Skip to content
Topos Uranos

Integration of rational functions

Summary

This lesson finds the antiderivatives of rational functions, quotients P(x)Q(x)\frac{P(x)}{Q(x)} of two polynomials. Division reduces the problem to a proper fraction; the theorem on factorization over the reals writes the denominator as a product of linear factors and of quadratic factors without real roots; and partial fraction decomposition turns the fraction into a sum of partial fractions, each with a known antiderivative. Algebra and Geometry admitted that decomposition when there are quadratic factors; here its existence and its uniqueness are proved, with Bézout’s identity for polynomials. It follows that every rational function has, on each interval free of roots of the denominator, an antiderivative made up of a rational function, logarithms and arctangents. The lesson closes with the Hermite-Ostrogradsky method, which obtains the rational part of the antiderivative without integrating and without knowing the roots of the denominator.

Learning objectives

  1. Prove Bézout’s identity for coprime polynomials and, with it, the existence and the uniqueness of the partial fraction decomposition with irreducible quadratic factors.
  2. Integrate each partial fraction, separating in the numerator a multiple of the derivative of the denominator, and give each antiderivative with the intervals on which it holds.
  3. Apply the Hermite-Ostrogradsky method, with the greatest common divisor of QQ and Q′Q', to obtain the rational part of an antiderivative without finding the roots of QQ.
  4. Recognize when a division, a simplification or a substitution shortens the computation before decomposing.

Rational functions and proper fractions

A rational function is a quotient P(x)Q(x)\frac{P(x)}{Q(x)} of two polynomials with real coefficients, with QQ nonzero, as in the lesson on algebraic fractions and partial fractions. Its domain excludes the real roots of QQ, which are finitely many. If they are r1<r2<⋯<rmr_1 < r_2 < \cdots < r_m, the domain is the union of the m+1m + 1 open intervals that they separate, and by the proposition on antiderivatives on a union of disjoint open intervals, from the lesson on antiderivatives, each antiderivative carries its own constant on each of them. If QQ has no real roots, the domain is R\mathbb{R} and one constant suffices.

If the fraction is improper, that is, if the degree of PP is not less than that of QQ, the division algorithm, from the lesson on the algebra of polynomials, gives P=QE+RP = QE + R, with RR zero or of degree less than that of QQ, and then PQ=E+RQ\frac{P}{Q} = E + \frac{R}{Q} on the domain. The polynomial EE has an antiderivative on R\mathbb{R}, by linearity and the antiderivatives of powers; what remains is the proper fraction RQ\frac{R}{Q}.

The denominator is factored with the theorem on factorization over the reals, from the lesson on the roots of higher-degree polynomials: every real polynomial of degree n≥1n \geq 1 is the product of its leading coefficient by factors x−rx - r, with rr real, and by monic quadratic factors x2+px+qx^2 + px + q irreducible over the reals, that is, with p2−4q<0p^2 - 4q < 0. Gathering equal factors into powers:

Q=c F1k1F2k2⋯FsksQ = c \, F_1^{k_1} F_2^{k_2} \cdots F_s^{k_s}

where cc is the leading coefficient, the factors F1,…,FsF_1, \ldots, F_s are pairwise distinct, each linear or irreducible quadratic, and the exponents kik_i are nonzero natural numbers. That lesson proved the theorem from the fundamental theorem of algebra (every polynomial of degree n≥1n \geq 1 with complex coefficients has at least one complex root), which it admitted without proof. Here it is admitted as well, as in Algebra and Geometry; it will be proved in the lesson on subsequences and the Bolzano-Weierstrass theorem. We will also use the roots of each factor: x−rx - r has only the root rr, which is real; an irreducible quadratic factor has two complex conjugate roots zz and z‾\conj{z}, which are not real, and it equals (x−z)(x−z‾)(x - z)(x - \conj{z}), as was seen in the proof of that theorem.