Summary
This lesson opens the seventh and last unit of the course, devoted to the Taylor polynomial and to power series. It presents its route, which brings together two lines: that of the Taylor polynomial, which approximates a function near a point with an error that one knows how to bound, and which rests on Calculus I; and that of sequences and series of functions, with uniform convergence and power series, which rests on the units on sequences, the definite integral and numerical series. The two meet when the error of the polynomial tends to as its order grows: then the function is the sum of its Taylor series, and with it one differentiates, integrates and computes constants. It reconstructs the history of those ideas, from the series of Gregory and of Newton, between 1668 and 1676, through Taylor’s Methodus incrementorum, of 1715, Maclaurin’s treatise, of 1742, Lagrange’s theory of functions, of 1797, and Cauchy’s function, of 1823, whose derivatives at are all zero, up to Cauchy’s error on the continuity of the sum of a series, Abel’s memoir of 1826 and Weierstrass’s uniform convergence; and the constants, from Euler’s formula of 1748 to the irrationality of , by Lambert in 1768 and by Niven in 1947. It formulates the problem that runs through it with a function differentiable on the whole line whose Taylor polynomials at approximate it only on . It ends by pointing out the place of the unit in the course, which it closes, and where its results are used in other courses and in science and technology.
Learning objectives
- Describe the route of the unit and explain what each lesson proves and which result of Calculus I or of the earlier units it relies on.
- Explain why a function with derivatives of all orders may fail to be the sum of its Taylor series, and why uniform convergence is needed in order to differentiate, integrate and pass to the limit in a series of functions.
- Place the results of the unit in their historical context, from Gregory, Newton and Taylor to Lagrange, Cauchy, Abel and Weierstrass, and distinguish the order in which they were discovered from the order in which they are proved.
- Point out the place of the unit in the course, as the close of what sequences and numerical series opened, and where the Taylor polynomial and power series intervene in other courses and in scientific and technical applications.
Entry assessment
Before working through the unit, it is advisable to measure the starting point. This self-assessment is not graded: it serves to know where to pause. The first six questions cover, in order, the six content lessons of the unit, one each, and measure a skill (computing a limit with the product of two expansions, choosing the order of a Taylor polynomial with the bound for the remainder, deciding with the supremum where a convergence is uniform, finding the set of convergence of a power series with its endpoints, deciding where a known series represents a composite function, approximating an integral without an elementary antiderivative with a series and the bound of the Leibniz test), not a statement to be recalled; the same skills are measured again at the end, in the synthesis lesson, with other data. The last five check what this unit takes as known from Calculus I and from the earlier units: higher-order derivatives, L’Hôpital’s rule applied repeatedly, Lagrange’s mean value theorem, the first mean value theorem for integrals and the root and ratio tests.
Before starting the unit, measure your starting point. It is a short self-assessment: it is not graded and its result is stored only in your browser.
Question 1What is the value of ?
The term of the sine has been omitted: in a product, each factor must be expanded up to the order sought, here .The term of the exponential has been omitted, which, multiplied by the of the sine, contributes .has been replaced by its equivalent inside a difference, and what is lost is precisely what matters; equivalents are substituted in factors, not in summands.By the proposition on expansions at zero of the elementary functions, and , and whatever is is also . By the proposition on the sum and product of expansions, the polynomial part of order of the product is that of the product of the two polynomial parts without the terms of degree greater than : . The numerator is , and the quotient tends to .Review: The Taylor polynomial
Question 2 is approximated with the Taylor polynomial of at , computed at . What is the smallest order for which the bound of the Lagrange remainder guarantees an error less than ?
This is what results if the derivative of order is bounded by , like that of the sine; that of the logarithm grows like , and the bound loses the factorial of the denominator.This is what results if the derivative of order is bounded by instead of by : the bound loses the factor .There is no need to know the intermediate point: a bound for the derivative of order at all the points between and suffices.By induction, the derivative of order of is on , and for strictly between and its absolute value is less than , because . By the corollary on the bound for the Taylor remainder, with , the error does not exceed , which equals for , more than , and for , less.Question 3Let for and each natural number . The sequence tends to at each point of . On which of these sets does it converge uniformly to ?
Each tends to , but the index depends on : at the function takes the value , so that the supremum of its values on is not below and does not tend to .The problem is not at but near it: the point , where takes the value , belongs to for every .The maximum on equals , but the supremum that decides is that of the set on which convergence is asked for: on it is , which tends to .At the terms are zero; for , since by the properties of the exponential, tends to by the orders of growth theorem, with base . The derivative is , which is not positive for : by the theorem on monotonicity from the sign of the derivative, decreases on , and the supremum of its values there is its value at , , which tends to by the same theorem. By the proposition on uniform convergence via the supremum, the convergence is uniform on . The index from which differs from by less than serves for all at once.Question 4On what set of real numbers does the series converge?
It is the interval of convergence, but the endpoints are studied separately: at the series is alternating and converges by the Leibniz test.At the terms tend to , but the series diverges: it is the series of reciprocals of powers with exponent , which does not exceed .The radius is the reciprocal of the limit of the ratio of the coefficients, , and not that limit.It is a power series with center , with coefficients , and the ratio tends to . By the corollary on the radius by the ratio, : the series converges absolutely on and diverges outside . At it is the series of , which diverges by the proposition on the series of reciprocals of powers, with exponent ; at it is that of , which converges by the Leibniz test, because decreases and tends to .Review: Power series
Question 5The lesson on the Taylor series of the elementary functions proves that for every in . On what set of real numbers does hold?
It is the set where the series of the logarithm holds for the variable ; on substituting , the endpoint corresponds to and to , and both are included.At the series is the alternating harmonic series, which converges, and its sum is , the value of there.Having derivatives of all orders is not enough: for the terms of the series do not tend to , and the series diverges although the function is defined and differentiable there.With , the equality holds if lies in , and since , this happens if and only if , that is, if . If , the absolute value of the term, , tends to , because its reciprocal, , tends to by the orders of growth theorem, with base , and by the proposition on the necessary condition for convergence the series diverges. At the sum is .Question 6On substituting in the series of the cosine and integrating term by term, is written as an alternating series. How many terms of that series must be added, at least, for the error bound of the Leibniz test to guarantee an error less than ?
The first omitted term would be , greater than : the first two terms give , and the integral equalsThe Leibniz test bounds the error by the first omitted term, , and not by the last term added, : three terms suffice.The series does not need an antiderivative: it gives the integral with any desired error, and the Leibniz test bounds it in advance.The lesson on the Taylor series of the elementary functions proves that for every real ; with , the series is a power series in that converges at every point, with infinite radius. By the theorem on term-by-term integration of a power series, , with , which decrease strictly and tend to . By the Leibniz test, with the index shifted, the error of adding the first terms is less than the first omitted term, : exceeds , and does not. With three terms,Question 7From earlier unitsLet , for . What is the value of for each natural number ?
The sign is missing: each differentiation of a power with negative exponent changes the sign.The factor that the chain rule contributes at each differentiation is missing.The factorial is missing: is the derivative of order at divided by .By induction, : for it is the chain rule, and on differentiating the formula of order the chain rule gives the factor , which leads to that of order . At one gets . Dividing by gives : the coefficients of the geometric progression with common ratio .Question 8From earlier unitsLet be a function twice differentiable on an open interval containing and three times differentiable at , with , and . What is the value of ?
The factorial is missing: on differentiating the denominator three times, appears.One has divided by , the exponent, and not by : the denominator is differentiated three times, and each differentiation contributes a factor.L’Hôpital’s rule is applied only twice; the last stage is the difference quotient of at , which requires the third derivative only at , and not its continuity.These are the hypotheses of the corollary on vanishing derivatives at a point, with and : the limit is . L’Hôpital’s rule is applied twice, on both sides of , and at the end what remains is the difference quotient of at divided by .Review: L’Hôpital’s rule
Question 9From earlier unitsFor , what inequality does Lagrange’s mean value theorem give when applied to the exponential on ?
The inequalities are reversed: the exponential increases, and lies between and , not below .The intermediate point is not in general the midpoint: for , andIt is the inequality opposite to the true one: the exponential lies above its tangent line at , .The exponential is continuous on and differentiable on , with derivative : there exists in with . Since the exponential is strictly increasing, , and multiplying by gives the inequality. The point is unknown, but it is bounded: this is the idea of the Lagrange remainder.Question 10From earlier unitsLet be a function continuous on . What does the first mean value theorem for integrals ensure about ?
The integral of the weight is missing: with , the integral equals , and not .The weight favors the points near , but it is not concentrated there: with , the integral equals , and not .The sign hypothesis falls on the weight, , which is not negative; need only be continuous.The theorem is applied with the continuous function and the weight , integrable and nonnegative on : there exists in such that the integral is , and by Barrow’s rule the last integral equals . It is the move that leads from the remainder in integral form to the Lagrange remainder.Question 11From earlier unitsLet if is even and if is odd. What do the root and ratio tests decide about the series ?
The ratio is not bounded: it does not stay below a number less than from any index on, and its limit superior is .The ratio test gives divergence if the ratio is greater than or equal to from some index on; here it is less than at all even indices, and the series converges.The limit superior of the roots is , less than : the root test decides. By the corollary “The root test decides whenever the ratio test decides”, from that lesson, the root test decides whenever the ratio test does so by its limit superior or by its limit inferior, and here it decides, moreover, where the ratio test does not.The th root of equals if is even and if it is odd, and its limit superior is : by the root test, the series converges. The ratio equals if is odd, which grows without bound, and if is even, which is less than : its limit superior is , and it is not greater than or equal to from any index on. Neither the convergence item nor the divergence item of the ratio test applies. The Cauchy-Hadamard theorem will do the same with the coefficients of a power series.Review: Ratio and root tests
Loading the content…
Completed
Sign in to save your progress