣an+1an∣\rho = \lim_{n\to\infty} \left|\frac{a_{n+1}}{a_n}\right| with the xx's still in it; (2) solve ρ<1\rho < 1 for xx (if ρ=0\rho = 0 for all xx, R=∞R = \infty; if ρ=∞\rho = \infty for all x≠ax \ne a, R=0R = 0); (3) plug each endpoint into the series and test it (harmonic, alternating series test, pp-series, divergence test).

  • Geometric series as a function: 11−x=∑n=0∞xn=1+x+x2+⋯\frac{1}{1-x} = \sum_{n=0}^{\infty} x^n = 1 + x + x^2 + \cdots for ∣x∣<1|x| < 1.

  • More generally a1−r=∑arn\frac{a}{1-r} = \sum a r^n for ∣r∣<1|r| < 1: rewrite your function so it looks like this.

Worked example (finite radius): ∑n=1∞(x−3)nn 2n\sum_{n=1}^{\infty} \frac{(x-3)^n}{n\,2^n}.

  • Ratio test: ρ=lim⁡nn+1⋅∣x−3∣2=∣x−3∣2<1  ⟺  1<x<5\rho = \lim \frac{n}{n+1} \cdot \frac{|x-3|}{2} = \frac{|x-3|}{2} < 1 \iff 1 < x < 5, so R=2R = 2.

  • Endpoints: x=1x = 1 gives ∑(−1)nn\sum \frac{(-1)^n}{n} (alternating harmonic, converges); x=5x = 5 gives ∑1n\sum \frac{1}{n} (diverges). Interval [1,5)[1, 5).

  • Also: ∑xnn!\sum \frac{x^n}{n!} has ρ=lim⁡∣x∣n+1=0\rho = \lim \frac{|x|}{n+1} = 0, so R=∞R = \infty; ∑n! xn\sum n!\,x^n has ρ=lim⁡(n+1)∣x∣=∞\rho = \lim (n+1)|x| = \infty for x≠0x \ne 0, so R=0R = 0.

  • Geometric representation: x5+x=x5⋅11−(−x/5)=∑n=0∞(−1)nxn+15n+1\frac{x}{5+x} = \frac{x}{5} \cdot \frac{1}{1 - (-x/5)} = \sum_{n=0}^{\infty} \frac{(-1)^n x^{n+1}}{5^{n+1}}, needs ∣x∣<5|x| < 5; interval (−5,5)(-5, 5) (geometric series never converge at ∣r∣=1|r| = 1).

Common mistakes

  • Forgetting to check the endpoints. The ratio test says nothing when ρ=1\rho = 1.

  • Writing the radius as the whole interval length. RR is half the length.

  • For a geometric-type series, "testing" the endpoints anyway: at ∣r∣=1|r| = 1 a geometric series always diverges.

7.2 Properties of Power Series

Big idea

  • If ∑cnxn=f(x)\sum c_n x^n = f(x) and ∑dnxn=g(x)\sum d_n x^n = g(x) on a common interval II: (i) ∑(cn±dn)xn=f±g\sum (c_n \pm d_n)x^n = f \pm g on II; (ii) ∑bxmcnxn=bxmf(x)\sum b x^m c_n x^n = b x^m f(x) on II; (iii) ∑cn(bxm)n=f(bxm)\sum c_n (b x^m)^n = f(b x^m) whenever bxm∈Ib x^m \in I. Different intervals: the result is good on the smaller (common) one.

  • Multiplying (Cauchy product): en=c0dn+c1dn−1+⋯+cnd0e_n = c_0 d_n + c_1 d_{n-1} + \cdots + c_n d_0, converging to f⋅gf \cdot g on the common interval.

  • Term-by-term calculus: if f=∑cn(x−a)nf = \sum c_n (x-a)^n for ∣x−a∣<R|x-a| < R, then f′=∑ncn(x−a)n−1f' = \sum n c_n (x-a)^{n-1} and ∫f dx=C+∑cn(x−a)n+1n+1\int f\,dx = C + \sum c_n \frac{(x-a)^{n+1}}{n+1}, same RR, but endpoint behavior can change.

  • Uniqueness: if two power series are equal on an open interval around aa, their coefficients are equal.

Formulas

  • 1(1−x)2=∑n=0∞(n+1)xn\frac{1}{(1-x)^2} = \sum_{n=0}^{\infty} (n+1)x^n for ∣x∣<1|x| < 1

  • ln⁡(1+x)=∑n=1∞(−1)n+1xnn\ln(1+x) = \sum_{n=1}^{\infty} (-1)^{n+1}\frac{x^n}{n} on (−1,1](-1, 1]

  • tan⁡−1x=∑n=0∞(−1)nx2n+12n+1\tan^{-1} x = \sum_{n=0}^{\infty} (-1)^n \frac{x^{2n+1}}{2n+1} on [−1,1][-1, 1]

  • Present value (annual payment CC, interest rate rr), nn years: P=C1+r+C(1+r)2+⋯+C(1+r)n=Cr(1−1(1+r)n)P = \frac{C}{1+r} + \frac{C}{(1+r)^2} + \cdots + \frac{C}{(1+r)^n} = \frac{C}{r}\left(1 - \frac{1}{(1+r)^n}\right)

  • Present value forever: P=CrP = \frac{C}{r}

Worked example (partial fractions, then two geometric series): f(x)=1(x−2)(x−4)=1/2x−4−1/2x−2=1/41−x/2−1/81−x/4f(x) = \frac{1}{(x-2)(x-4)} = \frac{1/2}{x-4} - \frac{1/2}{x-2} = \frac{1/4}{1 - x/2} - \frac{1/8}{1 - x/4}.

  • So f(x)=∑n=0∞(12n+2−12⋅4n+1)xnf(x) = \sum_{n=0}^{\infty} \left(\frac{1}{2^{n+2}} - \frac{1}{2 \cdot 4^{n+1}}\right) x^n; the pieces need ∣x∣<2|x| < 2 and ∣x∣<4|x| < 4, keep the smaller: interval (−2,2)(-2, 2). Check: n=0n = 0 gives 18=f(0)\frac{1}{8} = f(0).

  • New interval from an old one: if ∑anxn\sum a_n x^n works on (−1,1)(-1, 1), then ∑an(2x)n\sum a_n (2x)^n needs 2x∈(−1,1)2x \in (-1, 1), so (−12,12)\left(-\frac{1}{2}, \frac{1}{2}\right).

  • Differentiate to sum: 1(1−x)2=∑(n+1)xn\frac{1}{(1-x)^2} = \sum (n+1)x^n, so at x=13x = \frac{1}{3}, ∑n=0∞n+13n=1(1−1/3)2=94\sum_{n=0}^{\infty} \frac{n+1}{3^n} = \frac{1}{(1 - 1/3)^2} = \frac{9}{4}.

  • Integrate 11+x=1−x+x2−⋯\frac{1}{1+x} = 1 - x + x^2 - \cdots to get ln⁡(1+x)=C+x−x22+x33−⋯\ln(1+x) = C + x - \frac{x^2}{2} + \frac{x^3}{3} - \cdots; x=0x = 0 gives C=0C = 0. Endpoints: x=1x = 1 converges (alternating harmonic), x=−1x = -1 diverges: interval (−1,1](-1, 1].

  • Present value: 2 million dollars per year for 10 years at r=4%r = 4\%: P=20.04(1−1.04−10)≈16.222P = \frac{2}{0.04}\left(1 - 1.04^{-10}\right) \approx 16.222 million; forever, P=20.04=50P = \frac{2}{0.04} = 50 million.

Common mistakes

  • Assuming the differentiated or integrated series has the same interval. Same radius, yes; endpoints must be re-checked.

  • Forgetting the +C+C when integrating, then not solving for it with x=ax = a.

  • Starting the derivative series at n=0n = 0: the constant term dies, so it starts at n=1n = 1.

7.3 Taylor and Maclaurin Series

Big idea

  • If ff has a power series at aa, its coefficients must be cn=f(n)(a)n!c_n = \frac{f^{(n)}(a)}{n!} (uniqueness).

  • Taylor series at aa: ∑n=0∞f(n)(a)n!(x−a)n\sum_{n=0}^{\infty} \frac{f^{(n)}(a)}{n!}(x-a)^n. Maclaurin series: the Taylor series at a=0a = 0.

  • Taylor polynomial pnp_n = the nnth partial sum (stop at (x−a)n(x-a)^n). It approximates ff near aa.

  • Remainder Rn(x)=f(x)−pn(x)R_n(x) = f(x) - p_n(x). The series converges to ff exactly when Rn(x)→0R_n(x) \to 0.

Formulas

  • pn(x)=f(a)+f′(a)(x−a)+f′′(a)2!(x−a)2+⋯+f(n)(a)n!(x−a)np_n(x) = f(a) + f'(a)(x-a) + \frac{f''(a)}{2!}(x-a)^2 + \cdots + \frac{f^{(n)}(a)}{n!}(x-a)^n

  • Taylor's theorem with remainder: Rn(x)=f(n+1)(c)(n+1)!(x−a)n+1R_n(x) = \frac{f^{(n+1)}(c)}{(n+1)!}(x-a)^{n+1} for some cc between aa and xx.

  • Error bound: if ∣f(n+1)∣≤M|f^{(n+1)}| \le M between aa and xx, then ∣Rn(x)∣≤M(n+1)!∣x−a∣n+1|R_n(x)| \le \frac{M}{(n+1)!}|x-a|^{n+1}.

  • ex=∑xnn!e^x = \sum \frac{x^n}{n!}, sin⁡x=∑(−1)nx2n+1(2n+1)!\sin x = \sum \frac{(-1)^n x^{2n+1}}{(2n+1)!}, cos⁡x=∑(−1)nx2n(2n)!\cos x = \sum \frac{(-1)^n x^{2n}}{(2n)!} (all xx).

Worked example (Taylor polynomial, estimate, error bound): f(x)=xf(x) = \sqrt{x} at a=4a = 4.

  • f(4)=2f(4) = 2, f′(4)=14f'(4) = \frac{1}{4}, f′′(4)=−132f''(4) = -\frac{1}{32}, so p2(x)=2+14(x−4)−164(x−4)2p_2(x) = 2 + \frac{1}{4}(x-4) - \frac{1}{64}(x-4)^2 and 4.2≈p2(4.2)=2.049375\sqrt{4.2} \approx p_2(4.2) = 2.049375.

  • Bound: f′′′(x)=38x−5/2f'''(x) = \frac{3}{8}x^{-5/2} is largest on [4,4.2][4, 4.2] at x=4x = 4, so M=3256M = \frac{3}{256} and ∣R2∣≤3/2563!(0.2)3=1.5625×10−5|R_2| \le \frac{3/256}{3!}(0.2)^3 = 1.5625 \times 10^{-5}.

  • Taylor series with interval: f(x)=1xf(x) = \frac{1}{x} at a=2a = 2 has f(n)(x)=(−1)nn! x−(n+1)f^{(n)}(x) = (-1)^n n!\,x^{-(n+1)}, so cn=(−1)n2n+1c_n = \frac{(-1)^n}{2^{n+1}} and 1x=∑n=0∞(−1)n(x−2)n2n+1\frac{1}{x} = \sum_{n=0}^{\infty} \frac{(-1)^n (x-2)^n}{2^{n+1}}; ratio test gives 0<x<40 < x < 4, and at both endpoints the terms have size 12\frac{1}{2} (diverge): interval (0,4)(0, 4).

  • How many terms: to get e0.5e^{0.5} within 0.0010.001, every derivative is ex<2e^x < 2 on [0,0.5][0, 0.5], so ∣Rn∣≤2(0.5)n+1(n+1)!|R_n| \le \frac{2(0.5)^{n+1}}{(n+1)!}; n=3n = 3 gives 0.00520.0052 (too big), n=4n = 4 gives ≈0.00052<0.001\approx 0.00052 < 0.001, so n=4n = 4.

Common mistakes

  • Forgetting the n!n! in cn=f(n)(a)/n!c_n = f^{(n)}(a)/n!.

  • Plugging aa into f(n+1)f^{(n+1)} without checking it is the largest value on the interval. Pick MM as the max of ∣f(n+1)∣|f^{(n+1)}| between aa and xx.

  • For sin⁡x\sin x, p3=p4p_3 = p_4 (the x4x^4 coefficient is 00), so you can use the sharper bound with R4R_4.

7.4 Working with Taylor Series

Big idea

  • Binomial series (any real rr): (1+x)r=∑(rn)xn(1+x)^r = \sum \binom{r}{n} x^n for ∣x∣<1|x| < 1, where (rn)=r(r−1)⋯(r−n+1)n!\binom{r}{n} = \frac{r(r-1)\cdots(r-n+1)}{n!}. If rr is a whole number it stops (ordinary binomial expansion).

  • Get new series from the table by substituting, multiplying by xmx^m, differentiating, or integrating, instead of computing derivatives.

  • Solving differential equations: assume y=∑cnxny = \sum c_n x^n, plug in, match coefficients (uniqueness), use the initial conditions for c0,c1c_0, c_1.

  • Nonelementary integrals (like ∫e−x2 dx\int e^{-x^2}\,dx): expand the integrand, integrate term by term, and bound the error with the alternating series test.

Formulas

  • Table: 11−x=∑xn\frac{1}{1-x} = \sum x^n on (−1,1)(-1, 1); ex=∑xnn!e^x = \sum \frac{x^n}{n!}, sin⁡x=∑(−1)nx2n+1(2n+1)!\sin x = \sum \frac{(-1)^n x^{2n+1}}{(2n+1)!}, cos⁡x=∑(−1)nx2n(2n)!\cos x = \sum \frac{(-1)^n x^{2n}}{(2n)!} for all xx.

  • ln⁡(1+x)=∑(−1)n+1xnn\ln(1+x) = \sum \frac{(-1)^{n+1} x^n}{n} on (−1,1](-1, 1]; tan⁡−1x=∑(−1)nx2n+12n+1\tan^{-1} x = \sum \frac{(-1)^n x^{2n+1}}{2n+1} on [−1,1][-1, 1]; (1+x)r=∑(rn)xn(1+x)^r = \sum \binom{r}{n} x^n on (−1,1)(-1, 1).

  • Pendulum period: T=4Lg∫0π/2dθ1−k2sin⁡2θT = 4\sqrt{\frac{L}{g}} \int_0^{\pi/2} \frac{d\theta}{\sqrt{1 - k^2 \sin^2\theta}}, k=sin⁡θmax⁡2k = \sin\frac{\theta_{\max}}{2}. First term: T≈2πLgT \approx 2\pi\sqrt{\frac{L}{g}}; two terms: T≈2πLg(1+k24)T \approx 2\pi\sqrt{\frac{L}{g}}\left(1 + \frac{k^2}{4}\right).

Worked example (binomial series and an estimate): with r=13r = \frac{1}{3}, (1+x)1/3=1+x3+13(−23)2!x2+13(−23)(−53)3!x3+⋯=1+x3−x29+5x381−⋯(1+x)^{1/3} = 1 + \frac{x}{3} + \frac{\frac{1}{3}(-\frac{2}{3})}{2!}x^2 + \frac{\frac{1}{3}(-\frac{2}{3})(-\frac{5}{3})}{3!}x^3 + \cdots = 1 + \frac{x}{3} - \frac{x^2}{9} + \frac{5x^3}{81} - \cdots, so 1.13≈1+0.13−0.019≈1.03222\sqrt[3]{1.1} \approx 1 + \frac{0.1}{3} - \frac{0.01}{9} \approx 1.03222 (true value 1.032281.03228).

  • From the table: replace xx by x2x^2 in sin⁡x\sin x to get sin⁡(x2)=∑n=0∞(−1)nx4n+2(2n+1)!=x2−x66+x10120−⋯\sin(x^2) = \sum_{n=0}^{\infty} \frac{(-1)^n x^{4n+2}}{(2n+1)!} = x^2 - \frac{x^6}{6} + \frac{x^{10}}{120} - \cdots (all xx).

  • Nonelementary integral: ∫01sin⁡(x2) dx=13−142+11320−⋯≈0.31028\int_0^1 \sin(x^2)\,dx = \frac{1}{3} - \frac{1}{42} + \frac{1}{1320} - \cdots \approx 0.31028; alternating, so the error is less than the next term 115⋅7!=175600≈1.3×10−5\frac{1}{15 \cdot 7!} = \frac{1}{75600} \approx 1.3 \times 10^{-5}.

  • Differential equation y′=3yy' = 3y, y(0)=2y(0) = 2: with y=∑cnxny = \sum c_n x^n, matching powers gives (n+1)cn+1=3cn(n+1)c_{n+1} = 3c_n; c0=2c_0 = 2, so cn=2⋅3nn!c_n = \frac{2 \cdot 3^n}{n!} and y=2∑(3x)nn!=2e3xy = 2\sum \frac{(3x)^n}{n!} = 2e^{3x}.

  • Limit: lim⁡x→0ex−1−xx2=lim⁡x→0x22+x36+⋯x2=12\lim_{x\to 0} \frac{e^x - 1 - x}{x^2} = \lim_{x\to 0} \frac{\frac{x^2}{2} + \frac{x^3}{6} + \cdots}{x^2} = \frac{1}{2}.

Common mistakes

  • Binomial coefficients: the top keeps going down by 11: rr, r−1r-1, r−2,…r-2, \ldots (e.g. 12,−12,−32\frac{1}{2}, -\frac{1}{2}, -\frac{3}{2}). Watch the signs.

  • Substituting x2x^2 but forgetting to square every xx (so x33!\frac{x^3}{3!} becomes x63!\frac{x^6}{3!}, not x53!\frac{x^5}{3!}).

  • In differential-equation problems, matching coefficients with the wrong powers. Line up the same power of xx on both sides.

How the pieces connect

  • 7.1 supplies the tool every later series needs: the ratio test finds RR, and separate endpoint tests finish the interval.

  • 7.2 turns 11−x\frac{1}{1-x} into a factory: substitute, multiply by xmx^m, differentiate, or integrate term by term. The radius survives; endpoints must be re-checked.

  • 7.3 explains why the coefficients are forced: uniqueness means cn=f(n)(a)n!c_n = \frac{f^{(n)}(a)}{n!}, so a series built by 7.2 tricks is the same Taylor series derivatives would give.

  • Taylor's remainder (7.3) controls the error of any estimate; for alternating series the next-term bound (7.4) is the quicker option.

  • 7.4 is the payoff: the memorized table plus the binomial series solve differential equations, nonelementary integrals, and limits without repeated differentiation.

Checklist

  • Ratio test for RR, then test both endpoints

  • Rewrite a fraction as a1−r\frac{a}{1-r}

  • New interval after substituting bxmb x^m

  • Partial fractions, then two geometric series

  • Differentiate/integrate term by term (plus CC)

  • Present value, finite and perpetual

  • cn=f(n)(a)/n!c_n = f^{(n)}(a)/n!, Taylor polynomials

  • Remainder bound M(n+1)!∣x−a∣n+1\frac{M}{(n+1)!}|x-a|^{n+1}

  • Smallest nn for a given accuracy

  • Memorize the table of Maclaurin series

  • Binomial series for (1+x)r(1+x)^r

  • Series solutions of differential equations and integrals