Summary
This lesson opens the sixth unit of the course, devoted to numerical series: sums of infinitely many terms, defined as the limit of the sequence of their partial sums. It presents the route of the unit, which repeats with sums what the unit on sequences did with limits and the unit on improper integrals did with integrals: deciding whether a series converges without knowing its sum, with the Cauchy criterion, comparison, the integral test, the ratio and root tests and their refinements, and the tests for series that converge by cancellation; and deciding which operations preserve the sum, such as rearranging the terms or multiplying two series. It reconstructs the history of these ideas, from Zeno’s dichotomy and Oresme’s harmonic series, around 1350, through Mengoli, the Bernoullis and Euler’s sums of 1734 and 1735, up to Cauchy’s Cours d’analyse of 1821, Abel’s memoir of 1826, the tests of Raabe, Kummer and Bertrand, and the observations of Dirichlet in 1837 and of Riemann, published in 1867, on the order of the terms; it formulates the problem that runs through it, that of giving meaning to an infinite sum, and shows with the alternating harmonic series that the same terms, summed in another order, can give another sum. It ends by pointing out the place of the unit in the course, in particular its role in the unit on power series, and where series are used in other courses of the site 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 the sum of infinitely many terms must be defined as the limit of the partial sums, why that limit may fail to exist even though the terms tend to , and why the order of the terms can change it.
- Place the results of the unit in their historical context, from Zeno and Oresme to Euler, Cauchy, Abel, Dirichlet and Riemann, 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, its parallel with improper integrals, its role in the unit on power series, and where series come into play 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 (summing a telescoping series, deciding convergence by limit comparison with an equivalent infinitesimal, applying the ratio test at the borderline, deciding with Gauss’s test where Raabe’s test does not decide, distinguishing conditional from absolute convergence, recognizing when Mertens’s theorem applies), not a statement to be recalled; the same skills are measured again at the end, in the synthesis lesson, with other data. The last four check what this unit takes as known from Calculus I and from the earlier units: the limits superior and inferior, the squeeze theorem with th roots, an improper integral with a logarithm and L’Hôpital’s rule.
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 ?
It is what remains if only the first summand, , survives, multiplied by ; but the difference skips two indices, and in the partial sums two positive terms survive, and , not one.It is the sum of : the factor of the partial fractions is missing.It is the sum of , whose partial fractions differ by one index, not by two.By partial fractions, , which is with : on subtracting, the summands cancel. Since , by the proposition on telescoping series the series converges and its sum is .Question 2Which of these series converges?
Its terms tend to , but the ratio of to tends to , because at zero: the series diverges, like the harmonic one.The product has degree , but the square root reduces the exponent to : the ratio to is , which tends to , and the series diverges.The ratio to is , which tends to , and the series with exponent diverges: the difference of degrees, , does not exceed .Its terms are not negative. By the theorem on equivalents at zero, as , and since without vanishing, by the sequential characterization of the limit the ratio of to tends to . By the limit comparison test for series and the proposition on the series of reciprocals of powers, with exponent , the series converges.Review: Series with positive terms
Question 3For , what does the ratio test decide about the series ?
That the ratio tends to is not enough for the test to remain silent: at the ratio is greater than for every , the terms increase and do not tend to , and the test gives divergence. The case in which it does not decide is that of ratios less than that tend to .It is the borderline of the series of , whose coefficients are the reciprocals of these: here the factorial of is in the denominator, and the ratio tends to , not to .That the denominator grows faster than the numerator is not enough: the ratio of two consecutive terms tends to , and for it is greater than from some index on, so that the terms increase.The ratio of two consecutive terms is , which tends to . If , its limit superior is less than and the series converges; if , its limit inferior is greater than and it diverges. At the ratio is , which tends to but is greater than for every : by the second item of the ratio test, the series diverges to infinity.Review: Ratio and root tests
Question 4The sequence is given by and . What can be asserted about the series ?
Raabe’s test requires that quantity not to fall below a fixed number from some index on; here it tends to , and no works. It is the case that Raabe’s test leaves open and Gauss’s test decides.The ratio test and Raabe’s test do not decide, but Gauss’s test looks at the following term of the expansion of the ratio: with and a remainder of order , it gives divergence.That the terms decrease is not enough: those of the harmonic series also decrease, and it diverges.By induction, the terms are positive. The ratio equals , and the last summand does not exceed : it is . By Gauss’s test, with and , the series diverges. It is also the corollary on ratios of polynomials, with .Review: Finer convergence tests
Question 5What can be asserted about the series ?
They decrease faster, but not fast enough: the ratio to tends to , and the series of absolute values diverges by limit comparison with the harmonic series.Absolute convergence implies convergence, but not conversely: the alternating signs produce cancellations, and the Leibniz test gives convergence.The Leibniz test gives convergence, not absolute convergence: the alternating harmonic series converges by it, and the harmonic series diverges.For , and are positive and increasing, so that is positive, decreasing and tends to ; by the Leibniz test, applied to the series from on, it converges. The series of absolute values diverges: the ratio of to is , which tends to by the theorem on orders of growth, and by the limit comparison test for series, in the case of an infinite limit, it diverges like the harmonic series.Question 6For which of these pairs of series, with indices from on, does Mertens’s theorem guarantee that the Cauchy product converges to the product of the sums?
Both converge, but neither absolutely, and Mertens’s theorem does not apply; in fact, by the proposition on a divergent Cauchy product, for every , and the product diverges.The first converges absolutely, but the second diverges: Mertens’s theorem requires both to converge. Here all the products are positive and , so that the product diverges.That is the hypothesis of the corollary on the product of two absolutely convergent series, which also gives the absolute convergence of the product; Mertens’s theorem is content with one of the two converging absolutely.The series of the converges absolutely, because it is that of the reciprocals of the squares with the index shifted, and that of the converges by the Leibniz test, although not absolutely. Mertens’s theorem requires exactly this: one absolutely convergent series and another convergent one. That the absolutely convergent one is the second does not matter, because on replacing by the sum that defines does not change.Review: Absolute convergence, rearrangements and products of series
Question 7From earlier unitsLet . What are its limit superior and its limit inferior?
is the supremum of the whole sequence; the limit superior looks only at the tails, and disappears from them.Every bounded sequence has a limit superior and a limit inferior, whether it converges or not: they are the limits of the suprema and of the infima of the tails. They coincide if and only if the sequence converges.They are the terms of each parity, which depend on ; the limit superior and the limit inferior are numbers, the limits of the suprema and of the infima of the tails.The terms of even index are , and those of odd index, , which do not exceed . The supremum of the tail from on is its first term of even index, or , and it tends to . The infimum of the tail is for every : the terms of odd index in the tail are greater than and come as close to it as desired, and those of even index are greater than .Review: Subsequences, Bolzano-Weierstrass, limit superior and inferior
Question 8From earlier unitsWhat is the limit of the sequence ?
It is the limit of and of with fixed; but here the radicand grows like , and its th root stays above .The root of each summand has been taken separately, , which tends to ; the th root does not distribute over a sum.The radicand tends to infinity, but its th root stays between and .By Bernoulli’s inequality, , so that , and on taking the th root, which preserves order by the monotonicity of rational powers, . By the proposition on the th root of a positive number, , and by the squeeze theorem the sequence tends to . It is the computation that the root test requires.Review: Sequences and their limit
Question 9From earlier unitsFor which values of the real number does the improper integral converge?
With the antiderivative is , which grows without bound, although extremely slowly.That the integrand tends to faster than is not enough: with the antiderivative grows without bound.The factor , with , suffices for convergence: the antiderivative has limit .By the chain rule, an antiderivative on is if , and if . The integral from to is the difference of its values, and as , : by the lemma on limits of powers with real exponents, tends to if and to if , and tends to . By the definition of the improper integral, it converges exactly when , with value . It is the integral that the integral test compares with the series with a logarithm in the denominator.Question 10From earlier unitsWhat is ?
has been replaced by its equivalent inside a difference, and the numerator became zero; equivalents are substituted in factors, not in summands.It is the limit of , with the sign of the numerator changed.The factor from the derivative of in the denominator is missing.It is the form zero over zero. By L’Hôpital’s rule for that form, the limit is that of the quotient of the derivatives, , which tends to . With , it says that is equivalent to , a computation that limit comparison for series will use.Review: L’Hôpital’s rule
Loading the content…
Completed
Sign in to save your progress