Integrals That Give tan⁻¹ and sin⁻¹

Fractions whose antiderivatives are angles.

An integral that is an angle

1/(1 + x²) is not a power of x, and its numerator is not the derivative of its denominator, which would be 2x. Yet it is a derivative: the derivative of tan⁻¹x.

Here is why. If y = tan⁻¹x, then x = tan y, and dx/dy = sec²y = 1 + tan²y = 1 + x². So dy/dx = 1/(1 + x²). Read backwards, ∫ 1/(1 + x²) dx = tan⁻¹x + C.

An area of π/4

The area under y = 1/(1 + x²) from x = 0 to x = 1 is tan⁻¹1 − tan⁻¹0 = π/4 − 0 = π/4 ≈ 0.785. An area under a curve comes out as an angle: the angle whose tangent is 1.

From 0 to 10 the area is tan⁻¹10 ≈ 1.471. As the right-hand edge moves further out, the area keeps growing, but tan⁻¹x stays below π/2 ≈ 1.571, so the area never passes π/2.

xy

The curve y = 1/(1 + x²), from height 1 at x = 0 to height ½ at x = 1, shaded between them. The shaded area is tan⁻¹1 = π/4 ≈ 0.785.

sin⁻¹ in the same way

If y = sin⁻¹x, then x = sin y, and dx/dy = cos y. With y between −π/2 and π/2, cos y is not negative, so cos y = √(1 − sin²y) = √(1 − x²). So dy/dx = 1/√(1 − x²), and ∫ 1/√(1 − x²) dx = sin⁻¹x + C, for x between −1 and 1.

The area under y = 1/√(1 − x²) from 0 to ½ is sin⁻¹½ − sin⁻¹0 = π/6 ≈ 0.524.

xy

The curve y = 1/√(1 − x²), which has height 1 at x = 0 and climbs without limit toward x = −1 and x = 1. The region from 0 to ½ is shaded, and its area is sin⁻¹½ = π/6 ≈ 0.524.

With a² in place of 1

For 1/(a² + x²), with a > 0, substitute x = au, so dx = a du. The denominator becomes a² + a²u² = a²(1 + u²), and the integral becomes ∫ a/(a²(1 + u²)) du = (1/a) ∫ 1/(1 + u²) du = (1/a) tan⁻¹u + C. Putting u = x/a back: ∫ 1/(a² + x²) dx = (1/a) tan⁻¹(x/a) + C.

For 1/(9 + x²), a = 3, since 9 = 3 × 3. So the integral is (1/3) tan⁻¹(x/3) + C. Check: differentiating gives (1/3) × (1/3) × 1/(1 + x²/9) = (1/9) × 9/(9 + x²) = 1/(9 + x²). From 0 to 3 the integral is (1/3)(π/4) = π/12 ≈ 0.262.

For the root, the a cancels: √(a² − a²u²) = a√(1 − u²), and dx = a du divided by that leaves du/√(1 − u²). So ∫ 1/√(a² − x²) dx = sin⁻¹(x/a) + C, with no 1/a in front. From 0 to 1, ∫ 1/√(4 − x²) dx = sin⁻¹½ = π/6.

A number in front of x² is handled the same way. 1 + 4x² = 1 + (2x)², so ∫ 1/(1 + 4x²) dx = ½ tan⁻¹(2x) + C, the ½ undoing the 2 the chain rule brings out. From 0 to ½ that is ½ tan⁻¹1 = π/8 ≈ 0.393.

The sign decides

x² − 1 factors as (x − 1)(x + 1), so 1/(x² − 1) splits into partial fractions and integrates to logs: ½ ln|x − 1| − ½ ln|x + 1| + C. x² + 1 is never 0, so it has no factors, and 1/(x² + 1) integrates to tan⁻¹x + C instead.

The numerator matters too. x/(1 + x²) is half of 2x/(1 + x²), whose numerator is the derivative of its denominator, so it integrates to a log: ½ ln(1 + x²) + C. Only a constant over 1 + x² gives tan⁻¹.

Worked example: A Weld Inspection Source Beside a Walkway: A Dose That Comes Out as an Inverse Tangent

Question A small gamma source used for weld inspection stands 3 meters from a straight walkway. At a distance d meters from the source the dose rate is 30d2 microsieverts per second, and a technician walks the walkway at one meter per second, so each meter walked is a second spent. Writing x for the meters past the point of closest approach, d2 = 9 + x2. (a) What dose does the technician pick up between x = 0 and x = 3? (b) The walkway is moved so that it passes 4 meters from the source instead. What dose is picked up between x = 0 and x = 4 then? Take π = 3.14159 and give each answer to three significant figures.

  1. 1.The dose is ∫03309 + x2dx. Substitute x = 3tanθ, since 9 + 9tan2θ = 9sec2θ turns the sum of two squares into a single square.

    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)
    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)
    The dose is ∫03309 + x2dx. Substitute x = 3tanθ, since 9 + 9tan2θ = 9sec2θ.
  2. 2.The differential is dx = 3sec2θ dθ, so the integrand becomes 309sec2θ × 3sec2θ dθ = 10 dθ. Both squared secants cancel and nothing is left but a constant.

    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancel
    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancel
    The differential gives dx = 3sec2θ dθ, so the integrand becomes 309sec2θ × 3sec2θ dθ = 10 dθ.
  3. 3.The limits travel with the substitution. At x = 0, tanθ = 0, so θ = 0; at x = 3, tanθ = 1, so θ = π4.

    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4
    0123402346x, meters past the closest pointdose rate, per secondx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4
    The limits travel with it: x = 0 gives θ = 0, and x = 3 gives tanθ = 1, so θ = π4.
  4. 4.(a) The dose is 10[θ]0π4 = 10π4 = 5π2 = 7.85 microsieverts. That working, done once and for all, is the standard result ∫dxa2 + x2 = 1atan−1xa + C.

    0123402346x, meters past the closest pointdose rate, per second7.85x = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4(a) 10 × pi/4 = 7.85
    0123402346x, meters past the closest pointdose rate, per second7.85x = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4(a) 10 × pi/4 = 7.85
    (a) The dose is 10 × π4 = 5π2 = 7.85 microsieverts, which is the standard 1atan−1xa written out.
  5. 5.(b) With the offset 4, ∫043016 + x2dx = 304[tan−1x4]04 = 304 × π4 = 15π8 = 5.89 microsieverts. Check: each answer is 30a × π4, so moving the walkway from 3 meters out to 4 multiplies the dose by 34, and three quarters of 7.85 is 5.89.

    0123402346x, meters past the closest pointdose rate, per second7.855.89offset 3 moffset 4 mx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4(a) 10 × pi/4 = 7.85(b) offset 4: 7.5 × pi/4 = 5.89
    0123402346x, meters past the closest pointdose rate, per second7.855.89offset 3 moffset 4 mx = 3 tan θ, so 9 + x × x = 9/(cos θ × cos θ)dx = 3 dθ/(cos θ × cos θ), and the two cancellimits: x = 0 gives θ = 0; x = 3 gives θ = pi/4(a) 10 × pi/4 = 7.85(b) offset 4: 7.5 × pi/4 = 5.89
    (b) At an offset of 4 meters the dose is 304 × π4 = 15π8 = 5.89 microsieverts, three quarters of the first.

Answer: (a) 5π2 = 7.85 microsieverts; (b) 15π8 = 5.89 microsieverts, three quarters of the first

Common mistakes

  • Expecting a logarithm, and writing the answer as a multiple of ln(9 + x2). A logarithm appears when the top of the fraction is the derivative of the bottom; the derivative of 9 + x2 is 2x, and the top here is a constant. A constant over a sum of two squares gives an inverse tangent.
  • Leaving the limits as 0 and 3 after substituting x = 3tanθ. Those are meters, and after the substitution the variable is an angle, so the limits become 0 and π4 radians. Using 0 and 3 would give 30 microsieverts, nearly four times the true dose.

More techniques of integration problems, worked step by step →

The usual mistakes

Expecting a log. ln(1 + x²) differentiates to 2x/(1 + x²), which has an x on top that 1/(1 + x²) does not.

Mixing up the pair. tan⁻¹ goes with 1 + x² and no root; sin⁻¹ goes with the root of 1 − x².

Leaving out the 1/a. tan⁻¹(x/3) differentiates to 3/(9 + x²), three times too much.

Using a² for a. In 1/(4 + x²), a = 2, so the answer is ½ tan⁻¹(x/2) + C, not ¼ tan⁻¹(x/4) + C.

A factor 1/a in front of sin⁻¹(x/a). The a in the root cancels the a in dx, so nothing is left over.

Practice Integrals That Give tan⁻¹ and sin⁻¹ in the app