5.7 Integrals Resulting in Inverse Trigonometric Functions

The three inverse trigonometric integration formulas and their general forms with a constant, reaching them by substitution, completing the square for a quadratic denominator, the domains on which each is valid, and telling them apart from the logarithmic forms they resemble.

Subject: Calculus I · 69 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. Integrals Resulting in Inverse Trigonometric Functions

Title

Calculus I · Section 5.7

Algebraic integrands whose answers are angles

2. What this lesson gives you

Objectives

Section 3.7 found the derivatives of the inverse trigonometric functions. This lesson runs those formulas backwards, and the result is a new family of integrals you could not do before: ones whose answers are angles.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, pp. 526-532 — learning objective 5.7.1

So far every antiderivative you have found has been a power, an exponential, a logarithm or a trigonometric function. This last section of Chapter 5 adds one more family, and it comes with a surprise: integrands with no trigonometry in them at all, like one over one plus x squared, turn out to have angles as their antiderivatives.

There are only three formulas to learn, and you already know them, because they are the derivatives from Section 3.7 read backwards. The real skill is recognition: seeing that an integrand is one of the three forms in disguise, and choosing the substitution that reveals it.

The lesson also deals with the two things that most often go wrong. One is confusing these forms with logarithms, which share the same denominators. The other is forgetting that two of the three formulas only work on a restricted interval.

3. Before anything new: three derivatives from Section 3.7

Warm-up

Discussion prompt

Write down the derivatives of inverse sine, inverse tangent and inverse secant of x without looking back. For which x is each one defined?

Write your three answers before you reveal them. If you can produce these from memory, the rest of this lesson is pattern matching. If you cannot, this is the moment to notice, because every example that follows depends on them.

Pay attention to the last column as well as the formulas. The inverse sine is only defined between minus 1 and 1, the inverse secant only outside that interval, and the inverse tangent everywhere. Those domains come back in Part 4, where they decide whether a definite integral even makes sense.

If you got the inverse secant wrong, do not worry too much. It is the least used of the three, and the absolute value in it is a detail you will see explained properly later in this lesson.

4. Reading the derivative table backwards

Section

Part 1

5. An algebraic integrand with an angle for an answer

Intuition

Nothing in one over one plus x squared mentions a triangle or a circle. It is a plain rational function. Yet its antiderivative is an angle.

\[ \int\frac{dx}{1+x^2} = \tan^{-1}x + C \]

That is not a trick. It is just the derivative of the arctangent, read the other way. Every differentiation formula is an integration formula in disguise, and this section collects the three that produce inverse trigonometric functions.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 526 — introduction

Pause on how strange this is. One over one plus x squared is built from nothing but multiplication, addition and division. Its antiderivative is the angle whose tangent is x. Where did the angle come from?

The answer is that it was hiding in the derivative table all along. When you differentiated the inverse tangent in Section 3.7, you started with an angle and the trigonometry cancelled out, leaving a purely algebraic derivative. Integration runs that process backwards, so the angle reappears.

Keep this in mind as a general principle: any time you meet an integrand you do not recognise, ask whether it is the derivative of something you know. That question is what this whole section is built on.

6. Three derivatives you already know

Concept

Section 3.7 produced these by differentiating an inverse function. Each right-hand side is an algebraic function with no trigonometry left in it.

functionderivativedefined for
sin⁻¹ x1/√(1 − x²)−1 < x < 1
tan⁻¹ x1/(1 + x²)every x
sec⁻¹ x1/(|x|√(x² − 1))|x| > 1

The other three inverse functions, cosine, cotangent and cosecant, have the same derivatives with a minus sign in front, so they add nothing new for integration.

Stewart, Calculus: Early Transcendentals 8e, §3.5 Implicit Differentiation §3.5, pp. 208-217 — the inverse trigonometric derivatives in Stewart

These three lines are the raw material for the whole lesson. Look at the right-hand column of the table: each derivative is an algebraic expression, a fraction with a square root or a sum of squares, and no trigonometric function survives.

You might wonder why there are three formulas and not six, since there are six inverse trigonometric functions. The reason is that inverse cosine, inverse cotangent and inverse cosecant have derivatives that are exactly the negatives of these three. Anything they could integrate, these three can integrate with a minus sign in front.

Notice also that each derivative has a domain. That is not decoration. An antiderivative is only an antiderivative where both it and the integrand exist, so the domain in the last column will limit where each integration formula is allowed.

7. Slope on one graph, height on the other

Picture it

Figure (svg): Two panels. Left: the curve y equals inverse tangent of x from minus 3 to 3, with its tangent line at x equals 1, slope one half. Right: the curve y equals one over one plus x squared, with the point at x equals 1, height one half, marked.

At x = 1 the arctangent climbs with slope one half, and the integrand stands exactly one half tall there. That match holds at every x: that is what antiderivative means.

On the left, the arctangent curve; on the right, one over one plus x squared. Wherever you measure the slope on the left, the right-hand curve has exactly that height. That is the whole content of an antiderivative.

Look at the left panel first. The arctangent curve rises steeply through the origin and then flattens toward the dashed lines at plus and minus pi over two. The yellow tangent line at x equal to 1 has slope one half.

Now look at the right panel at the same x. The curve one over one plus x squared stands exactly one half tall there. Try it at other points in your head: at x equal to 0 the arctangent is steepest, slope 1, and the right-hand curve is at its peak, height 1. Far out, the arctangent is nearly flat, and the right-hand curve is nearly zero.

This is the meaning of the integration formula. It says the left curve is an antiderivative of the right curve, which is the same as saying the right curve records the slope of the left one at every point.

8. The three integration formulas

Concept

Put a positive constant a into the three derivatives and read them backwards. These are formulas 5.23 to 5.25.

\[ \int\frac{du}{\sqrt{a^2-u^2}} = \sin^{-1}\left(\frac{u}{a}\right) + C \]

\[ \int\frac{du}{a^2+u^2} = \frac{1}{a}\tan^{-1}\left(\frac{u}{a}\right) + C \]

\[ \int\frac{du}{u\sqrt{u^2-a^2}} = \frac{1}{a}\sec^{-1}\left(\frac{|u|}{a}\right) + C \]

the standing assumption — The constant a is positive. In every problem you find a by writing the constant term as a square.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, pp. 526-527 — Rule: formulas 5.23, 5.24 and 5.25

Here is the rule from the book, with a general positive constant a in place of 1. The u stands for whatever expression you substitute; in the simplest problems it is just x.

Compare the three carefully, because the differences are exactly what trips people up. The sine form has a square root and a constant squared minus a variable squared. The tangent form has no root and a sum. The secant form has a root with the variable squared minus the constant squared, and an extra u outside the root.

The factors in front differ too. The tangent and secant formulas have one over a; the sine formula does not. The next few slides explain where these factors come from, so that you never have to rely on memory alone to get them right.

9. Reading the inverse sine formula piece by piece

Notation

Annotate

On: \( \int \frac{du}{\sqrt{a^2 - u^2}} = \sin^{-1}\left(\frac{u}{a}\right) + C \)

  • A constant squared MINUS a variable squared, under a square root. The order matters: u² − a² is the secant form.
  • The differential must match u exactly. If u = 3x, you need 3 dx on top, and a factor of 1/3 appears to supply it.
  • The inverse sine is taken of u divided by a, never of u alone (unless a = 1).
  • Unlike the other two formulas there is no 1/a in front. The a in the root and the a in the substitution cancel.

Step through the annotations and connect each to what you will actually do in a problem. First you look for a constant minus a square under a root. The constant tells you a: if it is 9, then a is 3.

Then you look at the thing being squared. That is u, and the differential in the integral has to be exactly du. If u is 3x, the integral must contain 3 dx; if it only contains dx, you supply the 3 and pay for it with a factor of one third outside. That factor comes from the substitution, not from the formula.

The last note is the one that causes most lost marks. The inverse sine formula has no one over a in front. It is tempting to copy the shape of the tangent formula, and the result is wrong by a factor of a.

10. Reading the inverse tangent formula piece by piece

Notation

Annotate

On: \( \int \frac{du}{a^2 + u^2} = \frac{1}{a}\tan^{-1}\left(\frac{u}{a}\right) + C \)

  • A sum of two squares, with no root. The denominator is never zero, so this formula works for every u.
  • A factor of one over a in front. It survives because a² + u² = a²(1 + (u/a)²) leaves an a² that du = a dw only half cancels.
  • The same inner fraction as the sine formula: u measured in units of a.
  • An indefinite integral always carries the constant; in a definite integral it cancels.

The tangent form is the friendliest of the three. There is no square root, and the denominator is a sum of two squares, which can never be zero. That means the formula works on every interval without any checking.

The one over a in front is the part to understand rather than memorise. When you factor a squared out of a squared plus u squared, you get a squared times one plus u over a squared. Substituting w equal to u over a contributes one factor of a through du, which cancels only one of the two a's in a squared. The other one remains as one over a.

The next slide carries out that calculation in full, so you can see it happen rather than take it on trust.

11. Deriving the general inverse tangent form

Worked example

Only the case a equal to 1 comes straight from the derivative table. Derive formula 5.24 from it.

\[ \int\frac{du}{a^2+u^2} \]

Factor a squared out of the denominator

Why: This makes the denominator look like one plus something squared.

\[ a^2+u^2 = a^2\left(1+\frac{u^2}{a^2}\right) \]

Substitute w for u over a

Why: Then u is a times w, so du is a times dw.

\[ w = \frac{u}{a}, \quad du = a\,dw \]

Rewrite the integral in w

Why: One a cancels; one stays in the denominator.

\[ \int\frac{du}{a^2+u^2} = \int\frac{a\,dw}{a^2(1+w^2)} = \frac{1}{a}\int\frac{dw}{1+w^2} \]

Use the basic arctangent

Why: The table gives the integral of one over one plus w squared.

\[ \frac{1}{a}\int\frac{dw}{1+w^2} = \frac{1}{a}\tan^{-1}w + C \]

Return to u

Why: Replace w by u over a.

\[ \int\frac{du}{a^2+u^2} = \frac{1}{a}\tan^{-1}\frac{u}{a} + C \]

Check by differentiating

Why: The chain rule brings out another one over a, which rebuilds the a squared.

\[ \frac{d}{du}\left[\frac1a\tan^{-1}\frac ua\right] = \frac1a\cdot\frac{1/a}{1+u^2/a^2} = \frac{1}{a^2+u^2} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 527 — formula 5.24

The derivative table gives you only the case a equal to 1: the integral of one over one plus w squared is the inverse tangent of w. Every other value of a is reached by a substitution, and this slide does it once for all a.

The key move is the first one: factoring a squared out of the denominator so that it looks like one plus something squared. That something is u over a, so that is the natural substitution. Then du is a times dw, and one a cancels against the a squared, leaving one over a outside.

The check by differentiating is worth doing yourself. The chain rule on the inverse tangent of u over a produces one over a from the inner derivative, and together with the one over a in front that gives one over a squared, which is exactly what rebuilds the a squared in the denominator.

12. Proof: the inverse sine formula by implicit differentiation

Worked example

The book proves formula 5.23. Follow its argument one move at a time.

Name the angle

Why: Let y be the inverse sine of x over a, and undo the inverse.

\[ y = \sin^{-1}\frac{x}{a} \quad\Longrightarrow\quad a\sin y = x \]

Differentiate both sides with respect to x

Why: The left side needs the chain rule, since y depends on x.

\[ a\cos y\,\frac{dy}{dx} = 1 \]

Solve for the derivative

Why: Divide by a cos y.

\[ \frac{dy}{dx} = \frac{1}{a\cos y} \]

Figure (svg): A right triangle with the right angle at the lower right. The angle y sits at the lower left. The hypotenuse is labelled a, the vertical side opposite the angle is labelled x, and the horizontal side is labelled the square root of a squared minus x squared.

If sin y is x over a, the triangle has hypotenuse a and opposite side x, so the third side is the square root of a squared minus x squared. That side is where the root in the formula comes from.

Replace cos y using the Pythagorean identity

Why: The angle y lies between minus and plus a right angle, where cosine is not negative, so the positive root is correct.

\[ \cos y = \sqrt{1-\sin^2 y} = \sqrt{1 - \frac{x^2}{a^2}} \]

Move the a inside the root

Why: a equals the square root of a squared, since a is positive.

\[ a\cos y = \sqrt{a^2 - a^2\sin^2 y} = \sqrt{a^2 - x^2} \]

Read the derivative as an antiderivative

Why: Valid for x strictly between minus a and a.

\[ \frac{d}{dx}\sin^{-1}\frac{x}{a} = \frac{1}{\sqrt{a^2-x^2}} \;\Longrightarrow\; \int\frac{dx}{\sqrt{a^2-x^2}} = \sin^{-1}\frac{x}{a} + C \]

Check at a point

Why: Take a equal to 2 and x equal to 1: the chain-rule derivative and the integrand give the same number.

\[ \frac{1/2}{\sqrt{1-1/4}} = \frac{1}{\sqrt3} = \frac{1}{\sqrt{4-1}} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 527 — Proof of formula 5.23

This is the book's proof, and it is really the Section 3.7 method again. Rewrite y equals the inverse sine of x over a as a times sine y equals x, then differentiate both sides with respect to x, remembering that y is a function of x.

Solving gives one over a cosine y, which still has y in it. To get back to x you need cosine in terms of sine, and the triangle shows how: if the opposite side is x and the hypotenuse is a, the adjacent side is the square root of a squared minus x squared. That is where the root in the formula comes from.

The step people skip is the sign. Cosine could in principle be negative, but the inverse sine only returns angles between minus and plus a right angle, where cosine is never negative. That is why the positive square root is correct. The final check at a equal to 2 and x equal to 1 confirms the result with actual numbers.

13. Why the inverse sine carries no one over a

Picture it

Figure (svg): Two panels. Left: the curves one over the square root of a squared minus x squared for a equals 1 and a equals 2; the wider curve is lower. Right: the curves one over a squared plus x squared for a equals 1 and a equals 2; the second is much lower and only a little wider.

Scaling the inverse-sine integrand by a stretches it by a sideways and shrinks it by a vertically, so the area is unchanged. The arctangent integrand shrinks by a squared, so a factor of 1/a survives.

Replace 1 by a in each integrand and look at the whole area. The inverse-sine integrand gets a times wider and a times shorter, so the area stays pi. The arctangent integrand gets a times wider but a squared times shorter, so the area falls to pi over a. That leftover factor is the one over a in the formula.

This picture explains the missing factor without any algebra. On the left, the inverse-sine integrand for a equal to 1 lives between minus 1 and 1; for a equal to 2 it lives between minus 2 and 2, twice as wide, and it is half as tall. Stretching sideways by 2 and squashing by 2 leaves the area exactly the same: pi in both cases.

On the right, the arctangent integrand for a equal to 2 is also stretched sideways by 2, but its height at the centre drops from 1 to one quarter. Twice as wide and a quarter as tall halves the area, from pi to pi over two.

That leftover halving is the one over a. A square root turns a squared into a, which the stretch exactly undoes; without a root the a squared is too strong, and one factor of a is left over in front of the answer.

14. Trap: a one over a on the inverse sine

Trap

The trap

Memorising the arctangent formula first, then copying its shape:

\[ \int\frac{dx}{\sqrt{9-x^2}} \]

\[ = \frac13\sin^{-1}\frac{x}{3} + C \quad ? \]

Wrong. Only the tangent and secant carry the factor.

The fix

Differentiate the claimed answer. The chain rule gives one third times one third over a root, which is one third of the integrand, not the integrand.

\[ \int\frac{dx}{\sqrt{9-x^2}} = \sin^{-1}\frac{x}{3} + C \]

This mistake comes from learning the arctangent formula first and then assuming the inverse sine looks the same. The two formulas really do differ, and the only reliable defence is to differentiate your answer.

Try it on the wrong answer. The derivative of the inverse sine of x over 3 is one third over the root of one minus x squared over 9, which simplifies to one over the root of 9 minus x squared. That is already the integrand. So the extra one third in front makes the derivative a third too small.

Differentiating an answer takes a few seconds and catches every factor error in this lesson. Make it a habit on every indefinite integral you finish.

15. Integrals that give an inverse sine

Section

Part 2

16. Definite answers are angles

Concept

Evaluating a definite integral here means evaluating inverse trigonometric functions at the limits. The answers are angles in radians, and a handful of values come up again and again.

inputsin⁻¹tan⁻¹sec⁻¹
000—
1/2π/6
√2/2 (sine) or √2 (secant)π/4π/4
√3/2 (sine) or √3/3 (tangent)π/3π/6
1π/2π/40
√3 (tangent) or 2 (secant)π/3π/3

Each entry is a statement about a right triangle: for instance, the angle whose sine is one half is pi over six, a thirty-degree angle.

When you evaluate a definite integral in this section, you end up evaluating inverse trigonometric functions at the limits. Those are angles, and exam problems are designed so that they come out as familiar ones.

Read each entry as a triangle statement. The inverse sine of one half is pi over six because a thirty-degree angle has opposite side half the hypotenuse. The inverse tangent of 1 is pi over four because a forty-five-degree angle has equal legs. The inverse secant of 2 is pi over three because its cosine is one half.

If you forget a value, go back to the triangles with angles of thirty, forty-five and sixty degrees and read it off. Also remember that the answers are in radians, never degrees, because the calculus formulas are only true in radians.

17. Example 5.49: a definite integral straight from the formula

Worked example

Evaluate the definite integral.

\[ \int_0^{1/2}\frac{dx}{\sqrt{1-x^2}} \]

Match the form

Why: This is the inverse sine formula with a equal to 1 and u equal to x.

\[ \int\frac{dx}{\sqrt{1-x^2}} = \sin^{-1}x + C \]

Apply the Fundamental Theorem

Why: Evaluate the antiderivative at the limits.

\[ \int_0^{1/2}\frac{dx}{\sqrt{1-x^2}} = \sin^{-1}x\Big|_0^{1/2} \]

Subtract the two values

Why: The angle whose sine is one half is pi over six.

\[ = \sin^{-1}\tfrac12 - \sin^{-1}0 = \frac{\pi}{6} - 0 \]

Simplify

Why: Keep the exact value, then a decimal.

\[ = \frac{\pi}{6} \approx 0.5236 \]

Figure (svg): The curve y equals one over the square root of one minus x squared on the interval from minus 1 to 1, U-shaped with vertical asymptotes at x equals minus 1 and 1. The region under it from 0 to one half is shaded; its area is pi over six.

Between 0 and one half the curve barely rises above height 1, so the area is just over one half: pi over 6 is 0.5236.

Check against the picture

Why: On this interval the integrand rises from 1 to 1.1547, so the area lies between two rectangles of width one half.

\[ 0.5 \cdot 1 = 0.5 < 0.5236 < 0.5774 = 0.5 \cdot 1.1547 \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 527 — Example 5.49

This is the simplest possible case: a is 1 and u is just x, so there is no substitution at all. The whole problem is recognising the form, writing down the inverse sine, and evaluating it at the limits.

The inverse sine of one half is pi over six and the inverse sine of 0 is 0, so the integral is pi over six, about 0.5236. Keep the exact answer; the decimal is only for checking.

Now look at the picture and the check. On the interval from 0 to one half the integrand hardly rises, from 1 to about 1.15. So the area must be between that of a rectangle of height 1 and one of height 1.15, both of width one half. The answer 0.5236 sits between 0.5 and 0.577, just above the lower one, exactly as the flat-looking curve suggests.

18. What the inverse sine integral measures

Picture it

Figure (svg): The unit circle, with a radius drawn to the point at angle theta where the height is 0.6. The arc of the circle from the positive x-axis up to that point is highlighted; its length is theta, the inverse sine of 0.6, about 0.644.

The inverse sine of 0.6 is an angle, and on the unit circle an angle is an arc length. The integral from 0 to 0.6 of one over the root adds up exactly that arc.

Walk up the unit circle from the point 1, 0 until your height is x. The length of arc you walked is the inverse sine of x, and the integral from 0 to x of one over the root of one minus t squared adds up exactly that arc. The integrand is how fast arc length grows per unit of height, and it blows up near the top, where the circle turns horizontal.

This picture gives the inverse sine integral a meaning you can see. Start at the point 1, 0 on the unit circle and walk counterclockwise. When your height reaches 0.6, you have walked an arc whose length is the angle, and that angle is the inverse sine of 0.6, about 0.644.

The integral from 0 to x of one over the root of one minus t squared adds up that arc length a sliver at a time. The integrand is the rate at which arc length grows as your height grows. Near the bottom of the arc you are climbing almost straight up, so a small arc buys almost the same height and the rate is close to 1.

Near the top of the circle the path turns horizontal, and you need a lot of arc to gain a little height. That is why the integrand blows up as x approaches 1: the circle is going sideways, not up.

19. Checkpoint 5.40: sixteen x squared under the root

Fill the middle

\[ \int\frac{dx}{\sqrt{1-16x^2}} \]

Fill in the blanks

u = 4x, \quad \int\frac\frac{1}{4}4x} = ___\sin^___(___) + C

Why: Sixteen x squared is the square of 4x, so u is 4x and du is 4 dx, which forces a factor of one quarter outside. With a equal to 1 the answer is one quarter of the inverse sine of 4x, plus C. Differentiating: one quarter times 4 over the root of one minus 16 x squared, which is the integrand.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 528 — Checkpoint 5.40

Fill in the three blanks before checking. The first question is always what is being squared. Sixteen x squared is the square of 4x, so that is u.

Once u is 4x, du is 4 dx, and the integral only has dx. So you need a factor of one quarter outside to supply the 4. That quarter comes from the substitution, not from the formula, which is why it is allowed even though the inverse sine formula itself has no factor in front.

With a equal to 1, the answer is one quarter of the inverse sine of 4x. Differentiate it to see the chain rule's 4 cancel the quarter, and you will be sure it is right.

20. Example 5.50: substitute, then use a equal to 2

Worked example

Evaluate the integral.

\[ \int\frac{dx}{\sqrt{4-9x^2}} \]

Write both terms as squares

Why: Four is 2 squared and nine x squared is 3x squared.

\[ 4 - 9x^2 = 2^2 - (3x)^2 \]

Substitute u equal to 3x

Why: Then du is 3 dx, so dx is a third of du.

\[ u = 3x, \quad du = 3\,dx, \quad dx = \tfrac13\,du \]

Rewrite the integral in u

Why: The one third moves outside.

\[ \int\frac{dx}{\sqrt{4-9x^2}} = \frac13\int\frac{du}{\sqrt{4-u^2}} \]

Apply formula 5.23 with a equal to 2

Why: No extra factor from the formula itself.

\[ = \frac13\sin^{-1}\left(\frac u2\right) + C \]

Return to x

Why: Replace u by 3x.

\[ = \frac13\sin^{-1}\left(\frac{3x}{2}\right) + C \]

Check by differentiating

Why: The chain rule gives three halves inside; the root simplifies by pulling out a quarter.

\[ \frac13\cdot\frac{3/2}{\sqrt{1-9x^2/4}} = \frac{1/2}{\tfrac12\sqrt{4-9x^2}} = \frac{1}{\sqrt{4-9x^2}} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 528 — Example 5.50

Here two things happen at once, and it helps to separate them. The 4 is a constant squared, so a is 2. The nine x squared is the square of 3x, so u is 3x. Writing the denominator as 2 squared minus 3x squared makes both visible.

The substitution u equal to 3x needs du equal to 3 dx, so dx is a third of du. That one third is the only factor in the answer. The formula itself, being the inverse sine form, adds nothing; it only turns u over a into 3x over 2 inside the inverse sine.

The check shows all the pieces fitting back together. The chain rule brings out three halves, the one third makes it one half, and pulling a quarter out of the root turns the one half into exactly 1 over the root of 4 minus nine x squared.

21. Checkpoint 5.41: nine minus x squared

Prediction

\[ \int\frac{dx}{\sqrt{9-x^2}} \]

Predict first

Which is the antiderivative?

  • sin⁻¹(x/3) + C
  • (1/3) sin⁻¹(x/3) + C
  • sin⁻¹(3x) + C
  • (1/3) tan⁻¹(x/3) + C

Correct: sin⁻¹(x/3) + C

Why: Nine is 3 squared, so a is 3 and u is simply x. The inverse sine formula has no one over a in front, so the answer is the inverse sine of x over 3, plus C. Differentiating gives one third over the root of one minus x squared over 9, which is one over the root of 9 minus x squared.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 528 — Checkpoint 5.41

Choose an option before you reveal the answer. The distractors are built from the three most common slips in this section, so if you choose one, it tells you exactly what to fix.

Nine is 3 squared, so a is 3, and u is just x, so there is no substitution factor. The inverse sine formula has no factor of its own either. The answer is the inverse sine of x over 3, plus C.

If you picked the version with one third in front, you borrowed the tangent's factor. If you picked the inverse sine of 3x, you multiplied by a instead of dividing. If you picked the tangent, look again at the integrand: a square root around a difference of squares is always the sine form.

22. Example 5.51: an upper limit of root 3 over 2

Worked example

Evaluate the definite integral.

\[ \int_0^{\sqrt3/2}\frac{du}{\sqrt{1-u^2}} \]

Match the form

Why: Inverse sine with a equal to 1.

\[ \int_0^{\sqrt3/2}\frac{du}{\sqrt{1-u^2}} = \sin^{-1}u\Big|_0^{\sqrt3/2} \]

Evaluate at both limits

Why: The angle whose sine is root 3 over 2 is pi over three.

\[ = \sin^{-1}\left(\frac{\sqrt3}{2}\right) - \sin^{-1}(0) = \frac{\pi}{3} - 0 \]

Simplify

Why: Exact, then decimal.

\[ = \frac{\pi}{3} \approx 1.0472 \]

Check with bounding rectangles

Why: The integrand rises from 1 to 2 across an interval of width 0.866, so the area is between 0.866 and 1.732.

\[ 0.866 < 1.0472 < 1.732, \qquad \sin\frac{\pi}{3} = \frac{\sqrt3}{2} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 528 — Example 5.51

This definite integral matches the inverse sine form immediately, with a equal to 1. The only work is knowing the value of the inverse sine at root 3 over 2, which is pi over three: a sixty-degree angle has opposite side root 3 over 2 when the hypotenuse is 1.

Notice that the upper limit, about 0.866, is getting close to 1, where the integrand blows up. The integrand climbs from 1 at the lower limit to exactly 2 at the upper one. That is why this area, about 1.047, is noticeably bigger than the width of the interval.

The rectangle check confirms it: the area is trapped between 0.866 and 1.732, and 1.047 is inside. The final line also confirms the special value by going forwards: the sine of pi over three really is root 3 over 2.

23. The other three inverse functions

Concept

The derivatives of inverse cosine, cotangent and cosecant are the negatives of the three you know. So their integrals are the negatives too, and you never need them: factor out minus 1 and use the formulas you have.

\[ \int\frac{-dx}{\sqrt{1-x^2}} = -\sin^{-1}x + C \quad\text{or}\quad \cos^{-1}x + C \]

Figure (svg): Two decreasing curves on the interval from minus 1 to 1: y equals inverse cosine of x, running from pi down to 0, and y equals minus inverse sine of x, running from pi over two down to minus pi over two. A vertical arrow between them shows a constant gap of pi over two.

Both curves have the same slope everywhere, so they are both antiderivatives of the same integrand. They differ by a constant, pi over two, which the plus C absorbs.

Both answers are correct. Inverse cosine and minus inverse sine differ by the constant pi over two, and the constant of integration absorbs it.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 528 — Integrals Resulting in Other Inverse Trigonometric Functions

The book deliberately gives only three formulas, and this slide shows why that is enough. The derivative of the inverse cosine is the negative of the derivative of the inverse sine. So an integrand with a minus sign in front can be handled by factoring out minus 1 and using the inverse sine formula.

That gives minus the inverse sine as an answer. You could also write the inverse cosine, and the picture shows why both are right. The two curves have the same shape, one shifted by exactly pi over two. They have the same slope at every point, so they are antiderivatives of the same function.

The constant difference is absorbed by the plus C. That is why the answer to an indefinite integral can look quite different from one book or calculator to another and still be correct.

24. Is inverse cosine minus inverse sine?

Counterexample

\[ -\cos^{-1}t + C = \int\frac{dt}{\sqrt{1-t^2}} = \sin^{-1}t + C \]

Discussion prompt

Exercise 403 shows both sides are antiderivatives of the same integrand. Does that mean the inverse cosine of t equals minus the inverse sine of t? Test one value of t, then say what is really true.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 530 — Exercise 403

Write your answer before revealing. The exercise sets up a tempting argument: both sides equal the same integral, so surely they equal each other. A single test value shows the argument fails.

At t equal to 0, the inverse cosine is pi over two and minus the inverse sine is 0. They are not equal. What is true is that they differ by a constant, and the constant is pi over two, so the inverse cosine plus the inverse sine is always pi over two.

The general lesson is about the plus C. When two expressions are both antiderivatives of the same function, you may only conclude that they differ by a constant. The two C's in an equation like this one are different constants, and treating them as the same one is exactly the mistake the exercise is testing.

25. Integrals that give an inverse tangent

Section

Part 3

26. Example 5.52: one plus four x squared

Worked example

Evaluate the integral.

\[ \int\frac{1}{1+4x^2}\,dx \]

Write four x squared as a square

Why: It is 2x squared, so the denominator is one plus u squared.

\[ 1 + 4x^2 = 1 + (2x)^2 \]

Substitute u equal to 2x

Why: Then du is 2 dx, so dx is half of du.

\[ u = 2x, \quad du = 2\,dx, \quad dx = \tfrac12\,du \]

Rewrite the integral in u

Why: The half moves outside.

\[ \int\frac{dx}{1+4x^2} = \frac12\int\frac{du}{1+u^2} \]

Apply formula 5.24 with a equal to 1

Why: The factor one over a is just 1.

\[ = \frac12\tan^{-1}u + C \]

Return to x

Why: Replace u by 2x.

\[ = \frac12\tan^{-1}(2x) + C \]

Check by differentiating

Why: The chain rule's 2 cancels the half.

\[ \frac{d}{dx}\left[\frac12\tan^{-1}(2x)\right] = \frac12\cdot\frac{2}{1+4x^2} = \frac{1}{1+4x^2} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Example 5.52

The denominator is one plus four x squared. Four x squared is the square of 2x, so this is the tangent form with u equal to 2x and a equal to 1. The whole problem is managing the substitution factor.

Since u is 2x, du is 2 dx, and the integral only has dx. So dx is half of du, and the half comes outside. The formula with a equal to 1 contributes no extra factor, so the answer is half the inverse tangent of 2x.

The check is the best part: differentiate, and the chain rule produces a 2 that cancels the half, leaving exactly the integrand. If you had forgotten the half, the check would have produced twice the integrand and you would have caught the mistake yourself.

27. Trap: dropping the factor from du

Trap

The trap

Spotting the arctangent shape and writing the answer at once:

\[ \int\frac{dx}{1+4x^2} \]

\[ = \tan^{-1}(2x) + C \quad ? \]

Wrong by a factor of two.

The fix

Differentiating the claimed answer gives 2 over one plus four x squared: twice the integrand. The substitution u equal to 2x needs du equal to 2 dx, and the missing 2 must be paid for with a half outside.

\[ = \frac12\tan^{-1}(2x) + C \]

This is the most common error in the whole section, and it happens because the shape is so recognisable. You see one plus something squared, you write the inverse tangent of that something, and you forget that the something is not x.

The substitution u equal to 2x changes the differential as well as the denominator. The integral only contains dx, but the formula needs du, which is 2 dx. The half out front pays for that missing 2.

The fastest way to catch it is to differentiate your answer. The chain rule will produce the inner derivative, 2, and if nothing cancels it, your answer is off by that factor.

28. Checkpoint 5.42: twenty-five plus four x squared

Fill the middle

\[ \int\frac{dx}{25+4x^2} \]

Fill in the blanks

u = 2x,\; a = 5: \quad \int\frac105 = \frac______} }\tan^___\left(\frac______} }\right) + C

Why: Twenty-five is 5 squared, so a is 5, and four x squared is the square of 2x. The substitution contributes one half (from du equal to 2 dx) and the formula contributes one over a, one fifth, so the factor in front is one tenth. Check: one tenth times two fifths over one plus four x squared over 25 is one over 25 plus four x squared.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Checkpoint 5.42

This checkpoint combines both kinds of factor, which is why it is worth doing slowly. The constant 25 is 5 squared, so a is 5. The four x squared is the square of 2x, so u is 2x.

Now count the factors. The substitution contributes one half, because du is 2 dx. The formula contributes one over a, which is one fifth. Multiply them and you get one tenth in front. Inside the inverse tangent is u over a, which is 2x over 5.

If you got one fifth in front, you forgot the substitution factor; if you got one half, you forgot the formula's factor. The check by differentiating uses all three pieces: one tenth, times two fifths from the chain rule, over one plus four x squared over 25.

29. Example 5.53: nine plus x squared

Worked example

Evaluate the integral.

\[ \int\frac{1}{9+x^2}\,dx \]

Identify a and u

Why: Nine is 3 squared and u is x itself, so du is dx and no substitution is needed.

\[ a^2 = 9 \;\Longrightarrow\; a = 3, \quad u = x \]

Apply formula 5.24

Why: This time the factor one over a is one third.

\[ \int\frac{dx}{9+x^2} = \frac13\tan^{-1}\left(\frac{x}{3}\right) + C \]

Check by differentiating

Why: Two factors of one third combine to one ninth, which rebuilds the 9.

\[ \frac13\cdot\frac{1/3}{1+x^2/9} = \frac{1/9}{(9+x^2)/9} = \frac{1}{9+x^2} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Example 5.53

This is the tangent form with u simply equal to x, so no substitution is needed and the only factor is the formula's own one over a. Nine is 3 squared, so a is 3 and the answer is one third of the inverse tangent of x over 3.

The mistake to avoid is taking a to be 9. The constant in the denominator is a squared, not a. Always take the square root of the constant to find a.

In the check, notice how the two factors of one third combine. One comes from the formula, one from the chain rule on x over 3, and together they make one ninth, which is exactly what is needed to turn one plus x squared over 9 back into 9 plus x squared.

30. Checkpoint 5.43: sixteen plus x squared

Prediction

\[ \int\frac{dx}{16+x^2} \]

Predict first

Which is the antiderivative?

  • (1/4) tan⁻¹(x/4) + C
  • tan⁻¹(x/4) + C
  • (1/16) tan⁻¹(x/16) + C
  • (1/4) tan⁻¹(4x) + C

Correct: (1/4) tan⁻¹(x/4) + C

Why: Sixteen is 4 squared, so a is 4, not 16. The arctangent formula puts one over a in front and x over a inside: one quarter of the inverse tangent of x over 4. Differentiating gives one sixteenth over one plus x squared over 16, which is one over 16 plus x squared.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Checkpoint 5.43

Pick an answer before revealing. Each wrong option corresponds to one specific slip, so notice which one tempted you.

Sixteen is 4 squared, so a is 4. The tangent formula needs one over a in front and x over a inside, giving one quarter of the inverse tangent of x over 4. Leaving out the quarter is the classic mistake of treating the tangent like the sine. Using 16 instead of 4 forgets to take the square root. Writing 4x inside multiplies instead of dividing.

Differentiate the correct answer to confirm: one quarter times one quarter is one sixteenth, and one sixteenth over one plus x squared over 16 is one over 16 plus x squared.

31. Example 5.54: from root 3 over 3 to root 3

Worked example

Evaluate the definite integral.

\[ \int_{\sqrt3/3}^{\sqrt3}\frac{dx}{1+x^2} \]

Use the inverse tangent with a equal to 1

Why: Evaluate at the limits.

\[ \int_{\sqrt3/3}^{\sqrt3}\frac{dx}{1+x^2} = \tan^{-1}x\Big|_{\sqrt3/3}^{\sqrt3} \]

Evaluate at both limits

Why: Tangent of pi over three is root 3; tangent of pi over six is root 3 over 3.

\[ = \tan^{-1}\sqrt3 - \tan^{-1}\frac{\sqrt3}{3} = \frac{\pi}{3} - \frac{\pi}{6} \]

Subtract

Why: Exact, then decimal.

\[ = \frac{\pi}{6} \approx 0.5236 \]

Figure (svg): The curve y equals one over one plus x squared from minus 0.3 to 2.6, with the region under it from the square root of 3 over 3 to the square root of 3 shaded; the shaded area is pi over six.

The integrand falls from three quarters to one quarter across an interval about 1.15 wide, so the area must lie between 0.29 and 0.87. Pi over 6 is 0.524.

Check with bounding rectangles

Why: The integrand falls from 0.75 to 0.25 over an interval of width 1.1547.

\[ 0.25 \cdot 1.1547 = 0.289 < 0.5236 < 0.866 = 0.75 \cdot 1.1547 \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Example 5.54

This definite integral uses the basic inverse tangent with a equal to 1. The limits are chosen to be tangents of special angles: root 3 is the tangent of pi over three, and root 3 over 3 is the tangent of pi over six.

So the answer is pi over three minus pi over six, which is pi over six, about 0.524. It is a coincidence, and a nice one, that this is the same number as Example 5.49.

The picture and the check work together. Across the interval the curve falls from three quarters to one quarter, and the interval is about 1.15 wide. So the area lies between 0.29 and 0.87. The answer sits comfortably inside, closer to the middle because the curve falls fairly evenly across the interval.

32. Checkpoint 5.44: estimate before you integrate

Estimation

\[ \int_0^{2}\frac{dx}{4+x^2} \]

Predict first

The integrand falls from 1/4 at x = 0 to 1/8 at x = 2. Roughly what is the integral?

  • about 0.2
  • about 0.4
  • about 0.8
  • about 1.6

Correct: about 0.4

Why: A width of 2 times an average height between one eighth and one quarter gives between 0.25 and 0.5, and the trapezoid gives 0.375. The exact value is one half of the inverse tangent of 1, which is pi over 8, about 0.393.

\[ \int_0^{2}\frac{dx}{4+x^2} = \frac12\tan^{-1}\frac{x}{2}\Big|_0^{2} = \frac12\cdot\frac{\pi}{4} = \frac{\pi}{8} \approx 0.393 \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 529 — Checkpoint 5.44

Make a rough estimate before doing any integration. You know the integrand at both ends: one quarter at x equal to 0 and one eighth at x equal to 2. The interval is 2 wide, so the area must lie between one quarter and one half.

The trapezoid through the two end heights gives 0.375, and since the curve bulges slightly above the straight line between its ends, the true value should be a little more. The formula gives one half of the inverse tangent of 1, which is pi over eight, about 0.393. That fits.

Getting in the habit of estimating first protects you from factor errors. If you had forgotten the one half and got pi over four, about 0.785, your estimate would have told you instantly that the answer was too big.

33. Slide a and watch the arctangent flatten

Tweak it

Parameter explorer

The curve is the antiderivative (1/a) tan⁻¹(x/a), the area under 1/(a² + t²) from 0 to x. Slide a. What happens to the height it levels off at, and why?

\[ \int_0^{x}\frac{dt}{{a}^2+t^2} = \frac{1}{{a}}\tan^{-1}\frac{x}{{a}} \]

  • a — from 1 to 4: constant a

Start with a equal to 1: the curve is the ordinary inverse tangent, levelling off at about 1.57, which is pi over two. Now slide a up slowly and watch two things happen together.

The curve stretches sideways, because x over a changes more slowly, so the rise takes place over a wider range of x. And the whole curve squashes down, because of the one over a in front. The level it approaches is pi over two times one over a: at a equal to 2 it is about 0.79, at a equal to 4 about 0.39.

This is the same effect as the scaling picture earlier, now seen through the antiderivative. The total area under one over a squared plus t squared shrinks like one over a, and that is exactly the factor in the formula.

34. The inverse tangent integrand is defined everywhere

Intuition

Of the three forms, only the arctangent's denominator can never be zero: a sum of two squares with a positive constant is always positive.

\[ a^2 + u^2 \ge a^2 > 0 \quad \text{for every } u \]

So the formula needs no domain check, and the integrand can be integrated over any interval at all, even an infinitely long one. The next example does exactly that.

Of the three forms, only the tangent's denominator can never vanish. A positive constant squared plus any square is at least the constant squared, which is positive. So the integrand is defined and continuous for every real number.

That has a practical consequence: whenever you use the tangent formula on a definite integral, you do not need to check the interval. With the sine and secant forms you always do.

It also means you can ask what happens over an infinitely long interval. Because the integrand falls off like one over x squared far out, it turns out that the whole infinitely long region under it has finite area. The next example computes it.

35. Exercise 433: the area under the whole curve

Worked example

For positive A, compute I of A and its limit as A grows: the area under one over one plus t squared over the whole line.

\[ I(A) = \int_{-A}^{A}\frac{dt}{1+t^2} \]

Evaluate with the inverse tangent

Why: The formula applies on every interval.

\[ I(A) = \tan^{-1}t\Big|_{-A}^{A} = \tan^{-1}A - \tan^{-1}(-A) \]

Use that the inverse tangent is odd

Why: The inverse tangent of minus A is minus the inverse tangent of A.

\[ I(A) = 2\tan^{-1}A \]

Let A grow

Why: The inverse tangent approaches a right angle.

\[ \lim_{A\to\infty}I(A) = 2\cdot\frac{\pi}{2} = \pi \]

Figure (svg): Two panels. Left: the curve one over one plus t squared from minus 6 to 6 with the region from minus A to A shaded, A equal to 4. Right: the area 2 times inverse tangent of A plotted against A from 0 to 20, rising toward a dashed line at pi without reaching it.

However far out you go, the area never passes pi. The infinitely long region under the curve has a finite area, and it is exactly pi.

Check with numbers

Why: At A equal to 10 and 100 the areas are already close to pi and still below it.

\[ I(10) = 2.9423, \quad I(100) = 3.1216, \quad \pi = 3.1416 \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 531 — Exercise 433

Here you compute the area under one over one plus t squared from minus A to A, and then let A grow without bound. The antiderivative is the inverse tangent, and because it is an odd function, the value at minus A is the negative of the value at A. The two terms add to twice the inverse tangent of A.

As A grows, the inverse tangent of A approaches pi over two, so the total area approaches pi. Look at the right-hand panel: the area climbs quickly at first, then creeps toward the dashed line at pi without ever reaching it.

This is your first taste of an improper integral, which Calculus II studies properly. An infinitely long region can have a finite area, provided the height shrinks fast enough. The numbers in the check, 2.94 at A equal to 10 and 3.12 at A equal to 100, show how slowly the last bit of area arrives.

36. A spinning light and the inverse tangent

Real world

Figure (svg): The bell-like curve y equals one over pi times one plus x squared, for x from minus 6 to 6, peaking at about 0.32 at x equals 0. The region from minus 1 to 1 is shaded and labelled one half.

Dividing the arctangent integrand by pi makes the total area 1, so it is a probability density. Half the probability sits between minus 1 and 1.

Discussion prompt

A lamp 1 metre from a long straight wall points at a uniformly random angle toward the wall. Where the beam lands, x metres along the wall from the nearest point, has probability density 1/(π(1 + x²)). What is the probability the beam lands within 1 metre of the nearest point? Within 10 metres?

Work out both answers before you reveal them. The set-up is a lamp one metre from a wall, shining at a random angle. The spot on the wall is at x equal to the tangent of that angle, and that is why the inverse tangent appears.

Divide the arctangent integrand by pi and its total area becomes 1, so it can serve as a probability density. The probability of landing within a metre of the nearest point is the area from minus 1 to 1, which is two over pi times pi over four: exactly one half.

Within ten metres the probability is about 0.937, so about six percent of flashes land more than ten metres away. The tails of this curve are so heavy that this distribution, the Cauchy distribution, has no average value at all. That makes it a standard warning example in statistics.

37. The inverse secant, and domains

Section

Part 4

38. The inverse secant formula

Concept

The third formula, 5.25, has u outside the root as well as inside, and the square under the root is the other way round: the variable squared minus the constant squared.

\[ \int\frac{du}{u\sqrt{u^2-a^2}} = \frac{1}{a}\sec^{-1}\left(\frac{|u|}{a}\right) + C \]

Like the arctangent, it carries a factor of one over a. Unlike both others, the integrand only exists when u is further from zero than a is.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 527 — formula 5.25

The third formula is the least used, but you need to recognise it. It is the one with u outside the root as well as inside, and the square under the root is the variable squared minus the constant squared, the reverse of the sine form.

Like the tangent formula, it has one over a in front. And like the sine formula, it has a domain restriction, but the opposite one: the root is only real when u is further from zero than a is.

The absolute value inside the inverse secant is there so that the formula also works when u is negative. The next slide shows exactly why it is needed.

39. Why the absolute value in the secant formula

Concept

For negative u, the integrand u times the root is negative, so the integral must be decreasing there. The derivative of the inverse secant of the absolute value handles both signs at once.

\[ \frac{d}{du}\sec^{-1}|u| = \frac{1}{|u|\sqrt{u^2-1}}\cdot\frac{u}{|u|} = \frac{1}{u\sqrt{u^2-1}} \]

The factor u over its absolute value is plus 1 or minus 1, the sign of u. It turns the absolute value in the denominator back into plain u, exactly as the integrand needs.

For negative u, the integrand u times a positive root is negative, so any antiderivative must be decreasing there. The plain inverse secant is increasing on its whole domain, so it cannot be right for negative u on its own.

The absolute value fixes this. By the chain rule, differentiating the absolute value of u gives u over the absolute value of u, which is plus 1 for positive u and minus 1 for negative u. That sign factor turns the absolute value in the denominator back into plain u.

In practice, most problems use positive u, where the absolute value can simply be dropped, as in Exercise 396. But if an interval of integration lies to the left of minus a, remember that the absolute value is doing real work.

40. Where each formula lives

Picture it

Figure (svg): Three number lines, one per formula. Inverse sine: a thick segment strictly between minus a and a, with open circles at both ends. Inverse tangent: the whole line is thick, with arrows at both ends. Inverse secant: two thick rays, one to the left of minus a and one to the right of a, with open circles at minus a and a.

The arcsine formula works only strictly inside the interval, the arcsecant only strictly outside it, and the arctangent everywhere. A definite integral must stay inside the thick part.

Each formula comes with an interval. The inverse sine needs the quantity under the root to be positive, so u must lie strictly between minus a and a. The inverse secant needs the opposite. The inverse tangent works everywhere. A definite integral whose interval leaves the thick part of its line is not a real-number integral at all.

These three number lines are the domains of the three integrands, and they are the thing most often forgotten in this section. For the sine form, u must be strictly between minus a and a, with open circles because the integrand blows up at the ends.

For the tangent form, every u is allowed. For the secant form, u must be strictly outside the interval, further from zero than a, on either side.

The rule for definite integrals is simple: the whole interval of integration must lie inside the thick part of the line. If it does not, the integrand is not a real number somewhere on the interval, and the integral does not exist. The error-analysis slide in a moment shows what happens when this is ignored.

41. Exercise 396: an inverse secant from root 2 to 2

Worked example

Evaluate the definite integral in terms of an inverse trigonometric function.

\[ \int_{\sqrt2}^{2}\frac{dx}{|x|\sqrt{x^2-1}} \]

Check the interval lies in the domain

Why: Every x from root 2 to 2 is bigger than 1, so the integrand is defined.

\[ \sqrt2 \approx 1.414 > 1 \]

Use the inverse secant with a equal to 1

Why: For positive x the absolute value is just x.

\[ \int_{\sqrt2}^{2}\frac{dx}{x\sqrt{x^2-1}} = \sec^{-1}x\Big|_{\sqrt2}^{2} \]

Evaluate at both limits

Why: Secant of pi over three is 2, and secant of pi over four is root 2.

\[ = \sec^{-1}2 - \sec^{-1}\sqrt2 = \frac{\pi}{3} - \frac{\pi}{4} \]

Subtract

Why: Common denominator 12.

\[ = \frac{\pi}{12} \approx 0.2618 \]

Figure (svg): The curve y equals one over the absolute value of x times the square root of x squared minus 1, drawn in two pieces, for x greater than 1 and for x less than minus 1, each falling from a vertical asymptote. The strip between minus 1 and 1 is tinted red: the integrand is undefined there. The region from root 2 to 2 is shaded.

The inverse-secant integrand lives only outside the interval from minus 1 to 1. The shaded sliver from root 2 to 2 is the area in Exercise 396.

Check the special values

Why: The inverse secant of y is the inverse cosine of one over y, and a numerical integration agrees to six places.

\[ \sec^{-1}2 = \cos^{-1}\tfrac12 = \frac{\pi}{3}, \quad \sec^{-1}\sqrt2 = \cos^{-1}\frac{\sqrt2}{2} = \frac{\pi}{4} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 530 — Exercise 396

Start with the domain check, because this is the secant form. Every x from root 2 to 2 is bigger than 1, so the integrand is defined throughout and the formula may be used. For positive x the absolute value is just x.

The antiderivative is the inverse secant of x, and the limits are the secants of special angles: 2 is the secant of pi over three, and root 2 is the secant of pi over four. The answer is pi over three minus pi over four, which is pi over twelve, about 0.262.

If you find inverse secants awkward, convert them: the inverse secant of y is the inverse cosine of one over y. The check does exactly that, and a numerical integration of the original confirms the value. The picture shows how thin the shaded sliver is, which fits a small answer.

42. Find the error: an integral outside the domain

Error analysis

Figure (svg): The curve one over the square root of one minus t squared, drawn only between minus 1 and 1. On the t-axis, the interval from 1 to 2 is drawn in thick red, to the right of the asymptote at t equals 1, where no curve exists.

One minus t squared is negative for every t between 1 and 2, so the integrand is not a real number anywhere on that interval. There is no area to compute.

Annotate

On: \( \int_1^{2}\frac{dt}{\sqrt{1-t^2}} = \sin^{-1}t\Big|_1^{2} = \sin^{-1}2 - \frac{\pi}{2} \)

  • For every t between 1 and 2 this is negative, so the square root is not a real number anywhere on the interval.
  • No angle has sine 2. The answer is meaningless, and that is the formula telling you it was used outside its domain.
  • There is none: the integral does not exist as a real number. Check the interval against the formula's domain BEFORE integrating.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 530 — Exercise 405

Look at the picture before the annotations. The curve exists only between minus 1 and 1; the interval of integration, from 1 to 2, lies entirely to the right of that, where there is no curve at all.

The calculation goes through mechanically, and that is exactly the danger. It produces the inverse sine of 2, which is not a real number, since no angle has a sine bigger than 1. The formula is announcing, too late, that it was used outside its domain.

There is no fix to this integral: it simply does not exist as a real number. The lesson is to check the interval against the domain before you integrate, especially with the sine and secant forms. Exercise 406 is the same mistake with the secant form, integrating across the gap between minus 1 and 1.

43. Complete the table of the three forms

Comparison

Comparison matrix

integrandantiderivativevalid for
1/√(a² − u²)sin⁻¹(u/a) + C−a < u < a
1/(a² + u²)(1/a) tan⁻¹(u/a) + Cevery u
1/(u√(u² − a²))(1/a) sec⁻¹(|u|/a) + C|u| > a

Fill in each blank from memory before checking. This table is the whole lesson in compressed form, and being able to produce it from nothing is a good sign you are ready for the exercises.

Check two things in particular. The sine row has no factor in front, and the other two do. And the domains are three different shapes: inside an interval, the whole line, and outside an interval.

If you want a quick way to remember the domains, look at the expression under the root. A root needs a positive quantity, so a squared minus u squared forces u to be small, and u squared minus a squared forces it to be large. With no root, there is nothing to restrict.

44. Substitution and completing the square

Section

Part 5

45. Look for the square of something

Concept

Most integrands are not in a standard form yet. The clue is a constant plus or minus the square of some expression, together with that expression's derivative somewhere else in the integrand.

you seesquare ofudu supplies
1 + e²ˣeˣeˣeˣ dx
1 − x⁴x²x²2x dx
1 − (ln t)²ln tln tdt/t
4 − 9x²3x3x3 dx

If the derivative is present, up to a constant factor, the substitution goes through. If it is not, a different method is needed.

Most integrands in the exercises are not in a standard form yet, and the skill is spotting the substitution that makes them one. The clue has two parts, and you need both.

First, find a constant plus or minus the square of some expression. The table shows four typical disguises: e to the 2x is the square of e to the x, x to the fourth is the square of x squared, and so on. Second, check that the derivative of that expression appears elsewhere in the integrand, possibly up to a constant factor.

If only the first part holds, the substitution will leave stray pieces of x behind, and you will need a different method, such as the trigonometric substitutions of Calculus II. If both hold, the integral becomes one of the three standard forms in a single step.

46. The general scaling move

Concept

The most common case is a constant times x squared. Writing it as a square and substituting gives one rule for both sine and tangent forms.

\[ \int\frac{dx}{a^2+b^2x^2} = \frac{1}{ab}\tan^{-1}\left(\frac{bx}{a}\right) + C \]

\[ \int\frac{dx}{\sqrt{a^2-b^2x^2}} = \frac{1}{b}\sin^{-1}\left(\frac{bx}{a}\right) + C \]

The one over b always comes from du equal to b dx. The one over a appears only in the tangent. Check against Checkpoint 5.42: a is 5 and b is 2, and one over ab is one tenth.

Because a constant times x squared comes up so often, it is worth having one general rule instead of redoing the substitution every time. Write b squared x squared as the square of bx, substitute u equal to bx, and the two formulas on the slide follow.

The factors have two separate sources, and keeping them apart is the point. The one over b always comes from the substitution, because du is b dx. The one over a comes from the formula itself, so it appears in the tangent version and not in the sine version.

Test the rule on Checkpoint 5.42: a is 5 and b is 2, so the tangent rule gives one tenth of the inverse tangent of 2x over 5, exactly what you found. Use the rule as a check, but be able to do the substitution too.

47. Before substituting: what is u?

Step zero

\[ \int\frac{e^{x}}{1+e^{2x}}\,dx \]

Discussion prompt

Exercise 424 asks for this antiderivative. Before writing anything else, what should u be, and what two facts about the integrand make that choice work?

Write your answer before revealing it. The point of this slide is to slow down before substituting, and to name the two facts that make a substitution work.

The first fact is that e to the 2x is the square of e to the x, so the denominator is one plus a square. The second is that the derivative of e to the x is itself, so e to the x times dx is exactly du. Both are needed. If the numerator were 1 instead, the first fact would still hold but the substitution would leave a stray factor behind.

Once you have named u and checked that du is present, the rest of the problem is routine. The next slide carries it out.

48. Exercise 424: an exponential that gives an inverse tangent

Worked example

\[ \int\frac{e^{x}}{1+e^{2x}}\,dx \]

Substitute u equal to e to the x

Why: Its derivative is itself.

\[ u = e^{x}, \quad du = e^{x}\,dx, \quad e^{2x} = u^2 \]

Rewrite the integral in u

Why: Everything in x is used up.

\[ \int\frac{e^{x}\,dx}{1+e^{2x}} = \int\frac{du}{1+u^2} \]

Apply the inverse tangent

Why: Here a is 1.

\[ = \tan^{-1}u + C \]

Return to x

Why: Replace u by e to the x.

\[ = \tan^{-1}\left(e^{x}\right) + C \]

Check by differentiating

Why: The chain rule gives the derivative of e to the x on top.

\[ \frac{d}{dx}\tan^{-1}\left(e^{x}\right) = \frac{e^{x}}{1+\left(e^{x}\right)^2} = \frac{e^{x}}{1+e^{2x}} \]

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 531 — Exercise 424

With u equal to e to the x, every piece of the integrand converts: the numerator times dx becomes du, and e to the 2x becomes u squared. The integral becomes the basic inverse tangent with a equal to 1.

Returning to x gives the inverse tangent of e to the x. It is a slightly surprising answer, an angle whose tangent is an exponential, but the check by differentiating settles any doubt. The chain rule produces e to the x on top, and one plus the square of e to the x is one plus e to the 2x.

Compare Exercise 423, where the same numerator sits over the square root of one minus e to the 2t. The same substitution works, and the answer is the inverse sine of e to the t instead.

49. A fourth power under a root

Worked example

Evaluate the integral; it is on the homework list for this section.

\[ \int\frac{x}{\sqrt{1-x^4}}\,dx \]

Write the fourth power as a square

Why: x to the fourth is x squared, squared.

\[ 1 - x^4 = 1 - \left(x^2\right)^2 \]

Substitute u equal to x squared

Why: Its derivative is 2x, and x dx is already on top.

\[ u = x^2, \quad du = 2x\,dx, \quad x\,dx = \tfrac12\,du \]

Rewrite the integral in u

Why: The half moves outside.

\[ \int\frac{x\,dx}{\sqrt{1-x^4}} = \frac12\int\frac{du}{\sqrt{1-u^2}} \]

Apply the inverse sine and return to x

Why: Here a is 1.

\[ = \frac12\sin^{-1}u + C = \frac12\sin^{-1}\left(x^2\right) + C \]

Check by differentiating

Why: The chain rule gives 2x, which the half cancels.

\[ \frac{d}{dx}\left[\frac12\sin^{-1}\left(x^2\right)\right] = \frac12\cdot\frac{2x}{\sqrt{1-x^4}} = \frac{x}{\sqrt{1-x^4}} \]

The fourth power is a square in disguise: x to the fourth is the square of x squared. So one minus x to the fourth is one minus u squared, with u equal to x squared.

The substitution works because the rest of the integrand supplies du. The derivative of x squared is 2x, and there is an x on top, so x dx is half of du. The half moves outside, and the inverse sine form with a equal to 1 finishes the problem.

Notice what would happen without the x on top. The integral of one over the root of one minus x to the fourth has no elementary antiderivative at all. The x in the numerator is not a detail; it is what makes the substitution possible. The check confirms the answer by differentiating.

50. Which form does it lead to?

Sorting

Sort into buckets

Sort each integrand by what its antiderivative contains.

sin⁻¹
1/√(25 − x²); eᵗ/√(1 − e²ᵗ)
tan⁻¹
1/(7 + x²); eˣ/(1 + e²ˣ)
sec⁻¹
1/(x√(x² − 16))
ln
x/(7 + x²); 2x/(x² + 1)
sin
A constant minus a square under a root: 25 − x², and 1 − (eᵗ)² with du = eᵗ dt (Exercise 423).
tan
A constant plus a square with nothing else on top that is the bottom's derivative: 7 + x², and 1 + (eˣ)² with du = eˣ dx.
sec
u times the root of u² minus a constant: x√(x² − 16), with a = 4.
ln
The top is a constant times the derivative of the bottom: x is half of 2x, so these are logarithms.

Sort each integrand before checking. For each one, ask two questions: is there a square hidden in the denominator, and what is on top?

Two of the items are logarithms, and they are placed here on purpose. The integrand x over 7 plus x squared has a sum of squares in the bottom, just like an arctangent, but the x on top is a constant times the derivative of the bottom. That makes it a logarithm. The same goes for 2x over x squared plus 1.

The exponential items are the disguised ones. One minus e to the 2t under a root is a constant minus a square, the sine form, and one plus e to the 2x is a sum of squares, the tangent form. In both, e to the x is already in the numerator to supply du.

51. Completing the square finds a and u

Concept

A quadratic with an x term is not yet a sum or difference of squares. Completing the square makes it one, and the substitution u equal to x plus a constant costs nothing, because du equals dx.

\[ x^2 + 4x + 8 = \left(x^2 + 4x + 4\right) + 4 = (x+2)^2 + 2^2 \]

Figure (svg): Two panels. Left: the parabola y equals x squared plus 4x plus 8, with its lowest point at minus 2, 4 marked. Right: the curve one over x squared plus 4x plus 8, a bump centred at x equals minus 2 with height one quarter, shaded from minus 2 to 0.

Completing the square finds the parabola's vertex. The integrand is the ordinary arctangent bump shifted left by 2 and scaled: a is 2 because the lowest value is 2 squared.

Half the x coefficient is 2, so add and subtract 2 squared. What is left over, 4, is a squared, and x plus 2 is u.

When a quadratic has an x term, it is not yet a sum or difference of squares. Completing the square makes it one. Take half the coefficient of x, square it, and add and subtract that number.

For x squared plus 4x plus 8, half of 4 is 2, so you add and subtract 4. The first three terms become x plus 2, squared, and the leftover constant is 4, which is 2 squared. So u is x plus 2 and a is 2.

The picture shows what completing the square means geometrically. On the left, it finds the vertex of the parabola, at minus 2, 4. On the right, the integrand is the ordinary arctangent bump, just moved left by 2 and rescaled. Moving it does not change the method, because u equal to x plus 2 has du equal to dx: the shift costs nothing.

52. A quadratic denominator

Worked example

Evaluate the definite integral.

\[ \int_{-2}^{0}\frac{dx}{x^2+4x+8} \]

Complete the square

Why: Half of 4 is 2; its square is 4.

\[ x^2+4x+8 = (x+2)^2 + 4 \]

Substitute u equal to x plus 2

Why: du is dx; the limits minus 2 and 0 become 0 and 2.

\[ \int_{-2}^{0}\frac{dx}{(x+2)^2+4} = \int_{0}^{2}\frac{du}{u^2+4} \]

Apply the inverse tangent with a equal to 2

Why: One half in front, u over 2 inside.

\[ = \frac12\tan^{-1}\frac{u}{2}\Big|_0^{2} \]

Evaluate at both limits

Why: The inverse tangent of 1 is pi over four.

\[ = \frac12\left(\tan^{-1}1 - \tan^{-1}0\right) = \frac12\cdot\frac{\pi}{4} = \frac{\pi}{8} \]

Check by differentiating the antiderivative

Why: In x, the antiderivative is one half of the inverse tangent of x plus 2 over 2; a numerical integral also gives 0.3927.

\[ \frac{d}{dx}\left[\frac12\tan^{-1}\frac{x+2}{2}\right] = \frac14\cdot\frac{1}{1+\frac{(x+2)^2}{4}} = \frac{1}{x^2+4x+8} \]

Follow the method from the previous slide. Completing the square gives x plus 2 squared plus 4, so u is x plus 2 and a is 2. Because du equals dx, no factor comes from the substitution; the only factor is the formula's one half.

With a definite integral you can change the limits along with the variable, which saves returning to x. At x equal to minus 2, u is 0; at x equal to 0, u is 2. The answer is one half of the inverse tangent of 1, which is pi over eight.

The check differentiates the antiderivative written in x. It is worth doing once, to see that the one quarter from the two halves combines with one plus x plus 2 squared over 4 to rebuild x squared plus 4x plus 8. The shaded region on the right of the earlier figure is exactly this area.

53. A quadratic under a root, negative leading coefficient

Worked example

Evaluate the definite integral.

\[ \int_{2}^{3}\frac{dx}{\sqrt{4x-x^2}} \]

Factor out minus 1 before completing the square

Why: Completing the square needs x squared with coefficient plus 1.

\[ 4x - x^2 = -\left(x^2 - 4x\right) \]

Complete the square inside

Why: Half of minus 4 is minus 2; its square is 4.

\[ -\left(x^2 - 4x + 4 - 4\right) = -\left((x-2)^2 - 4\right) = 4 - (x-2)^2 \]

Figure (svg): The upper half of a circle of radius 2 centred at the point 2, 0, which is the graph of y equals the square root of 4x minus x squared for x from 0 to 4. A radius of length 2 is drawn from the centre.

Under the root, 4x minus x squared equals 4 minus the square of x minus 2: a circle of radius 2 centred at 2. The root is real only between 0 and 4.

Substitute u equal to x minus 2

Why: du is dx; the limits 2 and 3 become 0 and 1.

\[ \int_{2}^{3}\frac{dx}{\sqrt{4-(x-2)^2}} = \int_0^{1}\frac{du}{\sqrt{4-u^2}} \]

Apply the inverse sine with a equal to 2

Why: No factor in front.

\[ = \sin^{-1}\frac{u}{2}\Big|_0^{1} = \sin^{-1}\tfrac12 - 0 = \frac{\pi}{6} \]

Check the domain and the value

Why: The interval from 2 to 3 lies inside the semicircle's base from 0 to 4, and numerical integration gives 0.5236, which is pi over six.

\[ 0 < 2 < 3 < 4, \qquad \frac{\pi}{6} \approx 0.5236 \]

Here the quadratic is 4x minus x squared, and the x squared has a negative coefficient. Completing the square only works cleanly with x squared first and coefficient plus 1, so factor out minus 1 before you start. This is the step most often skipped, and skipping it gives a wrong sign on the square.

Inside the bracket, complete the square as usual, then distribute the minus sign back. The result is 4 minus x minus 2 squared: a constant minus a square, the sine form, with a equal to 2 and u equal to x minus 2.

The picture shows what the quadratic really is. The graph of the root of 4x minus x squared is a semicircle of radius 2 centred at 2. That tells you the domain at a glance: x from 0 to 4. The interval from 2 to 3 lies inside it, so the integral exists, and the answer is the inverse sine of one half, which is pi over six.

54. Put the completing-the-square steps in order

Ranking

\[ \int\frac{dx}{x^2+6x+13} \]

Put in order

Order the steps that evaluate this integral.

  1. Complete the square: (x + 3)² + 4
  2. Substitute u = x + 3, so du = dx
  3. Read off a = 2 from the constant 4
  4. Apply (1/a) tan⁻¹(u/a) + C
  5. Return to x: (1/2) tan⁻¹((x + 3)/2) + C

Why: Half of 6 is 3 and 13 minus 9 is 4, so the denominator is (x + 3) squared plus 2 squared. The shift costs nothing because du equals dx, and the arctangent formula with a equal to 2 finishes it.

Drag the steps into order before checking. The order is natural once you see that each step supplies what the next one needs.

Completing the square comes first, because until you have done it you cannot see what u and a are. Here half of 6 is 3, and 13 minus 9 leaves 4. The substitution then names u, reading off a comes from the leftover constant, the formula does the integration, and the last step translates back to x.

Differentiate the final answer as a check: one half times one half, over one plus x plus 3 squared over 4, gives one over x plus 3 squared plus 4, which expands back to the original denominator.

55. Telling them apart from logarithms

Section

Part 6

56. The numerator decides

Concept

The same denominator can lead to completely different antiderivatives. What decides is the numerator.

\[ \int\frac{dx}{1+x^2} = \tan^{-1}x + C, \qquad \int\frac{x\,dx}{1+x^2} = \frac12\ln\left(1+x^2\right) + C \]

Figure (svg): Two panels. Left: the integrands x over one plus x squared and one over one plus x squared for x from 0 to 8; the first decays slowly, the second quickly. Right: their antiderivatives from 0, one half the natural log of one plus x squared, which keeps climbing past 3, and inverse tangent of x, which levels off below a dashed line at pi over two.

An extra x on top makes the integrand decay like 1/x instead of 1/x², and the antiderivative switches from a bounded arctangent to an unbounded logarithm.

With x on top, the numerator is half the derivative of the denominator, so u equal to one plus x squared turns it into a logarithm. With a constant on top there is no such u, and the arctangent is the answer.

This is the most important distinction in the lesson. One over one plus x squared and x over one plus x squared share a denominator, and they look nearly identical, but their antiderivatives are completely different kinds of function.

The deciding question is whether the numerator is a constant times the derivative of the denominator. The derivative of one plus x squared is 2x, and x is half of that, so the substitution u equal to one plus x squared turns the second integral into half a logarithm. For a constant numerator there is no such u, and the arctangent is the answer.

The picture shows how different the results are. The x on top makes the integrand decay much more slowly, and its antiderivative, the logarithm, climbs forever. The arctangent, by contrast, levels off below pi over two. Same denominator, different numerator, entirely different behaviour.

57. Trap: an inverse tangent for x over one plus x squared

Trap

The trap

Seeing one plus x squared and reaching for the arctangent:

\[ \int\frac{x}{1+x^2}\,dx \]

\[ = x\tan^{-1}x + C \quad ? \]

Wrong. A factor of x cannot be carried outside an integral.

The fix

The x is half the derivative of the denominator, so this is a logarithm. Differentiate the claimed answer and you get the arctangent plus x over one plus x squared: not the integrand.

\[ = \frac12\ln\left(1+x^2\right) + C \]

This mistake comes from recognising the denominator and not looking at the numerator. It usually comes with a second error: treating the x on top as a constant and carrying it outside the integral, which is never allowed.

Differentiate the claimed answer and you can see it fail. The product rule gives the inverse tangent of x, plus x over one plus x squared. That extra arctangent term should not be there.

The correct approach sees that x is half the derivative of the denominator. The substitution w equal to one plus x squared gives half the natural logarithm. Whenever the numerator is a multiple of the denominator's derivative, think logarithm first.

58. Splitting a mixed numerator

Worked example

Evaluate the definite integral.

\[ \int_0^{1}\frac{x+3}{x^2+1}\,dx \]

Split the fraction into two

Why: Each piece has a numerator of one kind only.

\[ \frac{x+3}{x^2+1} = \frac{x}{x^2+1} + \frac{3}{x^2+1} \]

Integrate the first piece as a logarithm

Why: Let w be x squared plus 1, so dw is 2x dx.

\[ \int\frac{x\,dx}{x^2+1} = \frac12\ln\left(x^2+1\right) \]

Integrate the second piece as an inverse tangent

Why: Here a is 1.

\[ \int\frac{3\,dx}{x^2+1} = 3\tan^{-1}x \]

Evaluate from 0 to 1

Why: Both pieces vanish at 0.

\[ \left[\frac12\ln\left(x^2+1\right) + 3\tan^{-1}x\right]_0^1 = \frac12\ln 2 + \frac{3\pi}{4} \]

Check numerically

Why: The two pieces are 0.3466 and 2.3562; a numerical integral of the original gives the same total.

\[ 0.3466 + 2.3562 = 2.7028 \]

When the numerator has both an x term and a constant, neither method applies to the whole fraction. Split it into two fractions over the same denominator, one with only the x term on top and one with only the constant.

Each piece is now something you know. The x piece is half a logarithm, by the substitution w equal to x squared plus 1. The constant piece is three times the basic inverse tangent. The answer is the sum of the two.

Evaluating from 0 to 1, both parts vanish at 0, leaving half of ln 2 plus three pi over four, about 2.703. The check lists the two pieces separately so you can see that the arctangent part carries most of the value here, because the constant 3 is bigger than the x on this interval.

59. When the top is as big as the bottom, divide first

Concept

If the numerator's degree is at least the denominator's, none of the three forms applies yet. Divide first, exactly as with a fraction bigger than one.

\[ \frac{x^2}{x^2+1} = \frac{(x^2+1) - 1}{x^2+1} \]

\[ \frac{x^2}{x^2+1} = 1 - \frac{1}{x^2+1} \]

Adding and subtracting 1 in the numerator is long division in one line. What is left over is a standard form.

None of the three forms, and none of the logarithm tricks, can be applied directly when the numerator's degree is as large as the denominator's. You first have to split off the whole-number part, exactly as you would write seven thirds as 2 plus one third.

For x squared over x squared plus 1, the neat way to divide is to add and subtract 1 in the numerator. That makes the numerator the denominator minus 1, and the fraction splits into 1 minus one over x squared plus 1.

What is left over, one over x squared plus 1, is a standard form. In general, after dividing, the remainder is always of lower degree than the denominator, and one of the earlier methods then applies.

60. An improper quotient

Worked example

Evaluate the definite integral.

\[ \int_0^{1}\frac{x^2}{x^2+1}\,dx \]

Divide

Why: Add and subtract 1 in the numerator.

\[ \frac{x^2}{x^2+1} = 1 - \frac{1}{x^2+1} \]

Integrate each term

Why: The second is the basic inverse tangent.

\[ \int\left(1 - \frac{1}{x^2+1}\right)dx = x - \tan^{-1}x + C \]

Evaluate from 0 to 1

Why: The inverse tangent of 1 is pi over four.

\[ \left[x - \tan^{-1}x\right]_0^1 = 1 - \frac{\pi}{4} \]

Check the size

Why: The integrand lies between 0 and one half on this interval, so the area must be between 0 and one half; a numerical integral gives 0.2146.

\[ 1 - \frac{\pi}{4} \approx 0.2146, \quad 0 < 0.2146 < 0.5 \]

After dividing, the integrand is 1 minus one over x squared plus 1. The 1 integrates to x and the second term is the basic inverse tangent, so the antiderivative is x minus the inverse tangent of x.

Evaluating from 0 to 1 gives 1 minus pi over four, about 0.215. It is a small number, and the check explains why: on this interval x squared over x squared plus 1 starts at 0 and only reaches one half at x equal to 1, so the area must be less than one half.

If you tried the inverse tangent directly on the original integrand without dividing, you would get a function whose derivative is one over one plus x squared, which is a completely different integrand. Dividing first is not optional.

61. Same denominator, four numerators

Matching

Match the pairs

  • a. ∫ 1/(1 + x²) dx
  • b. ∫ x/(1 + x²) dx
  • c. ∫ x²/(1 + x²) dx
  • d. ∫ (2x + 1)/(1 + x²) dx
  • w. tan⁻¹ x + C
  • x. ½ ln(1 + x²) + C
  • y. x − tan⁻¹ x + C
  • z. ln(1 + x²) + tan⁻¹ x + C

Why: A constant on top gives the arctangent; x on top gives half a logarithm; x squared on top must be divided first, leaving x minus the arctangent; and 2x plus 1 splits into a full logarithm plus an arctangent.

Make all four matches before checking. The denominator is the same in every row, so the numerator alone decides the answer, which is the whole point of this part of the lesson.

A constant on top gives the arctangent. An x on top gives half a logarithm, because x is half the derivative of the bottom. An x squared on top has to be divided first, which leaves x minus the arctangent. And 2x plus 1 splits: 2x is exactly the derivative of the bottom, giving a full logarithm, and the 1 gives an arctangent.

If you can do all four of these without hesitating, you have the main skill of this section.

62. Putting it together

Section

Part 7

63. Pattern: an integrand with a quadratic in the bottom

Pattern

Figure (svg): A flow diagram. First box: is the top's degree at least the bottom's? If so, divide. Second box: is the top a multiple of the bottom's derivative? If so, a logarithm. Third box: complete the square and find u and a. From the third box three arrows lead to the three forms: a squared plus u squared gives inverse tangent, the root of a squared minus u squared gives inverse sine, u times the root of u squared minus a squared gives inverse secant.

Cheap questions first: divide if the top is too big, take a logarithm if the top is the bottom's derivative, and only then match one of the three forms.
  1. Divide first if the top's degree is at least the bottom's.
  2. Look for a logarithm: if the top is a constant times the bottom's derivative, let u be the bottom.
  3. Split a mixed numerator into a logarithm part and a constant part.
  4. Complete the square if there is a linear term; read off u and a.
  5. Match the form: sum of squares, tangent; constant minus square under a root, sine; u times root of square minus constant, secant.
  6. Check the domain for a definite integral, then differentiate your answer.

This is the order to work in when you meet a fraction with a quadratic in the denominator. It runs from the cheapest checks to the most involved, and each step can end the problem early.

Dividing comes first, because none of the other methods apply while the top is too big. Then look for a logarithm, since that needs only one substitution. Then split any mixed numerator, and only then complete the square and match one of the three forms.

The last step has two parts that are easy to forget. For a definite integral, check that the interval stays inside the form's domain. For any integral, differentiate your answer. Both take seconds and both catch errors that are otherwise invisible.

64. Check: the inverse sine form

Check

Check your understanding

Which is an antiderivative of 1/√(16 − x²)?

  • A. sin⁻¹(x/4) + C (correct)
  • B. (1/4) sin⁻¹(x/4) + C
  • C. (1/4) tan⁻¹(x/4) + C
  • D. sin⁻¹(4x) + C

Answer: A

Why: Sixteen is 4 squared, so a is 4 and u is x. The inverse sine formula has no factor in front, so the answer is the inverse sine of x over 4, plus C. Differentiating gives one quarter over the root of one minus x squared over 16, which is the integrand.

Why B tempts people
The one over a belongs to the tangent and secant formulas only; differentiating this gives a quarter of the integrand.
Why C tempts people
There is a square root and a difference of squares, so this is the sine form, not the tangent form.
Why D tempts people
The inner fraction is u over a, x divided by 4, not x multiplied by 4.

Identify a and u first, then decide which form you have. The square root around a constant minus a square tells you it is the sine form.

The correct answer has no factor in front, because the inverse sine formula has none and the substitution is trivial here. Each wrong option matches one slip: borrowing the tangent's factor, using the wrong form, or multiplying by a instead of dividing.

65. Check: completing the square

Check

Check your understanding

Which is an antiderivative of 1/(x² + 2x + 5)?

  • A. tan⁻¹(x + 1) + C
  • B. (1/2) tan⁻¹((x + 1)/2) + C (correct)
  • C. (1/2) ln(x² + 2x + 5) + C
  • D. (1/5) tan⁻¹(x/5) + C

Answer: B

Why: Completing the square gives (x + 1) squared plus 4, so u is x plus 1 and a is 2. The arctangent formula gives one half of the inverse tangent of (x + 1) over 2. Differentiating returns one quarter over one plus (x + 1) squared over 4, which is the integrand.

Why A tempts people
That would need the constant left over to be 1; here it is 4, so a is 2 and both halves are missing.
Why C tempts people
The top would have to be a multiple of 2x + 2 for a logarithm; the top is the constant 1.
Why D tempts people
The linear term 2x cannot be ignored, and a is the square root of the leftover constant, not the constant itself.

Complete the square before looking at the options. Half of 2 is 1, and 5 minus 1 leaves 4, so the denominator is x plus 1 squared plus 2 squared.

That gives u equal to x plus 1 and a equal to 2, and the tangent formula puts one half in front and x plus 1 over 2 inside. The logarithm option fails because the numerator is not a multiple of the denominator's derivative.

66. Check: a definite inverse tangent

Check

Check your understanding

What is the value of the integral of 1/(1 + x²) from 0 to 1?

  • A. π/4 (correct)
  • B. π/2
  • C. (1/2) ln 2
  • D. 1

Answer: A

Why: The antiderivative is the inverse tangent, and the inverse tangent of 1 minus the inverse tangent of 0 is pi over four minus 0, about 0.785. That fits the picture: the integrand falls from 1 to one half across a width of 1.

Why B tempts people
Pi over two is the limit of the inverse tangent as x grows without bound, the area from 0 all the way out, not from 0 to 1.
Why C tempts people
One half of ln 2 is the integral of x over one plus x squared, which has an x on top.
Why D tempts people
The integrand is below 1 for every x after 0, so the area over a width of 1 must be less than 1.

This is the simplest possible definite arctangent, and a good one to know by heart. The inverse tangent of 1 is pi over four and of 0 is zero, so the area is pi over four, about 0.785.

Sanity-check it against the curve: the integrand falls from 1 at x equal to 0 to one half at x equal to 1, so the area over a width of 1 must lie between one half and 1. Pi over four sits nicely in that range. Pi over two would be the area from 0 all the way to infinity.

67. Explain the one over a

Explain it to yourself

Discussion prompt

In two or three sentences: why does the arctangent formula have a factor of one over a in front while the inverse sine formula has none? Use the substitution u = a w.

Write your explanation before revealing. Being able to explain where the factor comes from means you will never need to memorise which formula has it.

The substitution u equal to a times w always contributes one factor of a, through du. Under a square root, a squared minus u squared becomes a times the root of one minus w squared: one factor of a, which cancels the one from du exactly. Without a root, a squared plus u squared becomes a squared times one plus w squared: two factors of a, and only one of them cancels.

That leftover a is the one over a in the tangent formula. The secant formula also has a leftover factor, because of the extra u outside its root, which brings in another a.

68. Exit ticket

Exit ticket

\[ \int_1^{\sqrt e}\frac{dt}{t\sqrt{1-(\ln t)^2}} \]

Discussion prompt

Exercise 425 asks for this antiderivative. Choose a substitution, name the form, and evaluate the definite integral exactly.

OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions §5.7, p. 531 — Exercise 425

This combines the main skills of the lesson in a single problem: recognise what is being squared, check that its derivative is present, name the form, and evaluate.

The thing being squared is the natural logarithm of t, and its derivative, one over t, is already in the integrand. So u is ln t, and du is dt over t. Changing the limits, t equal to 1 gives u equal to 0 and t equal to the root of e gives u equal to one half.

The integral becomes the inverse sine form with a equal to 1, and the answer is the inverse sine of one half, pi over six. Notice also that the new interval, from 0 to one half, lies inside the domain of the sine form, so the integral exists.

69. Recap

Recap

formantiderivativevalid for
1/√(a² − u²)sin⁻¹(u/a) + C, no factor−a < u < a
1/(a² + u²)(1/a) tan⁻¹(u/a) + Cevery u
1/(u√(u² − a²))(1/a) sec⁻¹(|u|/a) + C|u| > a

Next, Chapter 6 puts integrals to work on areas, volumes and physical quantities.

Stewart, Calculus: Early Transcendentals 8e, §7.3 Trigonometric Substitution §7.3, pp. 486-492 — the same integrals, reached by trigonometric substitution in Stewart

Three formulas, each a derivative from Section 3.7 read backwards. The sine form has no factor in front and works only strictly inside an interval. The tangent form has one over a and works everywhere. The secant form has one over a and works only strictly outside an interval.

Almost every problem needs some preparation before a formula applies: a substitution that makes the square visible, completing the square when there is a linear term, dividing when the top is too big, or splitting a mixed numerator. And always check whether the numerator makes the problem a logarithm instead.

Two habits make this section safe: differentiate every answer, and check the domain on every definite integral. With this, Chapter 5 is complete, and Chapter 6 turns integrals into areas between curves, volumes and physical quantities.

Sources

  1. OpenStax Calculus Volume 1, §5.7 Integrals Resulting in Inverse Trigonometric Functions — Strang & Herman, OpenStax / Rice University, CC BY-NC-SA 4.0, pp. 526-537
  2. Stewart, Calculus: Early Transcendentals 8e, §7.3 Trigonometric Substitution — James Stewart, Cengage Learning, 2016, pp. 486-492
  3. Stewart, Calculus: Early Transcendentals 8e, §3.5 Implicit Differentiation — James Stewart, Cengage Learning, 2016, pp. 208-217

Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.

Book on Wyzant · Text (657) 465-8108