Summary
This lesson opens the fifth unit of the course, devoted to improper integrals: those of a function on an unbounded interval, or of a function unbounded near an endpoint, defined as the limit of Riemann integrals over closed and bounded intervals that approach the endpoint. It presents the route of the unit, which repeats with integrals what the unit on sequences did with limits: deciding whether the limit exists without computing it, with the Cauchy criterion, comparison, absolute convergence and Dirichlet’s and Abel’s tests; differentiating an integral with respect to a parameter; and computing with these tools three classical values, the integral of over the half-line, the Gaussian integral and Stirling’s formula. It reconstructs the history of these ideas, from Torricelli’s infinite solid of finite volume and Wallis’s Arithmetica infinitorum, through Euler’s interpolation of the factorial, of 1729, and the formula of Stirling and De Moivre, of 1730, to Gauss, Cauchy’s Résumé des leçons of 1823, Frullani’s formula of 1828 and Dirichlet’s memoir of 1829; it formulates the problem that runs through it, that of giving an integral to an unbounded function, and shows with two functions that have an antiderivative that the limit defining it may exist or fail to exist. It ends by pointing out the place of the unit in the course, in particular its parallel with numerical series, and where improper integrals 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 an unbounded function, or an unbounded interval, falls outside the Riemann integral, and why the improper integral, defined as a limit, requires deciding whether that limit exists.
- Place the results of the unit in their historical context, from Torricelli and Wallis to Euler, Stirling, Cauchy, Frullani and Dirichlet, 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 numerical series, and where improper integrals 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 four questions cover, in order, the four content lessons of the unit, one each, and measure a skill (deciding the convergence of an integral improper at both endpoints by comparison, recognizing a conditionally convergent integral, differentiating an integral with respect to a parameter, reducing an integral to the gamma function with a change of variable), 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: the Wallis integrals, the fundamental theorem of calculus, the Cauchy criterion for sequences, L’Hôpital’s rule and equivalent infinitesimals.
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 1Which of these improper integrals converges?
At infinity it converges, like that of ; but near the quotient of the integrand by is , which tends to , and the integral of on diverges. An integral improper at both endpoints converges only if both pieces converge.The integrand is continuous on , so only infinity is improper, and there the quotient of the integrand by is , which tends to : the integral diverges, like that of , even though the integrand tends to .The integrand is not bounded near . By partial fractions, an antiderivative on is , and the integral from to equals , which tends to as .The integral is improper at both endpoints and is split at . For , . Near , the quotient of the integrand by is , which tends to : by the limit comparison test and the proposition on the integrals of powers, with exponent , the integral converges on . At infinity, the quotient by is , which also tends to , and the exponent is greater than : it converges on .Question 2What can be asserted about the integral ?
That a function tends to is not enough for its integral to converge: that of diverges, because on . The bound is of no use, and in fact the integral of the absolute value diverges.Absolute convergence implies convergence, but the converse is false: the signs of the cosine produce cancellations, and Dirichlet’s test ensures that the integral converges even though that of the absolute value diverges.The integrals of the cosine have no limit, but they are bounded, and the factor , decreasing and with limit , damps each arc more than the one before it: this is the situation of Dirichlet’s test, which gives convergence.The function is decreasing on and tends to , and the integrals of the cosine are bounded: . By Dirichlet’s test, the integral converges. On the other hand, ; the integral of converges by the same test, and that of diverges by comparison with , because : by the fundamental inequality of the exponential, , and the logarithm is increasing. By linearity, that of diverges, and by comparison so does that of the absolute value.Question 3Let for real . What is the value of ?
has been differentiated with respect to as , without the factor that the chain rule requires; with one is left with the integral of , which equals .It is , the integral of on , which is computed by parts; the question asks for the derivative of at .A common integrable bound is required for improper integrals, and for the derivative with respect to , as a sufficient condition for uniform convergence. On a closed and bounded interval it is enough that the integrand and its derivative with respect to be continuous on the rectangle, as here.The integrand and its derivative with respect to , , are continuous on the rectangle of the points with and . By the Leibniz rule for integrals with a parameter, one differentiates inside the integral: .Question 4What is the value of ?
It is : the factor contributed by the derivative of the change is missing.With one obtains half the integral of over the half-line, and that integral equals ; the value is that of the whole line.has been replaced by , as if the integrand had the factor that gives , and the factor has been lost.The integral of on converges and equals , by the corollary on the gamma function at one half. The function is strictly increasing on , with continuous derivative and limits and , and ; by the corollary on the increasing change of variable on an open interval, the two integrals coincide.Review: The Gaussian integral, the gamma function and Stirling’s formula
Question 5From earlier unitsWhat is the value of ?
It is the product without the last factor: the recurrence of the even orders ends at , not at .It is : the formula of the odd orders has been used, which ends at and does not contain .has been taken; but is the integral of the constant on , the length of that interval.It is the Wallis integral . By the proposition on the values of the Wallis integrals, , with .Question 6From earlier unitsLet for . What can be asserted about on ?
The value of the integrand at the lower endpoint has been subtracted; but that endpoint is constant and contributes nothing to the derivative.The fundamental theorem of calculus does not ask for a formula for the antiderivative: it ensures that the integral with a variable endpoint of a continuous function is differentiable, with derivative equal to the integrand.The growth of depends on the sign of the integrand, not on its monotonicity: each added piece contributes a positive area, albeit smaller and smaller.The integrand is continuous on the interval ; by the fundamental theorem of calculus, is differentiable at each point, with a one-sided derivative at , and . By the theorem on monotonicity from the sign of the derivative, is strictly increasing: this is what allows the unit to treat the integral of a nonnegative function as the limit of an increasing function.Question 7From earlier unitsWhich of these sequences is a Cauchy sequence?
The differences between consecutive terms tend to , but that is not enough: for every , so with there is no that works.The terms that lead from to are greater than or equal to , and for every : with there is no that works.It is bounded, but two consecutive terms have opposite signs and are apart, which is greater than : with there is no that works.If , each term with satisfies , and when summing from to the terms cancel in pairs: . Given , by the Archimedean property there is a natural number , and for and greater than or equal to one has .Review: Cauchy sequences
Question 8From earlier unitsWhat is ?
The logarithm tends to , but the cube root tends to more strongly: any positive power of beats the logarithm near .It is the coefficient of the quotient of the derivatives, ; the passage to the limit is missing, and the cube root tends to .That the form is indeterminate means that the limits of the factors do not decide that of the product, not that the latter does not exist; L’Hôpital’s rule computes it.The product is written as the quotient of by , which tends to as . By L’Hôpital’s rule for the form infinity over infinity, the limit is that of the quotient of the derivatives, divided by , that is, , which tends to .Review: L’Hôpital’s rule
Question 9From earlier unitsWhat is ?
has been taken; the equivalent is .The factor of the equivalent of is in the numerator, not in the denominator.The numerator is of order , but so is the denominator, because : the quotient tends to a nonzero number.By the theorem on equivalents at zero, , and . By the theorem on substitution of equivalents, factor by factor, the limit is that of . It is the computation that the limit comparison test calls for.Review: Equivalent infinitesimals and infinitely large functions
Loading the content…
Completed
Sign in to save your progress