Skip to content
Topos Uranos

Antiderivatives and integration: synthesis and challenges

Summary

This lesson closes the unit on antiderivatives and integration techniques. It gathers into a single deductive chain what the seven content lessons proved: from the corollary of Calculus I according to which two functions with the same derivative on an interval differ by a constant come the antiderivative, the indefinite integral and the table of basic antiderivatives; the chain rule and the product rule, read backward, give the change of variable and integration by parts; and from these two techniques, together with the trigonometric identities and Bézout’s identity, come the families of integrands that reduce to rational functions, whose antiderivatives are written with rational functions, logarithms and arctangents. It explains how the unit rests on Calculus I and on Algebra and Geometry, and what it leaves to the units that follow: the proof of the fundamental theorem of algebra, to the unit on sequences, and the existence of antiderivatives for every continuous function, to the fundamental theorem of calculus. 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 Algebra and Geometry each one depends on and in which proofs the change of variable, the change by the inverse function and the antiderivatives of rational functions are reused.
  2. Distinguish what the unit admits from what it proves, and recognize the hypotheses of each technique, in particular the interval on which each antiderivative holds.
  3. Relate the unit to Calculus I, to Algebra and Geometry and to the units on sequences and on the definite integral.
  4. Solve non-routine problems that combine several techniques, decide on which interval an antiderivative holds and extend it to the whole real line when possible.

What the unit achieved

Calculus I proved the differentiation rules; this unit traveled them backward. The starting point was a single corollary of the mean value theorem: if two functions have the same derivative on an interval, they differ by a constant. Written for two antiderivatives FF and GG of the same function on an interval II:

(∀x∈I)(G′(x)=F′(x))→(∃C∈R)(∀x∈I)(G(x)=F(x)+C)(\forall x \in I)\left(G'(x) = F'(x)\right) \rightarrow (\exists C \in \mathbb{R})(\forall x \in I)\left(G(x) = F(x) + C\right)

If this is so, it suffices to find one antiderivative to know them all, and it suffices to differentiate to check that a function is one. From this follows the method of the whole unit: each technique is a theorem whose proof consists in differentiating the result with a rule of Calculus I, and the proofs cite those rules instead of redoing them. Now, the corollary holds only on an interval; therefore, each antiderivative was given with the interval on which it holds, and on a split domain, with one constant for each interval.

The foundation and the table

The lesson on antiderivatives and the indefinite integral defined the antiderivative of a function on an interval, with the one-sided derivatives at the endpoints that belong to it, and proved with the corollary on functions with the same derivative that two antiderivatives on an interval differ by a constant, that a condition at a point fixes a single antiderivative and that, on a union of disjoint open intervals, there are as many constants as intervals, independent of one another, as ln⁡∣x∣\ln \abs{x} on R∖{0}\mathbb{R} \setminus \{0\} shows. It defined the indefinite integral as the family of antiderivatives on an interval, so that the equals sign between two indefinite integrals is an equality between families, and proved by differentiation linearity and the table of basic antiderivatives: the powers, with the interval corresponding to each exponent, the logarithm, the exponentials, the trigonometric functions and their inverses, and the hyperbolic functions and their inverses. It closed with Darboux’s theorem: the sign function has no antiderivative on the real line, although it has one on each half-line, and no function with a jump or removable discontinuity has an antiderivative on an open interval containing that point. That every continuous function has an antiderivative was not used: it is the fundamental theorem of calculus, from the unit on the definite integral.

The two general techniques

The lesson on integration by change of variable read the chain rule backward: if FF is an antiderivative of ff, then F(g(x))F\left(g(x)\right) is one of f(g(x))g′(x)f\left(g(x)\right) g'(x), without gg having to be injective; that theorem justifies writing du=g′(x) dx\mathrm{d}u = g'(x) \, \mathrm{d}x, and gives the linear change of variable, the antiderivatives with a parameter and the three forms g′g\frac{g'}{g}, (g(x))rg′(x)\left(g(x)\right)^r g'(x) and eg(x)g′(x)e^{g(x)} g'(x). When the substitution is written in the opposite direction, x=h(t)x = h(t), the theorem on change of variable by the inverse function requires that h′h' not vanish on an interval: then hh is strictly monotonic, its inverse is differentiable and the antiderivative goes back to the original variable. The lesson on integration by parts read the product rule backward, which exchanges one problem for another; repeated, it gave the tabular method, proved by induction, the cyclic integrals and the reduction formulas, among them the one for the powers of 1a2+x2\frac{1}{a^2 + x^2}. In the cyclic integrals and the reductions, solving for the integral, as the computation suggests, presupposes that the antiderivative exists; since without the fundamental theorem of calculus this is not known, the proof differentiates the candidate function. With the mean value theorem, the lesson also obtained the antiderivative of an inverse function, without assuming that the latter is differentiable.

The families

The four lessons that follow applied the two techniques to families of integrands. The lesson on trigonometric integrals split off, when an exponent is odd, a factor that is, up to sign, the derivative of the other function, and when both are even, lowered the degree with the power-reduction formulas, down to a sum of cosines; it proved the antiderivative of the secant, with one constant on each interval of its domain, the reduction of its powers, the products of powers of the tangent and the secant and, with the product-to-sum formulas, the products of sines and cosines with different frequencies. The lesson on trigonometric and hyperbolic substitutions eliminated the roots a2−x2\sqrt{a^2 - x^2}, a2+x2\sqrt{a^2 + x^2} and x2−a2\sqrt{x^2 - a^2} as cases of the change by the inverse function, choosing the interval of the new variable so that the substitution is invertible and the sign of the root known; it compared the trigonometric version with the hyperbolic one and, by completing the square, reduced every root of a polynomial of degree two with nonzero discriminant to one of the three. The lesson on integration of rational functions proved what Algebra and Geometry had admitted: Bézout’s identity for coprime polynomials and, with it, the existence and uniqueness of the partial fraction decomposition, also with irreducible quadratic factors; it integrated each partial fraction and concluded that every rational function has, on each interval without roots of the denominator, an antiderivative made up of a polynomial, a rational function, logarithms and arctangents, whose rational part the Hermite-Ostrogradsky method gives without knowing the roots of the denominator. Finally, the lesson on integration by rationalization brought other families to that one: the rational functions of sine and cosine, with the tangent half-angle substitution or, if parity allows, with shorter substitutions, and the joining of the antiderivatives of adjacent intervals; the roots of linear fractional functions; the roots of polynomials of degree two, with the three Euler substitutions; and the differential binomials, in Chebyshev’s three cases.

What is admitted, what is proved and what remains

The balance of the unit can be summarized as follows:

  • The results of Calculus I are admitted, in particular the differentiation rules, the mean value theorem and its corollaries, Darboux’s theorem and Bolzano’s theorem; those of Algebra and Geometry, such as the polynomial division algorithm and the trigonometric identities; and the fundamental theorem of algebra, which Algebra and Geometry admitted in the lesson on roots of higher-degree polynomials and whose proof is announced for the lesson on subsequences and the Bolzano-Weierstrass theorem.
  • All the techniques are proved, each by differentiating its result, together with what Algebra and Geometry had left admitted: Bézout’s identity and the partial fraction decomposition with irreducible quadratic factors.
  • Three tools are reused instead of repeating arguments: the change of variable, which proves the formulas of almost all the families; the change by the inverse function, in which every substitution x=h(t)x = h(t) ends; and the corollary on the antiderivatives of rational functions, in which all the rationalizations end.
  • What demands more than the derivative is left for later: that every continuous function has an antiderivative, that the definite integral is computed with an antiderivative, and the Wallis integrals and the orthogonality of sines and cosines, in the unit on the definite integral; and that certain antiderivatives, such as those of e−x2e^{-x^2} or those of differential binomials outside Chebyshev’s cases, are not elementary, which is mentioned as a fact from later courses.

The following figure gathers the three 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 or from Algebra and Geometry, 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

  • AFunctions with the same derivative. Proved in Calculus I with the mean value theorem: if ff and gg are continuous on an interval II and differentiable at each interior point of II, with f′(x)=g′(x)f'(x) = g'(x) at each of them, there exists a number CC such that f(x)=g(x)+Cf(x) = g(x) + C for every xx in II. It is the foundation of the unit. Admitted without proof in this unit.
  • BDerivatives and continuity. Proved in Calculus I: the differentiation rules for the sum, the product and the quotient, the chain rule, the derivative of the inverse function, the derivatives of the elementary functions and of their inverses, and the local character of the derivative; and, for continuous functions, Bolzano’s theorem and the fact that the image of an interval is an interval. Admitted without proof in this unit.
  • CMean value and Darboux. Proved in Calculus I: Lagrange’s mean value theorem; the limit of the derivative, which extends an antiderivative up to an endpoint; Darboux’s theorem, according to which a derivative takes all intermediate values; and the fact that a derivative without zeros on an interval gives a strictly monotonic function. Admitted without proof in this unit.
  • DAlgebra and Geometry. Admitted from Algebra and Geometry: the polynomial division algorithm, the fundamental trigonometric identities and the product-to-sum formulas, the bijective restrictions of the trigonometric functions and their inverses, and the hyperbolic functions and their inverses. Admitted without proof in this unit.
  • EFundamental theorem of algebra. It is admitted, as in Algebra and Geometry, and will be proved in the unit on sequences: every polynomial of positive degree has a complex root; consequently, every real polynomial is the product of a constant and of linear and irreducible quadratic factors. Admitted without proof in this unit.
  • FChange of variable. If FF is an antiderivative of ff on an interval JJ and gg is differentiable on an interval II with g(I)⊆Jg(I) \subseteq J, then F(g(x))F\left(g(x)\right) is an antiderivative of f(g(x))g′(x)f\left(g(x)\right) g'(x) on II: the tool that proves the formulas of the families. Established in: 6. Integration by change of variable.
  • GChange by the inverse. If hh is differentiable on an interval TT, with h′(t)≠0h'(t) \neq 0 on all of TT, and GG is an antiderivative of f(h(t))h′(t)f\left(h(t)\right) h'(t) on TT, then G(h−1(x))G\left(h^{-1}(x)\right) is an antiderivative of ff on h(T)h(T): the stage in which every substitution ends. Established in: 9. Change of variable by the inverse function.
  • HRational antiderivatives. Every rational function has an antiderivative on each interval without roots of its denominator, and there is one made up of a polynomial, a rational function, logarithms and arctangents: the destination of every rationalization. Established in: 26. The antiderivatives of rational functions.
  1. Segment 1 The foundation and the table

    • Define the antiderivative on an interval and prove, with the corollary on functions with the same derivative, that two antiderivatives differ by a constant, one for each interval if the domain is split.
    • Read the indefinite integral as a family and prove by differentiation linearity and the table of basic antiderivatives.
    • Show with Darboux’s theorem that a function with a jump has no antiderivative.
    1. Antiderivatives on an interval. If FF is an antiderivative of ff on an interval II, that is, if F′(x)=f(x)F'(x) = f(x) for every xx in II, with the one-sided derivatives at the endpoints that belong to II, a function GG is an antiderivative of ff on II if and only if there exists a number CC such that G(x)=F(x)+CG(x) = F(x) + C for every xx in II; given x0x_0 in II and a number y0y_0, there is a unique antiderivative GG with G(x0)=y0G(x_0) = y_0.

      Rests on: A (Functions with the same derivative), B (Derivatives and continuity). Proved in: Antiderivatives and the indefinite integral.

    2. Domains that are not intervals. If DD is the union of open intervals IkI_k, pairwise disjoint, and F′(x)=f(x)F'(x) = f(x) at each point of DD, the functions differentiable on DD with derivative ff are those equal to F(x)+CkF(x) + C_k on each IkI_k, with arbitrary and independent constants; thus ln⁡∣x∣\ln \abs{x} carries one constant on each half-line.

      Uses: 1. Antiderivatives on an interval. Rests on: B (Derivatives and continuity). Proved in: Antiderivatives and the indefinite integral.

    3. Linearity of antiderivatives. If FF and GG are antiderivatives of ff and of gg on an interval II, and α\alpha and β\beta are real numbers, αF+βG\alpha F + \beta G is an antiderivative of αf+βg\alpha f + \beta g on II; if α\alpha and β\beta are not both zero, ∫(αf(x)+βg(x)) dx=α∫f(x) dx+β∫g(x) dx\int \left(\alpha f(x) + \beta g(x)\right) \, \mathrm{d}x = \alpha \int f(x) \, \mathrm{d}x + \beta \int g(x) \, \mathrm{d}x as an equality of families.

      Uses: 1. Antiderivatives on an interval. Rests on: B (Derivatives and continuity). Proved in: Antiderivatives and the indefinite integral.

    4. Basic antiderivatives. On the intervals where they hold, ∫xr dx=xr+1r+1+C\int x^r \, \mathrm{d}x = \frac{x^{r + 1}}{r + 1} + C for r≠−1r \neq -1 and ∫1x dx=ln⁡∣x∣+C\int \frac{1}{x} \, \mathrm{d}x = \ln \abs{x} + C, as well as the antiderivatives of exe^x, axa^x, cos⁡x\cos x, sin⁡x\sin x, sec⁡2x\sec^2 x, 11+x2\frac{1}{1 + x^2}, 11−x2\frac{1}{\sqrt{1 - x^2}}, of the hyperbolic functions and of 1x2+1\frac{1}{\sqrt{x^2 + 1}}, 1x2−1\frac{1}{\sqrt{x^2 - 1}} and 11−x2\frac{1}{1 - x^2}, each obtained by reading backward a derivative from Calculus I.

      Rests on: B (Derivatives and continuity). Proved in: Antiderivatives and the indefinite integral.

    5. Functions without an antiderivative. The sign function has no antiderivative on R\mathbb{R}, although it has one on each open half-line; and a function defined on an open interval, with a removable or jump discontinuity at a point of it, has no antiderivative on that interval.

      Uses: 1. Antiderivatives on an interval. Rests on: C (Mean value and Darboux). Proved in: Antiderivatives and the indefinite integral.

  2. Segment 2 The two general techniques

    • Read the chain rule backward: the direct change of variable, which does not require injectivity, the linear change and the forms g′g\frac{g'}{g}, (g(x))rg′(x)\left(g(x)\right)^r g'(x) and eg(x)g′(x)e^{g(x)} g'(x).
    • Change the variable by a function whose derivative has no zeros and go back to the original variable with its inverse.
    • Read the product rule backward: integration by parts, the tabular method, the cyclic integrals and the reduction formulas, proved by differentiating the candidate.
    • Obtain the antiderivative of an inverse function without assuming it differentiable.
    1. Integration by change of variable. If FF is an antiderivative of ff on an interval JJ and gg is differentiable on an interval II with g(I)⊆Jg(I) \subseteq J, then F∘gF \circ g is an antiderivative of (f∘g) g′(f \circ g) \, g' on II: ∫f(g(x)) g′(x) dx=F(g(x))+C\int f(g(x)) \, g'(x) \, \mathrm{d}x = F(g(x)) + C; gg is not required to be injective.

      Uses: 1. Antiderivatives on an interval. Rests on: B (Derivatives and continuity). Proved in: Integration by change of variable.

    2. Linear change and antiderivatives with a parameter. If FF is an antiderivative of ff on an interval JJ and c≠0c \neq 0, then ∫f(cx+b) dx=1cF(cx+b)+C\int f(cx + b) \, \mathrm{d}x = \frac{1}{c} F(cx + b) + C on the interval of those xx for which cx+bcx + b lies in JJ; in particular, for a>0a > 0, ∫1a2+x2 dx=1aarctan⁡xa+C\int \frac{1}{a^2 + x^2} \, \mathrm{d}x = \frac{1}{a} \arctan \frac{x}{a} + C on R\mathbb{R} and ∫1a2−x2 dx=arcsin⁡xa+C\int \frac{1}{\sqrt{a^2 - x^2}} \, \mathrm{d}x = \arcsin \frac{x}{a} + C on (−a,a)\ioo{-a}{a}, with analogous formulas for 1x2+a2\frac{1}{\sqrt{x^2 + a^2}}, 1x2−a2\frac{1}{\sqrt{x^2 - a^2}} and 1a2−x2\frac{1}{a^2 - x^2}.

      Uses: 6. Integration by change of variable, 3. Linearity of antiderivatives, 4. Basic antiderivatives. Rests on: F (Change of variable). Proved in: Integration by change of variable.

    3. Three forms of the change of variable. If gg is differentiable on an interval II: ∫g′(x)g(x) dx=ln⁡∣g(x)∣+C\int \frac{g'(x)}{g(x)} \, \mathrm{d}x = \ln \abs{g(x)} + C if gg does not vanish on II; ∫eg(x)g′(x) dx=eg(x)+C\int e^{g(x)} g'(x) \, \mathrm{d}x = e^{g(x)} + C; and ∫(g(x))rg′(x) dx=(g(x))r+1r+1+C\int \left(g(x)\right)^r g'(x) \, \mathrm{d}x = \frac{\left(g(x)\right)^{r + 1}}{r + 1} + C for r≠−1r \neq -1 if gg is positive on II, with no condition if rr is a nonnegative integer and with gg without zeros if rr is an integer less than −1-1.

      Uses: 6. Integration by change of variable, 4. Basic antiderivatives. Rests on: F (Change of variable), B (Derivatives and continuity). Proved in: Integration by change of variable.

    4. Change of variable by the inverse function. If hh is differentiable on an interval TT, with h′(t)≠0h'(t) \neq 0 for every tt in TT, and I=h(T)I = h(T), then hh is strictly monotonic, II is an interval and h−1h^{-1} is differentiable on II; and if GG is an antiderivative of (f∘h) h′(f \circ h) \, h' on TT, then G∘h−1G \circ h^{-1} is an antiderivative of ff on II: ∫f(x) dx=G(h−1(x))+C\int f(x) \, \mathrm{d}x = G\left(h^{-1}(x)\right) + C.

      Rests on: B (Derivatives and continuity), C (Mean value and Darboux). Proved in: Integration by change of variable.

    5. Integration by parts. If uu and vv are differentiable on an interval II and HH is an antiderivative of u′vu' v on II, then uv−Hu v - H is an antiderivative of uv′u v' on II: ∫u(x) v′(x) dx=u(x) v(x)−∫u′(x) v(x) dx\int u(x) \, v'(x) \, \mathrm{d}x = u(x) \, v(x) - \int u'(x) \, v(x) \, \mathrm{d}x; in particular, uv′u v' has an antiderivative on II if and only if u′vu' v does.

      Uses: 1. Antiderivatives on an interval. Rests on: B (Derivatives and continuity). Proved in: Integration by parts.

    6. Tabular method. If PP is a polynomial of degree at most nn and G0,G1,…,Gn+1G_0, G_1, \dots, G_{n + 1} are functions defined on an interval II, each an antiderivative of the one before it, then ∫P(x) G0(x) dx=∑k=0n(−1)kP(k)(x) Gk+1(x)+C\int P(x) \, G_0(x) \, \mathrm{d}x = \sum_{k = 0}^{n} (-1)^k P^{(k)}(x) \, G_{k + 1}(x) + C on II.

      Uses: 10. Integration by parts, 3. Linearity of antiderivatives. Proved in: Integration by parts.

    7. Cyclic integrals. If aa and bb are not both zero, on R\mathbb{R}, ∫eaxsin⁡(bx) dx=eax(asin⁡(bx)−bcos⁡(bx))a2+b2+C\int e^{ax} \sin(bx) \, \mathrm{d}x = \frac{e^{ax} \left(a \sin(bx) - b \cos(bx)\right)}{a^2 + b^2} + C and ∫eaxcos⁡(bx) dx=eax(acos⁡(bx)+bsin⁡(bx))a2+b2+C\int e^{ax} \cos(bx) \, \mathrm{d}x = \frac{e^{ax} \left(a \cos(bx) + b \sin(bx)\right)}{a^2 + b^2} + C; solving for the integral, as two integrations by parts suggest, presupposes that the antiderivative exists, and the proof differentiates the candidate.

      Uses: 1. Antiderivatives on an interval. Rests on: B (Derivatives and continuity). Proved in: Integration by parts.

    8. Reduction formulas. For n≥2n \geq 2, if GG is an antiderivative of sin⁡n−2x\sin^{n - 2} x on R\mathbb{R}, the function −1nsin⁡n−1xcos⁡x+n−1nG(x)-\frac{1}{n} \sin^{n - 1} x \cos x + \frac{n - 1}{n} G(x) is one of sin⁡nx\sin^n x; and for a>0a > 0 and n≥1n \geq 1, if GG is an antiderivative of 1(a2+x2)n\frac{1}{\left(a^2 + x^2\right)^n} on R\mathbb{R}, the function x2na2(a2+x2)n+2n−12na2G(x)\frac{x}{2n a^2 \left(a^2 + x^2\right)^n} + \frac{2n - 1}{2n a^2} G(x) is one of 1(a2+x2)n+1\frac{1}{\left(a^2 + x^2\right)^{n + 1}}, so that each 1(a2+x2)n\frac{1}{\left(a^2 + x^2\right)^n} has on R\mathbb{R} an antiderivative Rn(x)+cnarctan⁡xaR_n(x) + c_n \arctan \frac{x}{a}, with RnR_n rational without real poles.

      Uses: 7. Linear change and antiderivatives with a parameter. Rests on: B (Derivatives and continuity). Proved in: Integration by parts.

    9. Antiderivative of the inverse function. If ff is continuous and strictly monotonic on an interval II, J=f(I)J = f(I) and FF is an antiderivative of ff on II, then JJ is an interval and ∫f−1(x) dx=xf−1(x)−F(f−1(x))+C\int f^{-1}(x) \, \mathrm{d}x = x f^{-1}(x) - F\left(f^{-1}(x)\right) + C on JJ, without ff having to be differentiable.

      Rests on: B (Derivatives and continuity), C (Mean value and Darboux). Proved in: Integration by parts.

  3. Segment 3 The families

    • Integrate the powers and products of trigonometric functions, splitting off a factor that is a derivative or lowering the degree, and the secant with one constant on each interval.
    • Eliminate the roots of polynomials of degree two with trigonometric and hyperbolic substitutions, choosing the interval of the new variable.
    • Prove Bézout’s identity and the partial fraction decomposition, and conclude that every rational function has an antiderivative on each interval without roots of the denominator.
    • Rationalize the rational functions of sine and cosine, the roots of linear fractional functions and of polynomials of degree two, and the differential binomials, and join the antiderivatives of adjacent intervals.
    1. An odd exponent. If mm and jj are nonnegative integers and PP is an antiderivative on R\mathbb{R} of the polynomial um(1−u2)ju^m \left(1 - u^2\right)^j, then, on R\mathbb{R}, ∫sin⁡mxcos⁡2j+1x dx=P(sin⁡x)+C\int \sin^m x \cos^{2j + 1} x \, \mathrm{d}x = P(\sin x) + C and ∫cos⁡mxsin⁡2j+1x dx=−P(cos⁡x)+C\int \cos^m x \sin^{2j + 1} x \, \mathrm{d}x = -P(\cos x) + C.

      Uses: 6. Integration by change of variable. Rests on: F (Change of variable), D (Algebra and Geometry). Proved in: Trigonometric integrals.

    2. Linearization of even powers. If jj and kk are nonnegative integers and N=j+kN = j + k, there exist real numbers c0,c1,…,cNc_0, c_1, \dots, c_N such that sin⁡2jxcos⁡2kx=c0+∑r=1Ncrcos⁡(2rx)\sin^{2j} x \cos^{2k} x = c_0 + \sum_{r = 1}^{N} c_r \cos(2rx) for every real xx, and then ∫sin⁡2jxcos⁡2kx dx=c0x+∑r=1Ncr2rsin⁡(2rx)+C\int \sin^{2j} x \cos^{2k} x \, \mathrm{d}x = c_0 x + \sum_{r = 1}^{N} \frac{c_r}{2r} \sin(2rx) + C on R\mathbb{R}.

      Uses: 7. Linear change and antiderivatives with a parameter. Rests on: D (Algebra and Geometry). Proved in: Trigonometric integrals.

    3. The secant and its powers. On each interval Ik=(−π2+kπ,π2+kπ)I_k = \ioo{-\frac{\pi}{2} + k\pi}{\frac{\pi}{2} + k\pi}, with kk an integer, ∫sec⁡x dx=ln⁡∣sec⁡x+tan⁡x∣+C\int \sec x \, \mathrm{d}x = \ln \abs{\sec x + \tan x} + C, with one constant per interval in the domain of the secant; and for n≥2n \geq 2, if GG is an antiderivative of sec⁡n−2x\sec^{n - 2} x on IkI_k, the function 1n−1sec⁡n−2xtan⁡x+n−2n−1G(x)\frac{1}{n - 1} \sec^{n - 2} x \tan x + \frac{n - 2}{n - 1} G(x) is one of sec⁡nx\sec^n x.

      Uses: 2. Domains that are not intervals, 8. Three forms of the change of variable. Rests on: B (Derivatives and continuity), D (Algebra and Geometry). Proved in: Trigonometric integrals.

    4. Powers of the tangent and the secant. On each IkI_k, with mm and ii nonnegative integers: if j≥1j \geq 1 and PP is an antiderivative on R\mathbb{R} of um(1+u2)j−1u^m \left(1 + u^2\right)^{j - 1}, then ∫tan⁡mxsec⁡2jx dx=P(tan⁡x)+C\int \tan^m x \sec^{2j} x \, \mathrm{d}x = P(\tan x) + C; and if n≥1n \geq 1 and QQ is an antiderivative on R\mathbb{R} of (u2−1)iun−1\left(u^2 - 1\right)^i u^{n - 1}, then ∫tan⁡2i+1xsec⁡nx dx=Q(sec⁡x)+C\int \tan^{2i + 1} x \sec^n x \, \mathrm{d}x = Q(\sec x) + C.

      Uses: 6. Integration by change of variable. Rests on: F (Change of variable), D (Algebra and Geometry). Proved in: Trigonometric integrals.

    5. The three trigonometric substitutions. For a>0a > 0: x=asin⁡tx = a \sin t on (−π2,π2)\ioo{-\frac{\pi}{2}}{\frac{\pi}{2}} gives a2−x2=acos⁡t\sqrt{a^2 - x^2} = a \cos t; x=atan⁡tx = a \tan t on the same interval gives a2+x2=asec⁡t\sqrt{a^2 + x^2} = a \sec t; and x=asec⁡tx = a \sec t gives x2−a2=atan⁡t\sqrt{x^2 - a^2} = a \tan t on (0,π2)\ioo{0}{\frac{\pi}{2}} and −atan⁡t-a \tan t on (π2,π)\ioo{\frac{\pi}{2}}{\pi}; in each case the derivative of the substitution does not vanish and its inverse is the corresponding one.

      Rests on: D (Algebra and Geometry). Proved in: Trigonometric and hyperbolic substitutions.

    6. Antiderivatives of the three roots. For a>0a > 0: ∫a2−x2 dx=12xa2−x2+a22arcsin⁡xa+C\int \sqrt{a^2 - x^2} \, \mathrm{d}x = \frac{1}{2} x \sqrt{a^2 - x^2} + \frac{a^2}{2} \arcsin \frac{x}{a} + C on [−a,a]\icc{-a}{a}; ∫x2+a2 dx=12xx2+a2+a22ln⁡(x+x2+a2)+C\int \sqrt{x^2 + a^2} \, \mathrm{d}x = \frac{1}{2} x \sqrt{x^2 + a^2} + \frac{a^2}{2} \ln\left(x + \sqrt{x^2 + a^2}\right) + C on R\mathbb{R}; and ∫x2−a2 dx=12xx2−a2−a22ln⁡∣x+x2−a2∣+C\int \sqrt{x^2 - a^2} \, \mathrm{d}x = \frac{1}{2} x \sqrt{x^2 - a^2} - \frac{a^2}{2} \ln \abs{x + \sqrt{x^2 - a^2}} + C on (−∞,−a]\ioc{-\infty}{-a} and on [a,∞)\ico{a}{\infty}, with one constant on each.

      Uses: 7. Linear change and antiderivatives with a parameter, 9. Change of variable by the inverse function, 17. The secant and its powers, 19. The three trigonometric substitutions. Rests on: G (Change by the inverse), C (Mean value and Darboux). Proved in: Trigonometric and hyperbolic substitutions.

    7. The three hyperbolic substitutions. For a>0a > 0: x=asinh⁡tx = a \sinh t on R\mathbb{R} gives a2+x2=acosh⁡t\sqrt{a^2 + x^2} = a \cosh t; x=acosh⁡tx = a \cosh t and x=−acosh⁡tx = -a \cosh t, with t>0t > 0, give x2−a2=asinh⁡t\sqrt{x^2 - a^2} = a \sinh t on (a,∞)\ioo{a}{\infty} and on (−∞,−a)\ioo{-\infty}{-a}; and x=atanh⁡tx = a \tanh t on R\mathbb{R} gives a2−x2=acosh⁡t\sqrt{a^2 - x^2} = \frac{a}{\cosh t}; the inverses are the inverse hyperbolic sine, cosine and tangent.

      Uses: 4. Basic antiderivatives. Rests on: D (Algebra and Geometry). Proved in: Trigonometric and hyperbolic substitutions.

    8. Completing the square. If a≠0a \neq 0 and Δ=b2−4ac\Delta = b^2 - 4ac, then ax2+bx+c=a(x+b2a)2−Δ4aax^2 + bx + c = a \left(x + \frac{b}{2a}\right)^2 - \frac{\Delta}{4a} for every real xx; with u=x+b2au = x + \frac{b}{2a}, the root of a polynomial of degree two reduces to one of the three roots, according to the signs of aa and of Δ\Delta.

      Rests on: D (Algebra and Geometry). Proved in: Trigonometric and hyperbolic substitutions.

    9. Bézout’s identity for polynomials. If PP and QQ are coprime polynomials, there exist polynomials AA and BB such that AP+BQ=1AP + BQ = 1; and two real polynomials, not both zero, without common complex roots are coprime.

      Rests on: D (Algebra and Geometry), E (Fundamental theorem of algebra). Proved in: Integration of rational functions.

    10. Partial fraction decomposition. If Q=c F1k1⋯FsksQ = c \, F_1^{k_1} \cdots F_s^{k_s} is a real polynomial of degree n≥1n \geq 1, a product of linear and irreducible quadratic factors, and PP is zero or of degree less than nn, there exist unique polynomials NijN_{ij}, zero or of degree less than that of FiF_i, such that PQ=∑i=1s∑j=1kiNijFij\frac{P}{Q} = \sum_{i = 1}^{s} \sum_{j = 1}^{k_i} \frac{N_{ij}}{F_i^j} at every real number that is not a root of QQ.

      Uses: 23. Bézout’s identity for polynomials. Rests on: D (Algebra and Geometry), E (Fundamental theorem of algebra). Proved in: Integration of rational functions.

    11. Antiderivatives of the partial fractions. On each half-line (−∞,r)\ioo{-\infty}{r} and (r,∞)\ioo{r}{\infty}, ∫Ax−r dx=Aln⁡∣x−r∣+C\int \frac{A}{x - r} \, \mathrm{d}x = A \ln \abs{x - r} + C and, for j≥2j \geq 2, ∫A(x−r)j dx=−A(j−1)(x−r)j−1+C\int \frac{A}{(x - r)^j} \, \mathrm{d}x = -\frac{A}{(j - 1)(x - r)^{j - 1}} + C; and if p2−4q<0p^2 - 4q < 0 and k=4q−p22k = \frac{\sqrt{4q - p^2}}{2}, on R\mathbb{R}, ∫Mx+Nx2+px+q dx=M2ln⁡(x2+px+q)+2N−Mp2karctan⁡2x+p2k+C\int \frac{Mx + N}{x^2 + px + q} \, \mathrm{d}x = \frac{M}{2} \ln\left(x^2 + px + q\right) + \frac{2N - Mp}{2k} \arctan \frac{2x + p}{2k} + C, with an analogous formula for the powers of the quadratic factor.

      Uses: 4. Basic antiderivatives, 7. Linear change and antiderivatives with a parameter, 8. Three forms of the change of variable, 22. Completing the square. Rests on: F (Change of variable). Proved in: Integration of rational functions.

    12. The antiderivatives of rational functions. If PQ\frac{P}{Q} is a rational function and II is an interval that contains no roots of QQ, PQ\frac{P}{Q} has an antiderivative on II, and there is one that is the sum of a polynomial, of a rational function whose denominator vanishes only at the real roots of QQ, of multiples of ln⁡∣x−r∣\ln \abs{x - r}, one for each real root rr of QQ, and of multiples of ln⁡(x2+px+q)\ln\left(x^2 + px + q\right) and of arctan⁡2x+p2k\arctan \frac{2x + p}{2k}, one of each kind for each irreducible quadratic factor.

      Uses: 3. Linearity of antiderivatives, 13. Reduction formulas, 24. Partial fraction decomposition, 25. Antiderivatives of the partial fractions. Rests on: D (Algebra and Geometry). Proved in: Integration of rational functions.

    13. Hermite-Ostrogradsky method. If QQ is a real polynomial of degree n≥1n \geq 1, PP is zero or of degree less than nn, Q1=gcd⁡(Q,Q′)Q_1 = \operatorname{gcd}(Q, Q') and Q2=QQ1Q_2 = \frac{Q}{Q_1}, there exist unique polynomials P1P_1 and P2P_2, zero or of degree less than that of Q1Q_1 and that of Q2Q_2, such that PQ=(P1Q1)′+P2Q2\frac{P}{Q} = \left(\frac{P_1}{Q_1}\right)' + \frac{P_2}{Q_2}; on each interval without roots of QQ, ∫P(x)Q(x) dx=P1(x)Q1(x)+∫P2(x)Q2(x) dx\int \frac{P(x)}{Q(x)} \, \mathrm{d}x = \frac{P_1(x)}{Q_1(x)} + \int \frac{P_2(x)}{Q_2(x)} \, \mathrm{d}x.

      Uses: 23. Bézout’s identity for polynomials, 24. Partial fraction decomposition, 25. Antiderivatives of the partial fractions. Rests on: B (Derivatives and continuity), E (Fundamental theorem of algebra). Proved in: Integration of rational functions.

    14. The tangent half-angle substitution. If RR is a rational function of two variables and f(x)=R(sin⁡x,cos⁡x)f(x) = R(\sin x, \cos x) is defined on an interval II contained in ((2k−1)π,(2k+1)π)\ioo{(2k - 1)\pi}{(2k + 1)\pi}, with kk an integer, the substitution t=tan⁡x2t = \tan \frac{x}{2} turns the integrand into the rational function R(2t1+t2,1−t21+t2)21+t2R\left(\frac{2t}{1 + t^2}, \frac{1 - t^2}{1 + t^2}\right) \frac{2}{1 + t^2}, which has an antiderivative GG, and G(tan⁡x2)G\left(\tan \frac{x}{2}\right) is an antiderivative of ff on II.

      Uses: 9. Change of variable by the inverse function, 26. The antiderivatives of rational functions. Rests on: G (Change by the inverse), H (Rational antiderivatives), D (Algebra and Geometry). Proved in: Integration by rationalization.

    15. Joining two antiderivatives. If JJ is an open interval, ff is defined on JJ and continuous at a point cc of JJ, and F1F_1 and F2F_2 are antiderivatives of ff on the subintervals of JJ to the left and to the right of cc, with finite limits L1L_1 and L2L_2 at cc, the function equal to F1F_1 on the left, L1L_1 at cc and F2+L1−L2F_2 + L_1 - L_2 on the right is an antiderivative of ff on JJ.

      Rests on: B (Derivatives and continuity), C (Mean value and Darboux). Proved in: Integration by rationalization.

    16. Substitutions by parity. If f(x)=R(sin⁡x,cos⁡x)f(x) = R(\sin x, \cos x) is defined on an interval II: when R(−u,v)=−R(u,v)R(-u, v) = -R(u, v), the substitution t=cos⁡xt = \cos x rationalizes the integrand on II; when R(u,−v)=−R(u,v)R(u, -v) = -R(u, v), the substitution t=sin⁡xt = \sin x; and when R(−u,−v)=R(u,v)R(-u, -v) = R(u, v) and II is contained in an interval (−π2+kπ,π2+kπ)\ioo{-\frac{\pi}{2} + k\pi}{\frac{\pi}{2} + k\pi}, the substitution t=tan⁡xt = \tan x.

      Uses: 6. Integration by change of variable, 9. Change of variable by the inverse function, 26. The antiderivatives of rational functions. Rests on: F (Change of variable), G (Change by the inverse), H (Rational antiderivatives). Proved in: Integration by rationalization.

    17. Root of a linear fractional function. If ad−bc≠0ad - bc \neq 0 and y(x)=ax+bcx+dy(x) = \frac{ax + b}{cx + d} is defined and does not vanish on an interval II, and is positive if nn is even, the substitution t=y(x)nt = \sqrt[n]{y(x)}, with inverse h(t)=dtn−ba−ctnh(t) = \frac{d t^n - b}{a - c t^n}, turns R(x,y(x)n) dxR\left(x, \sqrt[n]{y(x)}\right) \, \mathrm{d}x into a rational function of tt times dt\mathrm{d}t, and its antiderivative, composed with y(x)n\sqrt[n]{y(x)}, is an antiderivative on II.

      Uses: 9. Change of variable by the inverse function, 26. The antiderivatives of rational functions. Rests on: G (Change by the inverse), H (Rational antiderivatives). Proved in: Integration by rationalization.

    18. The Euler substitutions. If Q(x)=ax2+bx+cQ(x) = ax^2 + bx + c, with a≠0a \neq 0 and Δ=b2−4ac≠0\Delta = b^2 - 4ac \neq 0, is positive on an interval II, the function R(x,Q(x))R\left(x, \sqrt{Q(x)}\right) is rationalized on II by t=Q(x)+xat = \sqrt{Q(x)} + x \sqrt{a} if a>0a > 0; by t=Q(x)−cxt = \frac{\sqrt{Q(x)} - \sqrt{c}}{x} if c>0c > 0 and II does not contain 00; and by t=Q(x)x−αt = \frac{\sqrt{Q(x)}}{x - \alpha} if QQ has two real roots α\alpha and β\beta and II does not contain α\alpha.

      Uses: 9. Change of variable by the inverse function, 26. The antiderivatives of rational functions. Rests on: G (Change by the inverse), H (Rational antiderivatives). Proved in: Integration by rationalization.

    19. Chebyshev’s three cases. If mm, nn and pp are rational, with n≠0n \neq 0, aa and bb are nonzero reals and II is an interval contained in (0,∞)\ioo{0}{\infty} on which a+bxn>0a + b x^n > 0, the function xm(a+bxn)px^m \left(a + b x^n\right)^p has an antiderivative on II, obtained from that of a rational function, if pp is an integer, if m+1n\frac{m + 1}{n} is an integer or if p+m+1np + \frac{m + 1}{n} is an integer.

      Uses: 6. Integration by change of variable, 31. Root of a linear fractional function. Rests on: F (Change of variable). Proved in: Integration by rationalization.

Toward sequences and the definite integral. The chain stops where the derivative ceases to suffice. The whole unit found antiderivatives and checked them by differentiation, but it never proved that an arbitrary continuous function has an antiderivative: the fundamental theorem of calculus will do so, in the unit on the definite integral, by uniting the indefinite integral of this unit with the Riemann integral, which is defined with sums and their limits. Before that, the unit on sequences will provide the missing tool, uniform continuity, obtained with the Bolzano-Weierstrass theorem, and will prove the fundamental theorem of algebra that this unit admitted. The techniques of the unit will then pass to the definite integral, which is computed with any antiderivative, and to the areas, the volumes, the arc length and the improper integrals of the later units.