Two integrals recur throughout calculus and mechanics: ∫x2+a2dx
and ∫x2−a2dx (the syllabus's ∫dx/(x2±a2)). Both are
derived once here and then reused, via completing the square, for every quadratic denominator.
Derivation of ∫dx/(x2+a2) by trigonometric substitution. Let x=atanθ, so
dx=asec2θdθ and x2+a2=a2tan2θ+a2=a2sec2θ (using
1+tan2θ=sec2θ). Then
∫x2+a2dx=∫a2sec2θasec2θdθ=a1∫dθ=aθ+C=a1tan−1(ax)+C,
using θ=tan−1(x/a) from the substitution. This is verified directly by differentiating
the right side: dxd[a1tan−1(x/a)]=a1⋅1+x2/a21/a=x2+a21.
Derivation of ∫dx/(x2−a2) by partial fractions. Since x2−a2=(x−a)(x+a),
x2−a21=2a1[x−a1−x+a1] (Section 3), so
∫x2−a2dx=2a1lnx+ax−a+C.
Reducing a general quadratic denominator: complete the square. For
∫dx/(ax2+bx+c), first factor out a and complete the square inside:
ax2+bx+c=a[(x+2ab)2+4a24ac−b2].
Writing u=x+b/(2a) and k2=∣4ac−b2∣/(4a2), the bracket becomes u2+k2 (if
4ac>b2) or u2−k2 (if 4ac<b2) -- exactly the two forms derived above, now in u, giving
a tan−1 or a ln answer respectively. Example 8 carries this out for
∫dx/(x2+4x+13), where (x+2)2+9 gives the tan−1 case.
The linear-numerator form ∫(px+q)dx/(ax2+bx+c). The denominator's derivative is …