Rational algebraic integrands built from quadratic expressions ax2+bx+c (or their square roots) do not yield to the substitution and by-parts methods developed earlier in the chapter directly -- they need their own toolkit, organised here into four types that build on one another, one feeding into the next.
Type I. These are the six "base case" integrals built from a2±x2 and x2±a2, and every later type is eventually reduced to one of these six:
The first two are proved by factoring the denominator as a difference of squares and splitting into partial fractions: a2−x21=(a−x)(a+x)1=2a1[a+x1+a−x1], so integrating term by term gives 2a1[log∣a+x∣−log∣a−x∣]+c=2a1loga−xa+x+c; the x2−a2 case runs identically, with the sign of one partial fraction flipped.
The third and fourth are proved instead by a trigonometric substitution. For ∫a2+x2dx, put x=atanθ, so dx=asec2θdθ and a2+x2=a2sec2θ; the integral collapses to a1∫dθ=a1θ+c=a1tan−1(ax)+c. For ∫a2−x2dx, put x=asinθ, so dx=acosθdθ and a2−x2=a2cos2θ; the acosθ in the numerator cancels the square root exactly, leaving ∫dθ=θ+c=sin−1(ax)+c. The remaining two (via x=asecθ) work the same way, both ending in ∫secθdθ=log∣secθ+tanθ∣+c, which is converted back to x using a right triangle with hypotenuse x and one leg a; absorbing the leftover −loga into the constant of integration gives the tidy form logx+x2∓a2+c.
A companion memory aid runs through the whole chapter: whenever the surd is a2−x2, substitute x=asinθ; for a2+x2, substitute x=atanθ; for x2−a2, substitute x=asecθ.
Type II. Integrals ∫ax2+bx+cdx and ∫ax2+bx+cdx are handled by first forcing the coefficient of x2 to 1 and then completing the square:
ax2+bx+c=a[(x+2ab)2+4a24ac−b2].
Whatever constant is left over after completing the square plays the role of "a2" (or "−a2") in a Type I form, in the shifted variable x+2ab -- so every Type II integral reduces directly to one of the six Type I results. The only genuinely new skill is the algebra of completing the square (and, for the square-root case with a leading coefficient =1, factoring that coefficient out from under the radical first).
Type III. Integrals with a linear numerator, ∫ax2+bx+cpx+qdx and ∫ax2+bx+cpx+qdx, are solved by writing the numerator as a multiple of the denominator's derivative, plus a leftover constant:
px+q=Adxd(ax2+bx+c)+B=A(2ax+b)+B,
and finding A and B by comparing coefficients of x and of the constant term on both sides. Substituting back splits the integral into two pieces:
The first piece is an ∫f(x)f′(x)dx form and integrates directly to Alog∣ax2+bx+c∣; the second piece is exactly a Type II integral, evaluated by completing the square. The square-root version runs in parallel: its first piece is an ∫f′(x)[f(x)]ndx form with n=−21, giving 2Aax2+bx+c, while its second piece is a Type II square-root integral.
Remark (substitution table). As an alternative route to the same integrals, once the quadratic surd is written as a2−x2, a2+x2, or x2−a2 (after completing the square), it can also be attacked directly with the matching trigonometric substitution: x=asinθ for a2−x2, x=atanθ for a2+x2, and x=asecθ for x2−a2 -- the very substitutions used to prove the Type I square-root results.
Type IV. Integrals of the surd itself (not one over the surd), ∫a2−x2dx, ∫x2−a2dx, and ∫x2+a2dx, form Result 11.3: