Skip to content
Topos Uranos

Taylor polynomials and power series: synthesis and challenges

Summary

This lesson closes the unit on Taylor polynomials and power series, and with it the course. It gathers into a single deductive chain what the six content lessons proved: the Taylor polynomial of order nn is the only polynomial of degree at most nn with contact of order nn with the function, and its remainder is negligible compared with (x−a)n(x - a)^n; the Lagrange, Cauchy and integral formulas give the exact remainder and, with it, error bounds; uniform convergence, in which one and the same index serves for all points, is what allows passing to the limit in continuity, the integral and the derivative; power series converge on an interval whose radius is given by the Cauchy-Hadamard theorem, and inside it they are differentiated and integrated term by term, while Abel’s theorem carries continuity up to the endpoints; a function is the sum of its Taylor series exactly where its remainder tends to 00; and series compute integrals with no elementary antiderivative and constants such as ln⁡2\ln 2 and π4\frac{\pi}{4}, prove the irrationality of π\pi and give the complex exponential a series, whose sum at iθi\theta is the number cos⁡θ+isin⁡θ\cos \theta + i \sin \theta that Euler’s formula defines as eiθe^{i\theta}. It explains how the unit rests on Calculus I and on the preceding units, which of their promises it fulfills and what it leaves to the final review of the course and to Calculus III. It ends with an exit assessment, parallel to the entry assessment, and with ten challenges, harder than the problems of the lessons, each of which combines results from several of them.

Learning objectives

  1. Set out in logical order the results of the unit, indicating which results of Calculus I and of the preceding units each one depends on and in which proofs Taylor’s formula with Lagrange remainder, the Weierstrass M-test and Abel’s theorem are reused.
  2. Distinguish what each result gives: a local approximation (Peano remainder), an error bound (Lagrange, Cauchy and integral remainders), convergence at each point or on a whole set, and the representation of a function by its Taylor series.
  3. Relate the unit to Calculus I, to the units on sequences, the definite integral and numerical series, to the final review of the course and to Calculus III.
  4. Solve nonroutine problems that combine several lessons: expanding in order to compare, bounding a remainder, deciding uniformity with the supremum, summing a power series in the interior and on the boundary, and reading off coefficients in a known sum.

What the unit achieved

The unit on sequences studied when a sequence of numbers has a limit, and the unit on numerical series, when the sequence of partial sums of a series has one. This unit went further in two directions. The first, which rests on Calculus I, replaced a function, near a point, by the polynomial that has the same derivatives there up to a given order, and measured what is lost. The second, which rests on the units on sequences, the definite integral and numerical series, studied the sequences and series whose terms are functions. The two lines came together in the question that gives the unit its meaning: if the error of the Taylor polynomial tends to 00 as its order grows, the function is the sum of a power series,

f(x)=∑n=0∞f(n)(a)n!(x−a)nf(x) = \sum_{n=0}^{\infty} \frac{f^{(n)}(a)}{n!} (x - a)^n

and the theory of power series allows it to be differentiated, integrated and computed with. That being so, the chain has six segments: the Taylor polynomial, the Taylor remainder, sequences and series of functions, power series, Taylor series of the elementary functions and, finally, series, integrals and notable constants.

The Taylor polynomial

The lesson on the Taylor polynomial defined, for a function nn times differentiable at a point aa, the polynomial Pn,a(x)=∑k=0nf(k)(a)k!(x−a)kP_{n,a}(x) = \sum_{k=0}^{n} \frac{f^{(k)}(a)}{k!} (x - a)^k and the remainder Rn,a(x)=f(x)−Pn,a(x)R_{n,a}(x) = f(x) - P_{n,a}(x). It proved that a polynomial of degree at most nn that is negligible compared with (x−a)n(x - a)^n is zero, whence a function has at most one expansion of order nn; and Taylor’s formula with Peano remainder, under the exact hypotheses (n−1n - 1 derivatives on an open interval and the nnth at aa): the remainder is o⁡((x−a)n)\operatorname{o}\left((x - a)^n\right). Uniqueness turned the expansions into an algebra: those of the elementary functions are added, multiplied, composed with an infinitesimal, divided and integrated, and with them one computes limits that equivalent infinitesimals cannot reach and high-order derivatives without differentiating.

The Taylor remainder

The lesson on the Taylor remainder and its applications went from an asymptotic equality to a bound. With an auxiliary function and the mean value theorems of Cauchy and of Lagrange, it proved the formulas for the Lagrange and Cauchy remainders, with an intermediate point strictly between aa and xx, and, with integration by parts, that of the integral remainder, which requires the continuity of f(n+1)f^{(n + 1)}. From them came the bound for the remainder, which does not require knowing the intermediate point; the bound M2(x−a)2\frac{M}{2} (x - a)^2 for the error of the tangent line and of the differential; an explicit form of the higher derivative test for extrema; and the quadratic convergence of the Newton-Raphson method near a simple root.

Sequences and series of functions

The lesson on sequences and series of functions distinguished pointwise convergence, in which the index may depend on the point, from uniform convergence, in which one and the same index serves for all: the difference lies in the order of the quantifiers, and the supremum of ∣fn(x)−f(x)∣\abs{f_n(x) - f(x)} decides it. The powers xnx^n on [0,1]\icc{0}{1} showed that a pointwise limit of continuous functions can be discontinuous. It proved the uniform Cauchy criterion and the Weierstrass M-test, and three theorems on passing to the limit, each with a counterexample showing that its hypotheses are needed: the uniform limit of continuous functions is continuous; the integral of the uniform limit on a closed and bounded interval is the limit of the integrals; and a sequence of differentiable functions that converges at a point and whose derivatives converge uniformly has as its limit a differentiable function, whose derivative is the limit of the derivatives.

Power series

The lesson on power series proved, with Abel’s lemma and the completeness axiom, that every series ∑n=0∞cn(x−a)n\sum_{n=0}^{\infty} c_n (x - a)^n has a radius of convergence ρ\rho, and the Cauchy-Hadamard theorem computed it with the limit superior of the roots of the coefficients, which the root test of the preceding unit had prepared. On the closed interior intervals the convergence is uniform, by the M-test; from this, the sum is continuous, is differentiated and integrated term by term with the same radius, and its coefficients are the derivatives of the sum at the center divided by the factorials, so that two series with the same sum near the center are the same. At the endpoints anything can happen, and Abel’s theorem, proved with the summation by parts of the preceding unit, ensures that, if the series converges at an endpoint, it converges uniformly up to it and its sum is continuous there.

Taylor series of the elementary functions

The lesson on Taylor series of the elementary functions brought the two lines together. The Taylor series of an infinitely differentiable function converges to it at xx if and only if the remainder tends to 00 at xx, and a power series with sum ff near the center is the Taylor series of ff. With the bounds for the remainder and the limit ynn!→0\frac{y^n}{n!} \to 0, it proved that the exponential, the sine and the cosine are the sums of their series on the whole line; with term-by-term integration and Abel’s theorem, the series for the logarithm of 1+x1 + x on (−1,1]\ioc{-1}{1}, which gives ln⁡2\ln 2, and the series for the arctangent on [−1,1]\icc{-1}{1}, which gives π4\frac{\pi}{4}; and the binomial series of (1+x)α(1 + x)^{\alpha} on (−1,1)\ioo{-1}{1}, through the equation (1+x) S′(x)=αS(x)(1 + x) \, S'(x) = \alpha S(x). The function that extends e−1/x2e^{-1/x^2} by 00 has all its derivatives zero at 00: its Taylor series at 00 converges on the whole line, but not to it, and it is not analytic.

Series, integrals and notable constants

The lesson on series, integrals and notable constants applied all of the above. It wrote as series the antiderivative of e−t2e^{-t^2}, with the error bound of the Leibniz test, and the integral of x−xx^{-x} on [0,1]\icc{0}{1}, which is the sum of the 1nn\frac{1}{n^n}; it compared the sum ln⁡2\ln 2 with the route of the preceding unit and accelerated the computation of ln⁡2\ln 2 and of π4\frac{\pi}{4}; it returned to the irrationality of ee, which the theorem “Irrationality of e”, of the lesson on the number e and the natural exponential, already proved, and read it as a property of the tail of the exponential series; it proved that π\pi is irrational, with Niven’s proof; and, with the absolute convergence of the exponential series at a complex number, it proved that its sum is the complex exponential of Algebra and Geometry II: at iθi\theta, the sum of the series is cos⁡θ+isin⁡θ\cos \theta + i \sin \theta, the number that Euler’s formula defines there as eiθe^{i\theta}, so that the definition agrees with the sum of a series.

What is admitted, what is proved and what remains

The balance of the unit can be summarized as follows:

  • Admitted are the completeness axiom of Algebra and Geometry I; the results of Algebra and Geometry on summations, the trigonometric functions and complex numbers; those of Calculus I on limits, equivalent infinitesimals, higher-order derivatives, the mean value theorems, L’Hôpital’s rule, the exponential and the logarithm; and those of the preceding units on sequences, the definite integral, improper integrals and numerical series.
  • Proved are Taylor’s formula with its four remainders, the algebra of expansions, the theorems on uniform convergence, the theory of power series, the series of the elementary functions with their endpoints, the existence of infinitely differentiable functions that are not analytic, the irrationality of π\pi and the convergence of the exponential series in the complex numbers, whose sum is the complex exponential that Algebra and Geometry II defines with Euler’s formula. Thus the promises of Calculus I are fulfilled: that of the lesson on higher-order derivatives and the differential, which announced the Taylor polynomial and the error of the differential; that of the squeeze theorem, on how closely cos⁡x\cos x resembles 1−x221 - \frac{x^2}{2}; that of the theorems of Fermat, Rolle, Lagrange and Cauchy, on the constant M2\frac{M}{2}; that of L’Hôpital’s rule, which the Taylor polynomial generalizes; that of the Newton-Raphson method, on its quadratic convergence; and those of equivalent infinitesimals, monotonicity, the syntheses of the units on limits and on applications of derivatives and the syllabus. The promises of this course are also fulfilled: those of the introduction and the synthesis of the unit on sequences, which announced sequences of functions and their uniform convergence, on the model of uniform continuity; those of the introduction and the synthesis of the unit on the definite integral and those of the introduction and the synthesis of the unit on improper integrals, which announced term-by-term integration and the integral of the limit of a sequence of functions; that of the synthesis of the unit on applications of the integral, which left the Taylor polynomial to this unit; that of the lesson on the ratio and root tests, which announced the radius of convergence and the sum ∑n=1∞1n 2n=ln⁡2\sum_{n=1}^{\infty} \frac{1}{n \, 2^n} = \ln 2; and those of the introduction and the synthesis of the unit on numerical series, which announced the radius of convergence of power series and uniform convergence.
  • Three instruments are reused instead of repeating arguments: Taylor’s formula with Lagrange remainder, which gives the bound for the remainder, the extrema, the convergence of Newton-Raphson and, with the bounded derivatives criterion, the series for the exponential, the sine and the cosine; the Weierstrass M-test, which gives the uniform convergence of power series on the closed interior intervals and the term-by-term integration of the series of the sophomore’s dream and of the bell curve; and Abel’s theorem, which carries the series for the logarithm and the arctangent and the binomial series up to the endpoints of their interval of convergence.
  • Left for later courses, as facts on which nothing here depends, are Fourier series, complex analysis, in which the radius 11 of the series of 11+x2\frac{1}{1 + x^2} is explained by the points ±i\pm i, the Weierstrass approximation theorem, infinite products and the value ∑n=1∞1n2=π26\sum_{n=1}^{\infty} \frac{1}{n^2} = \frac{\pi^2}{6}.

The following figure gathers the six segments into a single chain. Each node of the column is a result, numbered according to the logical order and joined by an arc to the earlier links that its proof uses; around it float the supports, that is, the principles and tools to which the chain resorts again and again, each one joined by a solid line to the links that rest on it and by a dotted one to the link where it is established, except those that come from Calculus I, from Algebra and Geometry or from the preceding units, which are drawn as rings. Clicking a support highlights all the links that use it, and each link refers to the lesson in which it is proved.

Supports of the chain

  • ACompleteness axiom. Axiom of Algebra and Geometry I: every nonempty set of real numbers that is bounded above has a supremum. Here it gives the supremum of the differences ∣fn(x)−f(x)∣\abs{f_n(x) - f(x)}, the radius of convergence as the supremum of the distances to the center of the points of convergence, and the limits of Cauchy sequences. Admitted without proof in this unit.
  • BCalculus I and Algebra and Geometry. Proved in Calculus I and in Algebra and Geometry: higher-order derivatives and the Leibniz rule, the mean value theorems of Rolle, Lagrange and Cauchy, L’Hôpital’s rule and the corollary on derivatives vanishing at a point, the symbol oo and equivalent infinitesimals, the exponential, the logarithm and powers with real exponents, the trigonometric formulas and complex numbers with De Moivre’s formula. Admitted without proof in this unit.
  • CSequences. From the unit on sequences: the limit theorems, monotone convergence, the limits superior and inferior, the Cauchy criterion, the characterization of the limit and of continuity by sequences and the orders of growth. Admitted without proof in this unit.
  • DIntegrals. From the units on the definite integral and improper integrals: the Riemann integral and the integrability of continuous functions, the bounds and the additivity of the integral, the fundamental theorem of calculus, integration by parts and improper integrals. Admitted without proof in this unit.
  • ENumerical series. From the unit on numerical series: the geometric series, comparison, the ratio and root tests, the Leibniz test with the bound for the remainder, Abel’s summation by parts, absolute convergence and the Cauchy product. Admitted without proof in this unit.
  • FLagrange remainder. If ff is n+1n + 1 times differentiable on an open interval containing aa and x≠ax \neq a, there exists cc strictly between aa and xx with Rn,a(x)=f(n+1)(c)(n+1)!(x−a)n+1R_{n,a}(x) = \frac{f^{(n + 1)}(c)}{(n + 1)!} (x - a)^{n + 1}: the way to bound a remainder without knowing the intermediate point. Established in: 11. Taylor’s formula with Lagrange remainder.
  • GM-test. If ∣fn(x)∣≤Mn\abs{f_n(x)} \leq M_n for every nn and every xx in AA, and ∑n=1∞Mn\sum_{n=1}^{\infty} M_n converges, the series ∑n=1∞fn\sum_{n=1}^{\infty} f_n converges uniformly on AA: the usual way to prove a uniform convergence. Established in: 25. Weierstrass M-test.
  • HAbel’s theorem. If ∑n=0∞cn(x−a)n\sum_{n=0}^{\infty} c_n (x - a)^n converges at a+ra + r, with r>0r > 0, it converges uniformly on [a,a+r]\icc{a}{a + r} and its sum is continuous there: the way to carry an equality up to an endpoint of the interval of convergence. Established in: 38. Abel’s theorem.
  1. Segment 1 The Taylor polynomial

    • Define the Taylor polynomial and its remainder, and prove the uniqueness of the expansion.
    • Prove Taylor’s formula with Peano remainder under its exact hypotheses.
    • Add, multiply, compose, divide and integrate expansions, and compute limits with them.
    1. Polynomials in powers of x minus a. Every polynomial QQ of degree at most nn can be written in a unique way as ∑k=0nck(x−a)k\sum_{k=0}^{n} c_k (x - a)^k, with ck=Q(k)(a)k!c_k = \frac{Q^{(k)}(a)}{k!}.

      Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    2. Uniqueness of the expansion. If a polynomial of degree at most nn is o⁡((x−a)n)\operatorname{o}\left((x - a)^n\right) as xx tends to aa, all its coefficients are zero; a function has at most one expansion of order nn at aa.

      Uses: 1. Polynomials in powers of x minus a. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    3. Taylor’s formula with Peano remainder. If n≥1n \geq 1 and ff is n−1n - 1 times differentiable on an open interval containing aa and nn times differentiable at aa, then Rn,a(x)=o⁡((x−a)n)R_{n,a}(x) = \operatorname{o}\left((x - a)^n\right), and Pn,aP_{n,a} is the only polynomial of degree at most nn with contact of order nn with ff at aa.

      Uses: 1. Polynomials in powers of x minus a, 2. Uniqueness of the expansion. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    4. Principal part of the remainder. If ff is nn times differentiable on an open interval containing aa and n+1n + 1 times differentiable at aa, then Rn,a(x)=f(n+1)(a)(n+1)!(x−a)n+1+o⁡((x−a)n+1)R_{n,a}(x) = \frac{f^{(n + 1)}(a)}{(n + 1)!} (x - a)^{n + 1} + \operatorname{o}\left((x - a)^{n + 1}\right).

      Uses: 3. Taylor’s formula with Peano remainder. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    5. Expansions at zero of the elementary functions. As xx tends to 00, exe^x, ln⁡(1+x)\ln(1 + x), sin⁡x\sin x, cos⁡x\cos x and (1+x)r(1 + x)^r are equal to their Taylor polynomials at 00 plus an oo of the corresponding power.

      Uses: 3. Taylor’s formula with Peano remainder. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    6. Sum and product of expansions. If ff and gg have expansions of order nn at 00 with polynomial parts QQ and SS, the sum αf+g\alpha f + g has the polynomial part αQ+S\alpha Q + S, and the product fgf g, the polynomial part of degree at most nn of QSQ S.

      Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    7. Composition of expansions. If gg is continuous at 00 and has an expansion of order nn at 00, and uu has an expansion of order nn with zero polynomial part at 00, the composition g(u(x))g(u(x)) has as its polynomial part the terms of degree at most nn of the composition of the polynomial parts.

      Uses: 6. Sum and product of expansions. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    8. Expansion of a quotient. If ff and gg have expansions of order nn at 00 and the polynomial part of gg does not vanish at 00, the quotient fg\frac{f}{g} has an expansion of order nn at 00, which is obtained with the truncated geometric series.

      Uses: 7. Composition of expansions, 6. Sum and product of expansions. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

    9. Expansion of an antiderivative. If ff is differentiable on an open interval containing 00 and f′(x)=∑k=0nckxk+o⁡(xn)f'(x) = \sum_{k=0}^{n} c_k x^k + \operatorname{o}(x^n), then f(x)=f(0)+∑k=0nckk+1xk+1+o⁡(xn+1)f(x) = f(0) + \sum_{k=0}^{n} \frac{c_k}{k + 1} x^{k + 1} + \operatorname{o}\left(x^{n + 1}\right).

      Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor polynomial.

  2. Segment 2 The Taylor remainder

    • Prove the formulas for the Lagrange, Cauchy and integral remainders, and bound the remainder.
    • Bound the error of the tangent line and of the differential, and decide extrema with the remainder.
    • Prove the quadratic convergence of the Newton-Raphson method.
    1. The auxiliary function of the remainder. If ff is n+1n + 1 times differentiable on an open interval II and xx is in II, the function F(t)=f(x)−∑k=0nf(k)(t)k!(x−t)kF(t) = f(x) - \sum_{k=0}^{n} \frac{f^{(k)}(t)}{k!} (x - t)^k satisfies F(x)=0F(x) = 0, F(a)=Rn,a(x)F(a) = R_{n,a}(x) and F′(t)=−f(n+1)(t)n!(x−t)nF'(t) = -\frac{f^{(n + 1)}(t)}{n!} (x - t)^n.

      Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor remainder and its applications.

    2. Taylor’s formula with Lagrange remainder. If ff is n+1n + 1 times differentiable on an open interval II containing aa, for every x≠ax \neq a in II there exists cc strictly between aa and xx with Rn,a(x)=f(n+1)(c)(n+1)!(x−a)n+1R_{n,a}(x) = \frac{f^{(n + 1)}(c)}{(n + 1)!} (x - a)^{n + 1}.

      Uses: 10. The auxiliary function of the remainder. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor remainder and its applications.

    3. Taylor’s formula with Cauchy remainder. Under the same hypotheses, there exists cc strictly between aa and xx with Rn,a(x)=f(n+1)(c)n!(x−c)n(x−a)R_{n,a}(x) = \frac{f^{(n + 1)}(c)}{n!} (x - c)^n (x - a).

      Uses: 10. The auxiliary function of the remainder. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor remainder and its applications.

    4. Bound for the Taylor remainder. If moreover ∣f(n+1)(t)∣≤M\abs{f^{(n + 1)}(t)} \leq M for every tt strictly between aa and xx, then ∣Rn,a(x)∣≤M(n+1)!∣x−a∣n+1\abs{R_{n,a}(x)} \leq \frac{M}{(n + 1)!} \abs{x - a}^{n + 1}.

      Rests on: F (Lagrange remainder). Proved in: The Taylor remainder and its applications.

    5. Taylor’s formula with integral remainder. If ff is n+1n + 1 times differentiable on an open interval II containing aa, with f(n+1)f^{(n + 1)} continuous on II, then Rn,a(x)=1n!∫axf(n+1)(t) (x−t)n dtR_{n,a}(x) = \frac{1}{n!} \int_a^x f^{(n + 1)}(t) \, (x - t)^n \, \mathrm{d}t for every xx in II.

      Rests on: D (Integrals), B (Calculus I and Algebra and Geometry). Proved in: The Taylor remainder and its applications.

    6. The error of the tangent line. If ff is twice differentiable on an open interval and ∣f′′(t)∣≤M\abs{f''(t)} \leq M between aa and xx, then ∣f(x)−f(a)−f′(a) (x−a)∣≤M2(x−a)2\abs{f(x) - f(a) - f'(a) \, (x - a)} \leq \frac{M}{2} (x - a)^2.

      Uses: 13. Bound for the Taylor remainder. Rests on: F (Lagrange remainder). Proved in: The Taylor remainder and its applications.

    7. Extrema with the Lagrange remainder. If n≥2n \geq 2, ff is nn times differentiable on an open interval, f′(a)=⋯=f(n−1)(a)=0f'(a) = \cdots = f^{(n - 1)}(a) = 0 and f(n)>0f^{(n)} > 0 on an interval containing aa, except perhaps at aa, then aa is a strict minimum if nn is even, and f−f(a)f - f(a) changes sign at aa if nn is odd.

      Rests on: F (Lagrange remainder). Proved in: The Taylor remainder and its applications.

    8. Quadratic convergence of the Newton-Raphson method. If f(r)=0f(r) = 0, ∣f′∣≥m>0\abs{f'} \geq m > 0 and ∣f′′∣≤M\abs{f''} \leq M on [r−δ,r+δ]\icc{r - \delta}{r + \delta}, and Kδ<1K \delta < 1 with K=M2mK = \frac{M}{2m}, the iterations from any point of that interval converge to rr, with ∣xn+1−r∣≤K∣xn−r∣2\abs{x_{n + 1} - r} \leq K \abs{x_n - r}^2.

      Rests on: F (Lagrange remainder), C (Sequences). Proved in: The Taylor remainder and its applications.

    9. Quadratic convergence near a simple root. If f′′f'' is continuous, f(r)=0f(r) = 0 and f′(r)≠0f'(r) \neq 0, there is an interval around rr from whose points the Newton-Raphson iterations converge to rr with order two.

      Uses: 17. Quadratic convergence of the Newton-Raphson method. Rests on: B (Calculus I and Algebra and Geometry). Proved in: The Taylor remainder and its applications.

  3. Segment 3 Sequences and series of functions

    • Distinguish pointwise from uniform convergence and decide the latter with the supremum.
    • Prove the uniform Cauchy criterion and the Weierstrass M-test.
    • Prove the continuity, integration and differentiation theorems for the uniform limit, with their counterexamples.
    1. Uniform convergence via the supremum. The sequence (fn)(f_n) converges uniformly to ff on AA if and only if, from some index on, the suprema dn=sup⁡x∈A∣fn(x)−f(x)∣d_n = \sup_{x \in A} \abs{f_n(x) - f(x)} exist and tend to 00.

      Rests on: A (Completeness axiom), C (Sequences). Proved in: Sequences and series of functions.

    2. The powers on the unit interval. The powers xnx^n converge pointwise on [0,1]\icc{0}{1} to a function that is discontinuous at 11; the convergence is not uniform on [0,1)\ico{0}{1}, and it is uniform on [0,r]\icc{0}{r} if 0≤r<10 \leq r < 1.

      Rests on: C (Sequences). Proved in: Sequences and series of functions.

    3. Uniform Cauchy criterion. The sequence (fn)(f_n) converges uniformly on AA if and only if for every ε>0\varepsilon > 0 there exists NN with ∣fn(x)−fm(x)∣<ε\abs{f_n(x) - f_m(x)} < \varepsilon for all m,n≥Nm, n \geq N and every xx in AA.

      Rests on: C (Sequences), A (Completeness axiom). Proved in: Sequences and series of functions.

    4. The uniform limit of continuous functions is continuous. If fn→ff_n \to f uniformly on AA and each fnf_n is continuous at a point aa of AA, then ff is also continuous at aa.

      Rests on: B (Calculus I and Algebra and Geometry). Proved in: Sequences and series of functions.

    5. Integration of the uniform limit. If the fnf_n are integrable on [a,b]\icc{a}{b} and fn→ff_n \to f uniformly there, ff is integrable and ∫axfn→∫axf\int_a^x f_n \to \int_a^x f uniformly on [a,b]\icc{a}{b}.

      Rests on: D (Integrals). Proved in: Sequences and series of functions.

    6. Differentiation of the uniform limit. If the fnf_n are differentiable on (a,b)\ioo{a}{b}, (fn(x0))(f_n(x_0)) converges at a point and fn′→gf_n' \to g uniformly on (a,b)\ioo{a}{b}, then fnf_n converges uniformly to a differentiable function ff with f′=gf' = g.

      Uses: 21. Uniform Cauchy criterion. Rests on: B (Calculus I and Algebra and Geometry), C (Sequences). Proved in: Sequences and series of functions.

    7. Weierstrass M-test. If ∣fn(x)∣≤Mn\abs{f_n(x)} \leq M_n for every nn and every xx in AA, and the series of the MnM_n converges, the series ∑n=1∞fn\sum_{n=1}^{\infty} f_n converges absolutely at each point and uniformly on AA.

      Rests on: E (Numerical series). Proved in: Sequences and series of functions.

    8. Term-by-term continuity, integration and differentiation. A uniformly convergent series of continuous functions has a continuous sum; on [a,b]\icc{a}{b} it is integrated term by term; and on (a,b)\ioo{a}{b}, if it converges at a point and the series of the derivatives converges uniformly, it is differentiated term by term.

      Uses: 22. The uniform limit of continuous functions is continuous, 23. Integration of the uniform limit, 24. Differentiation of the uniform limit. Rests on: D (Integrals). Proved in: Sequences and series of functions.

  4. Segment 4 Power series

    • Prove the existence of the radius of convergence and compute it with the Cauchy-Hadamard theorem.
    • Differentiate and integrate a power series term by term on its interval of convergence.
    • Study the endpoints and prove Abel’s theorem.
    1. Abel’s lemma. If the terms cn(x0−a)nc_n (x_0 - a)^n are bounded, with x0≠ax_0 \neq a, the series converges absolutely at every xx with ∣x−a∣<∣x0−a∣\abs{x - a} < \abs{x_0 - a}.

      Rests on: E (Numerical series), C (Sequences). Proved in: Power series.

    2. Existence of the radius of convergence. There is a unique ρ\rho, a nonnegative real number or ∞\infty, such that the series converges absolutely if ∣x−a∣<ρ\abs{x - a} < \rho and diverges if ∣x−a∣>ρ\abs{x - a} > \rho.

      Uses: 27. Abel’s lemma. Rests on: A (Completeness axiom). Proved in: Power series.

    3. Cauchy-Hadamard theorem. With L=lim sup⁡n→∞∣cn∣nL = \limsup_{n \to \infty} \sqrt[n]{\abs{c_n}}, the radius is 1L\frac{1}{L} if 0<L<∞0 < L < \infty, is ∞\infty if L=0L = 0 and is 00 if L=∞L = \infty.

      Uses: 28. Existence of the radius of convergence. Rests on: E (Numerical series), C (Sequences). Proved in: Power series.

    4. The radius by the ratio. If cn≠0c_n \neq 0 from some index on and ∣cn+1cn∣→ℓ\abs{\frac{c_{n + 1}}{c_n}} \to \ell, the radius is 1ℓ\frac{1}{\ell}, with the conventions for ℓ=0\ell = 0 and ℓ=∞\ell = \infty.

      Uses: 29. Cauchy-Hadamard theorem. Rests on: C (Sequences), E (Numerical series). Proved in: Power series.

    5. Uniform convergence on the closed interior intervals. If 0<r<ρ0 < r < \rho, the series converges uniformly on [a−r,a+r]\icc{a - r}{a + r}, and its remainder does not exceed ∑k=n+1∞∣ck∣rk\sum_{k=n + 1}^{\infty} \abs{c_k} r^k there.

      Uses: 28. Existence of the radius of convergence. Rests on: G (M-test), E (Numerical series). Proved in: Power series.

    6. Continuity of the sum. If ρ>0\rho > 0, the sum of the series is continuous on its interval of convergence.

      Uses: 31. Uniform convergence on the closed interior intervals, 26. Term-by-term continuity, integration and differentiation. Proved in: Power series.

    7. The same radius for the differentiated and for the integrated series. The differentiated series and the integrated series of a power series have the same radius of convergence as the series itself.

      Uses: 28. Existence of the radius of convergence. Rests on: E (Numerical series), C (Sequences). Proved in: Power series.

    8. Term-by-term differentiation of a power series. If ρ>0\rho > 0, the sum SS is differentiable on the interval of convergence, and S′(x)=∑n=1∞ncn(x−a)n−1S'(x) = \sum_{n=1}^{\infty} n c_n (x - a)^{n - 1}.

      Uses: 33. The same radius for the differentiated and for the integrated series, 31. Uniform convergence on the closed interior intervals, 26. Term-by-term continuity, integration and differentiation. Proved in: Power series.

    9. Higher-order derivatives and uniqueness of the coefficients. The sum has derivatives of all orders on the interval of convergence, ck=S(k)(a)k!c_k = \frac{S^{(k)}(a)}{k!}, and two series with the same sum on an interval (a−δ,a+δ)\ioo{a - \delta}{a + \delta} have the same coefficients.

      Uses: 34. Term-by-term differentiation of a power series. Proved in: Power series.

    10. Term-by-term integration of a power series. If ρ>0\rho > 0, for every xx in the interval of convergence, ∫axS(t) dt=∑n=0∞cnn+1(x−a)n+1\int_a^x S(t) \, \mathrm{d}t = \sum_{n=0}^{\infty} \frac{c_n}{n + 1} (x - a)^{n + 1}.

      Uses: 32. Continuity of the sum, 33. The same radius for the differentiated and for the integrated series, 34. Term-by-term differentiation of a power series. Rests on: D (Integrals). Proved in: Power series.

    11. The four behaviors at the endpoints. The series of xnx^n, of xnn2\frac{x^n}{n^2}, of xnn\frac{x^n}{n} and of (−1)nxnn\frac{(-1)^n x^n}{n} have radius 11 and converge exactly on (−1,1)\ioo{-1}{1}, on [−1,1]\icc{-1}{1}, on [−1,1)\ico{-1}{1} and on (−1,1]\ioc{-1}{1}.

      Uses: 30. The radius by the ratio. Rests on: E (Numerical series), C (Sequences). Proved in: Power series.

    12. Abel’s theorem. If the series converges at a+ra + r, with r>0r > 0, it converges uniformly on [a,a+r]\icc{a}{a + r}, its sum is continuous there and S(xj)→∑n=0∞cnrnS(x_j) \to \sum_{n=0}^{\infty} c_n r^n for every sequence xj→a+rx_j \to a + r in that interval; likewise on a−ra - r.

      Uses: 21. Uniform Cauchy criterion, 26. Term-by-term continuity, integration and differentiation. Rests on: E (Numerical series), C (Sequences). Proved in: Power series.

  5. Segment 5 Taylor series of the elementary functions

    • Decide where a function is represented by its Taylor series, with the remainder or with the theory of power series.
    • Prove the series for the exponential, the sine, the cosine, the logarithm and the arctangent and the binomial series, with their endpoints.
    • Construct an infinitely differentiable function that is not analytic.
    1. Representation of a function by its Taylor series. The Taylor series of ff at aa converges at xx with sum f(x)f(x) if and only if Rn,a(x)→0R_{n,a}(x) \to 0.

      Rests on: C (Sequences). Proved in: Taylor series of the elementary functions.

    2. The power series of a function is its Taylor series. If f(x)=∑n=0∞cn(x−a)nf(x) = \sum_{n=0}^{\infty} c_n (x - a)^n on (a−δ,a+δ)\ioo{a - \delta}{a + \delta}, ff is infinitely differentiable there and cn=f(n)(a)n!c_n = \frac{f^{(n)}(a)}{n!}.

      Uses: 28. Existence of the radius of convergence, 35. Higher-order derivatives and uniqueness of the coefficients. Proved in: Taylor series of the elementary functions.

    3. Powers divided by the factorial. For every real number yy, ynn!→0\frac{y^n}{n!} \to 0.

      Rests on: C (Sequences). Proved in: Taylor series of the elementary functions.

    4. Bounded derivatives criterion. If ∣f(n)(t)∣≤MBn\abs{f^{(n)}(t)} \leq M B^n for every nn and every tt strictly between aa and xx, the Taylor series of ff at aa converges at xx with sum f(x)f(x).

      Uses: 13. Bound for the Taylor remainder, 39. Representation of a function by its Taylor series, 41. Powers divided by the factorial. Rests on: F (Lagrange remainder), C (Sequences). Proved in: Taylor series of the elementary functions.

    5. Series for the exponential, the sine and the cosine. For every real xx, ex=∑n=0∞xnn!e^x = \sum_{n=0}^{\infty} \frac{x^n}{n!}, sin⁡x=∑n=0∞(−1)nx2n+1(2n+1)!\sin x = \sum_{n=0}^{\infty} \frac{(-1)^n x^{2n + 1}}{(2n + 1)!} and cos⁡x=∑n=0∞(−1)nx2n(2n)!\cos x = \sum_{n=0}^{\infty} \frac{(-1)^n x^{2n}}{(2n)!}.

      Uses: 42. Bounded derivatives criterion. Rests on: B (Calculus I and Algebra and Geometry), E (Numerical series). Proved in: Taylor series of the elementary functions.

    6. Series for the logarithm. For every xx in (−1,1]\ioc{-1}{1}, ln⁡(1+x)=∑n=1∞(−1)n+1xnn\ln(1 + x) = \sum_{n=1}^{\infty} \frac{(-1)^{n + 1} x^n}{n}; in particular, ln⁡2=∑n=1∞(−1)n+1n\ln 2 = \sum_{n=1}^{\infty} \frac{(-1)^{n + 1}}{n}.

      Uses: 28. Existence of the radius of convergence, 36. Term-by-term integration of a power series, 37. The four behaviors at the endpoints, 40. The power series of a function is its Taylor series. Rests on: H (Abel’s theorem), D (Integrals), E (Numerical series), C (Sequences), B (Calculus I and Algebra and Geometry). Proved in: Taylor series of the elementary functions.

    7. Series for the arctangent. For every xx in [−1,1]\icc{-1}{1}, arctan⁡x=∑n=0∞(−1)nx2n+12n+1\arctan x = \sum_{n=0}^{\infty} \frac{(-1)^n x^{2n + 1}}{2n + 1}; in particular, π4=∑n=0∞(−1)n2n+1\frac{\pi}{4} = \sum_{n=0}^{\infty} \frac{(-1)^n}{2n + 1}.

      Uses: 36. Term-by-term integration of a power series, 44. Series for the logarithm, 40. The power series of a function is its Taylor series. Rests on: H (Abel’s theorem), E (Numerical series), D (Integrals), C (Sequences), B (Calculus I and Algebra and Geometry). Proved in: Taylor series of the elementary functions.

    8. Binomial series. For every real α\alpha and every xx in (−1,1)\ioo{-1}{1}, (1+x)α=∑n=0∞cnxn(1 + x)^{\alpha} = \sum_{n=0}^{\infty} c_n x^n, with c0=1c_0 = 1 and cn=α(α−1)⋯(α−n+1)n!c_n = \frac{\alpha (\alpha - 1) \cdots (\alpha - n + 1)}{n!}.

      Uses: 30. The radius by the ratio, 34. Term-by-term differentiation of a power series, 40. The power series of a function is its Taylor series. Rests on: B (Calculus I and Algebra and Geometry), E (Numerical series). Proved in: Taylor series of the elementary functions.

    9. The binomial series at the endpoints. If α>0\alpha > 0, the binomial series converges absolutely on ±1\pm 1, with sums 2α2^{\alpha} and 00.

      Uses: 46. Binomial series. Rests on: H (Abel’s theorem), E (Numerical series), C (Sequences). Proved in: Taylor series of the elementary functions.

    10. The exponential of minus the reciprocal of the square. For every integer k≥0k \geq 0, e−1/x2xk→0\frac{e^{-1/x^2}}{x^k} \to 0 as xx tends to 00.

      Uses: 43. Series for the exponential, the sine and the cosine. Rests on: B (Calculus I and Algebra and Geometry), E (Numerical series). Proved in: Taylor series of the elementary functions.

    11. A function with all its derivatives zero at zero. The function equal to e−1/x2e^{-1/x^2} if x≠0x \neq 0 and to 00 at 00 is infinitely differentiable on the whole line, and all its derivatives at 00 are zero.

      Uses: 48. The exponential of minus the reciprocal of the square. Rests on: B (Calculus I and Algebra and Geometry). Proved in: Taylor series of the elementary functions.

    12. An infinitely differentiable function that is not analytic. The Taylor series of that function at 00 converges on the whole line with sum 00, which agrees with the function only at the origin: the function is not analytic at 00.

      Uses: 49. A function with all its derivatives zero at zero, 39. Representation of a function by its Taylor series, 40. The power series of a function is its Taylor series. Rests on: B (Calculus I and Algebra and Geometry). Proved in: Taylor series of the elementary functions.

  6. Segment 6 Series, integrals and notable constants

    • Write as series integrals with no elementary antiderivative and bound the error.
    • Compute constants with series and prove the irrationality of π\pi.
    • Prove the absolute convergence of the exponential series at a complex number and identify its sum with the complex exponential.
    1. The bell curve integral as a series. For every real xx, ∫0xe−t2 dt=∑n=0∞(−1)nx2n+1n! (2n+1)\int_0^x e^{-t^2} \, \mathrm{d}t = \sum_{n=0}^{\infty} \frac{(-1)^n x^{2n + 1}}{n! \, (2n + 1)}, and if 0<x≤10 < x \leq 1 the error of each partial sum does not exceed the first omitted term.

      Uses: 26. Term-by-term continuity, integration and differentiation, 43. Series for the exponential, the sine and the cosine, 41. Powers divided by the factorial. Rests on: G (M-test), E (Numerical series), D (Integrals). Proved in: Series, integrals and notable constants.

    2. Integrals of powers times powers of the logarithm. If 0≤k≤n0 \leq k \leq n, the function xn(ln⁡x)kx^n (\ln x)^k, extended by 00 at 00, is continuous on [0,1]\icc{0}{1}, and its integral there is (−1)k k!(n+1)k+1\frac{(-1)^k \, k!}{(n + 1)^{k + 1}}.

      Rests on: D (Integrals), B (Calculus I and Algebra and Geometry). Proved in: Series, integrals and notable constants.

    3. The sophomore’s dream. The function x−xx^{-x}, extended by 11 at 00, is continuous on [0,1]\icc{0}{1}, and its integral there is ∑n=1∞1nn\sum_{n=1}^{\infty} \frac{1}{n^n}, also as an improper integral.

      Uses: 26. Term-by-term continuity, integration and differentiation, 52. Integrals of powers times powers of the logarithm, 43. Series for the exponential, the sine and the cosine. Rests on: G (M-test), D (Integrals), B (Calculus I and Algebra and Geometry). Proved in: Series, integrals and notable constants.

    4. The error of the partial sums of the arctangent. If 0<x≤10 < x \leq 1, the error of the partial sum of order nn of the arctangent series has the sign of (−1)n+1(-1)^{n + 1} and does not exceed x2n+32n+3\frac{x^{2n + 3}}{2n + 3}.

      Uses: 45. Series for the arctangent. Rests on: E (Numerical series), C (Sequences). Proved in: Series, integrals and notable constants.

    5. Irrationality of pi. The number π\pi is irrational: if it were ab\frac{a}{b}, the integral of xn(a−bx)nn!sin⁡x\frac{x^n (a - b x)^n}{n!} \sin x on [0,π]\icc{0}{\pi} would be a positive integer less than 11 for nn large.

      Uses: 41. Powers divided by the factorial. Rests on: B (Calculus I and Algebra and Geometry), D (Integrals). Proved in: Series, integrals and notable constants.

    6. Absolute convergence of a series of complex numbers. If the series of the moduli of a series of complex numbers converges, the series converges.

      Rests on: E (Numerical series), B (Calculus I and Algebra and Geometry). Proved in: Series, integrals and notable constants.

    7. The exponential series at a complex number. For every complex z=a+biz = a + bi, the series ∑n=0∞znn!\sum_{n=0}^{\infty} \frac{z^n}{n!} converges absolutely with sum ea(cos⁡b+isin⁡b)e^a \left(\cos b + i \sin b\right); at iθi\theta, its sum is cos⁡θ+isin⁡θ\cos \theta + i \sin \theta, the number that Euler’s formula defines as eiθe^{i\theta}.

      Uses: 56. Absolute convergence of a series of complex numbers, 43. Series for the exponential, the sine and the cosine. Rests on: E (Numerical series), B (Calculus I and Algebra and Geometry), C (Sequences). Proved in: Series, integrals and notable constants.

Toward the final review of the course and Calculus III. The chain stops where the course ends. The final review lesson of Calculus II goes through its seven units, from the integration techniques to power series, as a single spine: passing to the limit, in Riemann sums, in improper integrals, in sequences, in series and in sequences of functions. Calculus III will carry the Taylor polynomial over to functions of several variables, with partial derivatives in place of higher-order derivatives. Fourier series, complex analysis and the Weierstrass approximation theorem are left for later courses, and nothing proved here depends on them.