Why trig integrals need their own toolkit
The previous guide used trigonometric substitution to remove roots like sqrt(a^2 - x^2). This one is its mirror image: here you already have sines and cosines and you want to integrate them. An integral such as 'integral of sin^4(x) cos^2(x) dx' looks harmless, but a blind u-substitution stalls — the derivative of sin is cos, so the cosines and sines tangle instead of cancelling. The fix is not cleverness but bookkeeping: a small set of powers-of-trig patterns covers essentially every case.
Two identities do almost all the work. The Pythagorean identity sin^2 + cos^2 = 1 (and its cousins 1 + tan^2 = sec^2, 1 + cot^2 = csc^2) lets you convert between functions, and the double-angle / power-reduction identities cos^2(x) = (1 + cos 2x)/2 and sin^2(x) = (1 - cos 2x)/2 let you trade a square for a lower-power expression in 2x. Knowing which identity to reach for, and when, is the whole skill — and it depends on whether your exponents are odd or even.
Powers of sine and cosine: odd is easy, even needs halving
For 'integral of sin^m(x) cos^n(x) dx', look at the parities of m and n. If either exponent is odd, you win immediately: peel off one factor of that odd function to be the dx-partner, and use sin^2 + cos^2 = 1 to rewrite the remaining even power in terms of the other function. For 'integral of sin^5(x) cos^2(x) dx', split sin^5 = sin^4 * sin = (1 - cos^2 x)^2 * sin x; now let u = cos x, du = -sin x dx, and you are integrating a plain polynomial in u — the same substitution you learned in Volume I, finally with a clear target.
If both exponents are even, no single factor can become du, so you climb down instead. Apply the power-reduction identities to halve every even power, multiply everything out, and you get a sum of cosines of 2x, 4x, and so on — each of which integrates in one line. 'integral of sin^2 x dx' becomes 'integral of (1 - cos 2x)/2 dx = x/2 - sin(2x)/4 + C'. Higher even powers just mean applying the halving twice; it is repetitive, never hard.
Secants and tangents: a different pairing
The tangent-secant family plays by similar rules, anchored on two facts: the derivative of tan x is sec^2 x, and the derivative of sec x is sec x tan x. So for 'integral of tan^m(x) sec^n(x) dx', an even power of sec (n even, n >= 2) lets you peel off sec^2 x for the du-partner and convert the rest with sec^2 = 1 + tan^2, then set u = tan x. An odd power of tan together with at least one sec lets you peel off sec x tan x and convert the remaining even tan power, then set u = sec x.
The awkward leftover cases — an odd power of tangent with no secant, or an odd power of secant alone — do not yield to a clean substitution. 'integral of sec x dx = ln|sec x + tan x| + C' is the famous trick (multiply and divide by sec x + tan x so the numerator becomes the derivative of the denominator), and 'integral of sec^3 x dx' is the textbook case for a reduction formula or integration by parts — exactly the recursion machinery the next guide builds. Recognizing that a problem belongs to this harder pocket is itself progress.
The Weierstrass substitution: one trick to rule them all
What about something like 'integral of dx/(2 + cos x)', where there is no power to peel and no factor to save? Here the Weierstrass substitution — the tangent half-angle substitution t = tan(x/2) — is a genuine universal solvent. The magic is that it converts every rational expression in sin x and cos x into a rational function of the single algebraic variable t, with no trig left at all. You then finish with the workhorses of the previous guide: partial fractions applied to an integral of a rational function.
Let t = tan(x/2). Then the three pieces become:
sin x = 2t / (1 + t^2)
cos x = (1 - t^2) / (1 + t^2)
dx = 2 / (1 + t^2) dt
Example: integral of dx / (2 + cos x)
2 + cos x = 2 + (1 - t^2)/(1 + t^2) = (3 + t^2)/(1 + t^2)
integral becomes integral of [2/(1+t^2)] * [(1+t^2)/(3+t^2)] dt
= integral of 2/(3 + t^2) dt
= (2/sqrt(3)) arctan(t/sqrt(3)) + C
back-substitute t = tan(x/2):
= (2/sqrt(3)) arctan( tan(x/2)/sqrt(3) ) + CWhere do those formulas come from? They are pure double-angle bookkeeping. Writing x = 2 * (x/2) and using the double-angle identities, sin x = 2 sin(x/2) cos(x/2) and cos x = cos^2(x/2) - sin^2(x/2); dividing top and bottom by cos^2(x/2) turns every term into tan(x/2) = t, which is where the 1 + t^2 denominators are born. And dx follows because t = tan(x/2) gives dt/dx = (1/2) sec^2(x/2) = (1/2)(1 + t^2), so dx = 2/(1 + t^2) dt.
The procedure, and its honest limits
- Set t = tan(x/2) and write down dt = (1/2)(1 + t^2) dx, i.e. dx = 2/(1 + t^2) dt.
- Replace every sin x by 2t/(1 + t^2) and every cos x by (1 - t^2)/(1 + t^2); the integrand becomes a rational function of t.
- Simplify and integrate that rational function — usually by partial fractions, sometimes by a quick arctan or log.
- Substitute t = tan(x/2) back in; for a definite integral, instead convert the limits (x = 0 gives t = 0, and watch out near x = pi).
Step back and the whole landscape is one connected idea: every method here ends by reducing a trig integral to something elementary you can already do — a polynomial in u, a sum of cosines, or a rational function. None of these antiderivatives is mysterious or non-elementary; they always exist in closed form. What you are building is judgment — recognizing at a glance which lever (parity, identity, or half-angle) opens a given integral with the least work.