Skip to content
Topos Uranos

Integration by rationalization

Summary

This lesson gathers the substitutions that turn other families of integrands into rational functions of a new variable, whose antiderivatives are always given by the lesson on the integration of rational functions. The first is the tangent half-angle substitution, t=tan⁡x2t = \tan \frac{x}{2}, which rationalizes every rational function of sine and cosine on each interval lying between two consecutive odd multiples of π\pi; the lesson studies how the antiderivatives on those intervals are joined into an antiderivative on the whole line, and when the parity of the integrand allows the shorter substitutions t=sin⁡xt = \sin x, t=cos⁡xt = \cos x and t=tan⁡xt = \tan x. Then the rational functions of xx and of a root ax+bcx+dn\sqrt[n]{\frac{ax + b}{cx + d}} are rationalized, and those of xx and of ax2+bx+c\sqrt{ax^2 + bx + c}, with the three Euler substitutions. The lesson closes with the differential binomials xm(a+bxn)px^m \left(a + b x^n\right)^p and Chebyshev’s three cases, in which the integral reduces to one of the preceding ones.

Learning objectives

  1. Rationalize a rational function of sine and cosine with t=tan⁡x2t = \tan \frac{x}{2}, state the intervals on which the antiderivative obtained holds, and join those antiderivatives into an antiderivative on the whole line when the integrand is continuous.
  2. Recognize from the parity of the integrand when the substitutions t=sin⁡xt = \sin x, t=cos⁡xt = \cos x or t=tan⁡xt = \tan x suffice, and prove that they rationalize it.
  3. Rationalize the rational functions of xx and of ax+bcx+dn\sqrt[n]{\frac{ax + b}{cx + d}}, and those of xx and of ax2+bx+c\sqrt{ax^2 + bx + c} with the Euler substitution that corresponds to the coefficients.
  4. Decide whether a differential binomial falls under one of Chebyshev’s three cases and, if it does, find its antiderivative.

Rational functions of two variables

The lesson on the integration of rational functions settled one type of integrand completely: by the corollary on the antiderivatives of rational functions, every quotient of polynomials has an antiderivative on each interval that contains no roots of its denominator, and that antiderivative is written with polynomials, rational functions, logarithms and arctangents. This lesson looks for substitutions that turn other integrands into rational functions of the new variable. To describe them, one more notion is needed.

DefinitionRational function of two variables

A polynomial in two variables is a finite sum of terms of the form a uivja \, u^i v^j, with a real coefficient aa and integer exponents i≥0i \geq 0 and j≥0j \geq 0; its degree is the largest value of i+ji + j among the terms with nonzero coefficient. A rational function of two variables is a quotient R(u,v)=P(u,v)Q(u,v)R(u, v) = \frac{P(u, v)}{Q(u, v)} of two polynomials in two variables, with QQ nonzero, defined at the points (u,v)(u, v) at which Q(u,v)≠0Q(u, v) \neq 0.

For example, 12+cos⁡x\frac{1}{2 + \cos x} and sin⁡3x2+cos⁡x\frac{\sin^3 x}{2 + \cos x} are of the form R(sin⁡x,cos⁡x)R(\sin x, \cos x), with R(u,v)=12+vR(u, v) = \frac{1}{2 + v} and R(u,v)=u32+vR(u, v) = \frac{u^3}{2 + v}. The whole lesson rests on the following lemma: when the two variables are replaced by rational functions of one variable tt, the result is a rational function of tt whose denominator does not vanish wherever everything is defined.

LemmaComposite rational functions

Let R=PQR = \frac{P}{Q} be a rational function of two variables, rr and ss two rational functions of one variable, and TT an interval on which rr and ss are defined and Q(r(t),s(t))≠0Q\left(r(t), s(t)\right) \neq 0. Then there exist two polynomials AA and BB in one variable, with B(t)≠0B(t) \neq 0 for every tt in TT, such that R(r(t),s(t))=A(t)B(t)R\left(r(t), s(t)\right) = \frac{A(t)}{B(t)} on TT. Consequently, that function has an antiderivative on TT, of the form given by the corollary on the antiderivatives of rational functions. The numerator and the denominator are multiplied by a power of the denominators of rr and of ss high enough to remove them from all the terms.

Proof

  1. r2Ns2N⋅a risj=a r1i r2N−i s1j s2N−jr_2^N s_2^N \cdot a \, r^i s^j = \resaltar{a \, r_1^i \, r_2^{N - i} \, s_1^j \, s_2^{N - j}}

    Let r=r1r2r = \frac{r_1}{r_2} and s=s1s2s = \frac{s_1}{s_2}, with polynomials r2r_2 and s2s_2 without roots in TT, and let NN be an integer greater than or equal to the degrees of PP and of QQ. Every term a uivja \, u^i v^j of PP or of QQ has i≤Ni \leq N and j≤Nj \leq N; replacing uu by r(t)r(t) and vv by s(t)s(t) and multiplying by r2(t)Ns2(t)Nr_2(t)^N s_2(t)^N leaves a polynomial in tt.

  2. R(r(t),s(t))=r2Ns2N P(r(t),s(t))r2Ns2N Q(r(t),s(t))=A(t)B(t),t∈TR\left(r(t), s(t)\right) = \frac{r_2^N s_2^N \, P\left(r(t), s(t)\right)}{r_2^N s_2^N \, Q\left(r(t), s(t)\right)} = \resaltar{\frac{A(t)}{B(t)}}, \qquad t \in T

    Let AA and BB be the polynomials that result from PP and from QQ. On TT, the factor r2(t)Ns2(t)Nr_2(t)^N s_2(t)^N does not vanish, and neither does Q(r(t),s(t))Q\left(r(t), s(t)\right); therefore, B(t)≠0B(t) \neq 0. The quotient does not change when the numerator and the denominator are multiplied by the same factor, and the corollary on the antiderivatives of rational functions applies on TT, which contains no roots of BB.

If, in addition, R(r(t),s(t))R\left(r(t), s(t)\right) is multiplied by another rational function without poles in TT, such as the derivative of a substitution, the product is still a quotient of polynomials whose denominator does not vanish on TT. Each technique of this lesson therefore consists in choosing a substitution x=h(t)x = h(t) in which hh, and the root or the trigonometric functions of the integrand, are rational functions of tt.