5.6 Integrals Involving Exponential and Logarithmic Functions

Integrating the natural exponential and its substitutions, general exponentials with other bases, the reciprocal with its absolute value, the logarithmic form for a quotient whose numerator is the denominator's derivative, and applications to growth and decay.

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 Involving Exponential and Logarithmic Functions

Title

Calculus I · Section 5.6

The functions of growth, integrated

2. What this lesson gives you

Objectives

Section 3.9 differentiated the exponential and logarithm functions, and Section 5.5 gave you substitution. This lesson puts the two together, running those derivative rules backwards.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 516-525 — learning objectives 5.6.1 and 5.6.2

Exponentials and logarithms are the functions of growth and decay: populations, bank balances, radioactive samples, drugs leaving the bloodstream. In Section 3.9 you learned how fast they change. This lesson asks the reverse question: given how fast something changes, how much has built up?

There is almost nothing new to memorise. Every formula here is a derivative you already know, read backwards, and the technique is the substitution you learned in Section 5.5. What is new is recognising the shapes: an exponential whose exponent has its derivative sitting beside it, and a fraction whose top is the derivative of its bottom.

Two warnings run through the whole lesson. The exponent of e is not a power, so the power rule never applies to it. And one over x needs an absolute value in its antiderivative, which still does not let you integrate across zero.

3. Read the derivative table backwards

Warm-up

Discussion prompt

Write the derivative of each function from Section 3.9, then turn each line into an integral formula.

Write your three derivatives before revealing. If any of them is shaky, that is the thing to fix first, because each one becomes an integration rule in the next few minutes.

The first line is the cleanest fact in calculus: the exponential is its own derivative, so it is its own antiderivative too. The second tells you that differentiating a to the x produces a factor of ln a, which means integrating it will have to divide by ln a. The third says the logarithm undoes one over x.

Notice the gap in that third line. The logarithm is only defined for positive x, but one over x is perfectly happy at x equal to minus 2. Keep that gap in mind; Part 5 closes it with an absolute value.

4. The natural exponential

Section

Part 1

5. The exponential is its own antiderivative

Concept

The exponential function is its own derivative. Read that sentence backwards and it becomes an integral.

\[ \frac{d}{dx}e^{x} = e^{x} \quad\Longrightarrow\quad \int e^{x}\,dx = e^{x} + C \]

Figure (svg): The curve y equals e to the x from x equals minus 3 to 2, with the region under it shaded from 0 to 1.5; a dashed vertical segment marks the height e to the 1.5, about 4.48, at the right edge of the shading.

The shaded area is exactly the height at the right edge minus 1: e to the 1.5 is 4.48, and the area is 3.48. The exponential accumulates area at the same rate as its own height.

\[ \int_0^{b} e^{x}\,dx = e^{b} - e^{0} = e^{b} - 1 \]

The area under the curve from 0 to b is the height at b, less one. No other function does that.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 516 — rule (5.21)

Because the exponential is its own derivative, the question which function has derivative e to the x has the answer e to the x itself, plus any constant. That is the first line of rule 5.21.

Look at the figure. The shaded area from 0 to 1.5 is 3.48, and the height of the curve at 1.5 is 4.48. The area is exactly the height minus one, and that is no coincidence: the Fundamental Theorem says the area is the antiderivative at 1.5 minus the antiderivative at 0, and the antiderivative is the curve itself.

Put differently, the exponential piles up area at a rate equal to its own height. Everything in the first part of the lesson is this one fact plus the chain rule.

6. A linear exponent: divide by the slope

Concept

With kx in the exponent, substitute for the exponent. The derivative of kx is the constant k, and a constant can always be moved outside.

\[ u = kx, \qquad du = k\,dx, \qquad dx = \frac{du}{k} \]

\[ \int e^{kx}\,dx = \frac1k\int e^{u}\,du = \frac1k e^{u} + C \]

\[ \int e^{kx}\,dx = \frac{1}{k}e^{kx} + C \qquad (k \ne 0) \]

Differentiating gives back a factor k from the chain rule; the one over k is there to cancel it.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 516 — rule (5.21) with substitution

The moment the exponent is anything other than a bare x, you need substitution. With k times x in the exponent, let u be the exponent. Its derivative is the constant k, so dx is du over k, and the one over k slides out in front of the integral.

The result is worth knowing by heart: the integral of e to the kx is one over k times e to the kx. You can see why it must be so by differentiating: the chain rule multiplies by k, and the one over k is there precisely to cancel it.

The condition that k is not zero matters only formally. With k equal to zero the integrand is the constant 1, whose integral is x, not anything with a one over zero in it.

7. Squeeze the exponent, shrink the area

Picture it

Figure (svg): Two curves, y equals e to the x and y equals e to the 2x, for x from minus 4 to 1. The region under each from far left up to x equals 0 is shaded: the wide region under e to the x has area 1, the narrow region under e to the 2x has area one half.

Doubling the exponent squeezes the graph sideways by a factor of 2, so every region under it has half the width and half the area. That half is the 1/k in the formula.

\[ \int_{-\infty}^{0} e^{x}\,dx = 1, \qquad \int_{-\infty}^{0} e^{2x}\,dx = \frac12 \]

Doubling the exponent makes the graph change twice as fast, which squeezes it sideways by a factor of two. Half the width means half the area, and that half is the one over k.

This is the one-over-k rule as a picture. The orange curve, e to the 2x, reaches every height at half the x-distance from the origin that the blue curve e to the x needs. So it is the blue curve squeezed sideways by a factor of two.

Squeezing a region sideways by two halves its area. The whole region under e to the x to the left of the origin has area 1, so the region under e to the 2x has area one half. That half is exactly the one over k with k equal to 2.

Hold on to the picture for the general case. A bigger k means a steeper, narrower curve and a smaller area; a small k means a gentle, wide curve and a large area. That will matter when you meet decay rates in Part 3.

8. Example 5.37: e to the minus x

Worked example

Find the antiderivative of the exponential function below.

\[ \int e^{-x}\,dx \]

Let u be the exponent

Why: The inside of the exponential is minus x.

\[ u = -x, \qquad du = -dx \]

Solve for dx

Why: Multiply the differential equation by minus one.

\[ dx = -du \]

Rewrite the integral in u

Why: The minus sign comes outside.

\[ \int e^{-x}\,dx = -\int e^{u}\,du \]

Integrate

Why: The exponential is its own antiderivative.

\[ -\int e^{u}\,du = -e^{u} + C \]

Substitute back

Why: Put minus x back in place of u.

\[ \int e^{-x}\,dx = -e^{-x} + C \]

Check by differentiating

Why: The chain rule brings down a factor of minus one, which cancels the leading minus sign.

\[ \frac{d}{dx}\left(-e^{-x}\right) = -e^{-x}\cdot(-1) = e^{-x} \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 516 — Example 5.37

This is the linear-exponent rule with k equal to minus one, done slowly so you can see every sign. Let u be minus x. Then du is minus dx, and so dx is minus du.

Replacing dx by minus du brings a minus sign out in front of the integral. The integral of e to the u is e to the u, so the answer is minus e to the minus x plus C.

Always check an exponential antiderivative by differentiating it, because sign errors are the commonest mistake here. The chain rule brings down a minus one from the exponent, which multiplies the minus sign already in front, and the two minuses give back e to the minus x.

9. A non-linear exponent needs its derivative present

Concept

Once the exponent is not linear, its derivative is no longer a constant, and substitution only works if that derivative is already sitting in the integrand.

\[ \int 2x\,e^{x^2}\,dx: \quad u = x^2,\; du = 2x\,dx \quad\Longrightarrow\quad \int e^{u}\,du = e^{x^2} + C \]

\[ \int e^{x^2}\,dx: \quad du = 2x\,dx \text{ but there is no } x \text{ to use} \]

The second integral has no antiderivative that can be written with the usual functions at all. A missing constant can be supplied; a missing variable cannot.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 516-517 — the remark after Checkpoint 5.31

With a linear exponent, the derivative of the exponent is a constant, and you can always make up a missing constant by multiplying and dividing. With x squared in the exponent, the derivative is 2x, which is not a constant.

So the question becomes: is that 2x, or some constant multiple of it, already in the integrand? In the first line it is, and the substitution goes through in one step. In the second line it is not, and there is nothing you can legally do to create an x from nowhere.

In fact the integral of e to the x squared has no antiderivative that can be written with the ordinary functions at all. It is the famous bell-curve integral, and statisticians tabulate it numerically for that reason. So the test for any e to a power is: find the derivative of the exponent, and look for it in the integrand.

10. Trap: the power rule on the exponent

Trap

The trap

The power rule is a habit, and here it fires on the wrong thing:

\[ \int e^{3x}\,dx = \frac{e^{3x+1}}{3x+1} + C \]

Wrong. The exponent of e is not a power of x.

The fix

The power rule integrates a variable raised to a fixed number. Here the variable is up in the exponent. Substitute for the exponent instead:

\[ \int e^{3x}\,dx = \frac13 e^{3x} + C \]

Differentiate to check: the chain rule gives three times the exponential, over three.

The book singles this out as the most common error with exponentials, and it is easy to see why. You have used the power rule hundreds of times: add one to the exponent, divide by the new exponent. Here there is an exponent in sight, so the habit fires.

But the power rule is for a variable raised to a fixed number, like x to the fifth. Here a fixed number, e, is raised to a variable. These are entirely different functions and they integrate differently. The correct move is to substitute for the exponent, which gives one third of e to the 3x.

If you are ever unsure, differentiate your answer. The wrong answer would need the quotient rule and produces a mess that looks nothing like e to the 3x.

11. Checkpoint 5.31: a cubic in the exponent

Fill the middle

\[ \int x^2 e^{-2x^3}\,dx \]

Fill in the blanks

\int x^2 e^-\frac{1}{6}\,dx = ___\,e^___ + C

Why: Let u be minus two x cubed, so du is minus six x squared dx, and x squared dx is du over minus six. The integral becomes minus one sixth of the integral of e to the u. Differentiating minus one sixth of the answer brings down minus six x squared, which cancels to x squared.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 516 — Checkpoint 5.31

Fill in the constant before checking. The shape is the same as Example 5.39 on the way: the exponent is a cubic, and its derivative is a multiple of x squared, which is exactly the other factor.

Let u be minus two x cubed. Then du is minus six x squared dx, so x squared dx is du divided by minus six. The integral becomes minus one sixth of the integral of e to the u, which is minus one sixth of e to the minus two x cubed.

The sign is where most people slip. The exponent has a minus sign, so its derivative has one too, and that minus comes out in front of the answer. Differentiate to confirm: minus one sixth times minus six x squared is x squared.

12. Example 5.38: a square root of 1 plus e to the x

Worked example

Find the antiderivative.

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

Write the root as a power

Why: A rational exponent makes the power rule available later.

\[ \int e^{x}\sqrt{1 + e^{x}}\,dx = \int e^{x}\left(1 + e^{x}\right)^{1/2}dx \]

Let u be the inside of the root

Why: Its derivative, e to the x, is the other factor.

\[ u = 1 + e^{x}, \qquad du = e^{x}\,dx \]

Rewrite in u

Why: Both factors are used up.

\[ \int e^{x}\left(1 + e^{x}\right)^{1/2}dx = \int u^{1/2}\,du \]

Apply the power rule

Why: Now the variable really is raised to a fixed power.

\[ \int u^{1/2}\,du = \frac{u^{3/2}}{3/2} + C = \frac23 u^{3/2} + C \]

Substitute back

Why: Replace u by 1 plus e to the x.

\[ \int e^{x}\sqrt{1 + e^{x}}\,dx = \frac23\left(1 + e^{x}\right)^{3/2} + C \]

Figure (svg): The curve y equals e to the x times the square root of 1 plus e to the x, for x from minus 4 to 1.2, rising from almost zero; the region under it from minus 2 to 0 is shaded.

The antiderivative gives the shaded area as two thirds of the difference of (1 + eˣ) to the three halves at 0 and at minus 2, which is 1.079. A numerical integration gives the same four digits.

Check by differentiating

Why: The power rule gives the root, and the chain rule the e to the x.

\[ \frac{d}{dx}\left[\frac23\left(1+e^{x}\right)^{3/2}\right] = \left(1+e^{x}\right)^{1/2}e^{x} \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 516-517 — Example 5.38 and Figure 5.37

Here the exponential is not the outer function; the square root is. So the substitution is not for the exponent but for the inside of the root. Ask yourself which piece has its derivative sitting beside it: the derivative of 1 plus e to the x is e to the x, and there it is, multiplying the root.

Once u is 1 plus e to the x, both factors are absorbed and you are left with the integral of the square root of u, a plain power-rule integral. Two thirds of u to the three halves, then put 1 plus e to the x back in.

The figure is the book's Figure 5.37 with a strip shaded from minus 2 to 0. Using the antiderivative, that area is 1.079, and a numerical integration agrees to four digits. The derivative check confirms the formula; the area check confirms you can use it.

13. Checkpoint 5.32: choose u before you start

Step zero

\[ \int e^{x}\left(3e^{x} - 2\right)^2 dx \]

Discussion prompt

Before integrating: what should u be, what is du, and what constant has to be fixed? Then finish the integral.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 517 — Checkpoint 5.32

Answer the three questions before you integrate anything. Choosing u is the whole problem; the integration afterwards is routine.

The inside of the square is 3e to the x minus 2, and its derivative is 3e to the x, which is three times the factor outside. So u is the inside, and e to the x dx is one third of du. The integral becomes one third of the integral of u squared, which is u cubed over nine.

You could expand the square instead and integrate three exponentials separately, and you would get the same answer up to a constant. But notice how much more work that is. Spotting the substitution first is almost always faster.

14. Example 5.39: when du is off by a constant

Worked example

Use substitution to evaluate the indefinite integral.

\[ \int 3x^2 e^{2x^3}\,dx \]

Let u be the exponent

Why: Its derivative is a multiple of x squared, which is present.

\[ u = 2x^3, \qquad du = 6x^2\,dx \]

Match the constant

Why: The integrand has 3x squared, which is half of 6x squared.

\[ 3x^2\,dx = \frac12\,du \]

Rewrite in u

Why: The half comes outside.

\[ \int 3x^2 e^{2x^3}\,dx = \frac12\int e^{u}\,du \]

Integrate

Why: The exponential is its own antiderivative.

\[ \frac12\int e^{u}\,du = \frac12 e^{u} + C \]

Substitute back

Why: Put 2x cubed back for u.

\[ \int 3x^2 e^{2x^3}\,dx = \frac12 e^{2x^3} + C \]

Check by differentiating

Why: The chain rule brings down 6x squared; half of it is 3x squared.

\[ \frac{d}{dx}\left(\frac12 e^{2x^3}\right) = \frac12 e^{2x^3}\cdot 6x^2 = 3x^2 e^{2x^3} \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 517 — Example 5.39

The exponent is 2x cubed, and its derivative is 6x squared. The integrand has 3x squared, which is exactly half of that. The book's phrase is that du is off by a constant multiplier, and a constant multiplier is always fixable.

So write 3x squared dx as one half du, pull the half out in front, and integrate e to the u. The answer is one half of e to the 2x cubed.

The check shows why the half is right. Differentiating e to the 2x cubed brings down 6x squared, and half of that is 3x squared, the original factor. Compare this with e to the x squared on its own, where the needed factor was a variable and could not be supplied.

15. Checkpoint 5.33: predict the constant

Prediction

\[ \int 2x^3 e^{x^4}\,dx \]

Predict first

Which is the antiderivative?

  • (1/2) e^(x⁴) + C
  • 2 e^(x⁴) + C
  • (1/2) x⁴ e^(x⁴) + C
  • e^(x⁵)/5 + C

Correct: (1/2) e^(x⁴) + C

Why: With u equal to x to the fourth, du is four x cubed dx, so two x cubed dx is half of du. The integral is one half of e to the u. The option with x to the fourth multiplied in front forgets that the x cubed was used up by du, and the last option applies the power rule to the exponent.

\[ u = x^4,\; du = 4x^3\,dx \;\Longrightarrow\; \int 2x^3 e^{x^4}\,dx = \frac12\int e^{u}\,du = \frac12 e^{x^4} + C \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 517 — Checkpoint 5.33

Commit to an option before revealing. Each wrong option is a specific mistake, so it is worth knowing which one you would have made.

Let u be x to the fourth; du is 4x cubed dx, and the integrand has 2x cubed dx, which is half of du. So the answer is one half of e to the x to the fourth.

If you picked the option with x to the fourth multiplied in front, you integrated the x cubed separately, forgetting that du had already used it up. If you picked the last option, you applied the power rule to the exponent, the trap from a few slides ago.

16. Does the exponent's derivative appear?

Sorting

Sort into buckets

Sort each integral: does u equal the exponent work, or is the needed factor wrong or missing?

u = exponent works
∫ e^(5x) dx; ∫ x e^(x²) dx; ∫ x² e^(x³) dx; ∫ e^(√x)/√x dx
factor wrong or missing
∫ e^(x²) dx; ∫ x e^(x³) dx
yes
e^(5x) needs only the constant 5. For x e^(x²) and x² e^(x³) the derivatives 2x and 3x² are present up to a constant. For e^(√x)/√x the derivative of √x is 1/(2√x), also present up to a constant.
no
e^(x²) would need a factor of x, and x e^(x³) would need x², not x. A missing or wrong power of x cannot be fixed by moving a constant, so substitution fails.

For each integral, find the derivative of the exponent, then look for it among the other factors, allowing any constant multiple. That is the whole test.

The four that work have what they need. The derivative of 5x is a constant. The derivatives of x squared and x cubed are 2x and 3x squared, present up to a constant. And the derivative of the square root of x is one over twice the square root of x, which is present as the division by the root.

The two that fail are missing a power of x. The integral of e to the x squared would need an x. The integral of x times e to the x cubed has an x but needs an x squared, and having the wrong power is just as fatal as having none: you cannot move a variable across the integral sign.

17. Other bases

Section

Part 2

18. Every base is e in disguise

Concept

Any positive base a equals e raised to its own natural logarithm. Raise both sides to the power x.

\[ a = e^{\ln a} \quad\Longrightarrow\quad a^{x} = \left(e^{\ln a}\right)^{x} = e^{x\ln a} \]

Figure (svg): Four exponential curves through the point (0, 1): one half to the x falling, and 2 to the x, e to the x and 3 to the x rising, steeper in that order. Short tangent segments at x equals 0 have slopes ln 2, 1 and ln 3.

Every exponential passes through (0, 1), and its slope there is ln a. Only e has slope exactly 1, which is why e to the x alone needs no correction factor.

So a to the x is the natural exponential with a linear exponent, and the slope of that exponent is ln a.

Stewart, Calculus: Early Transcendentals 8e, §3.6 Derivatives of Logarithmic Functions §3.6, pp. 218-223 — the base-change identity

The identity on this slide is what lets one rule handle every base. Any positive number a can be written as e raised to ln a, because the exponential and the natural logarithm undo each other. Raise both sides to the power x and the rules of exponents give e to the x ln a.

Look at the figure. All four exponentials pass through the point zero, one, but they leave it at different slopes. The slope of a to the x at zero is ln a: about 0.69 for base 2, exactly 1 for base e, about 1.10 for base 3, and negative for base one half, which is why that curve falls.

So a to the x is nothing but the natural exponential with a linear exponent whose slope is ln a. And you already know how to integrate an exponential with a linear exponent.

19. Deriving the integral of a to the x

Concept

With the base rewritten, the integral is the linear-exponent case with k equal to ln a.

\[ \int a^{x}\,dx = \int e^{x\ln a}\,dx \]

\[ \int e^{x\ln a}\,dx = \frac{1}{\ln a}e^{x\ln a} + C \]

\[ \int a^{x}\,dx = \frac{a^{x}}{\ln a} + C \qquad (a > 0,\; a \ne 1) \]

A quick check on a definite integral from Exercise 322: two to the x from 0 to 3.

\[ \int_0^{3} 2^{x}\,dx = \frac{2^3 - 2^0}{\ln 2} = \frac{7}{\ln 2} \approx 10.099 \]

A numerical integration gives 10.099 as well. The base 1 is excluded because its logarithm is zero, and 1 to the x is just the constant 1.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 516, 523 — rule (5.21) and Exercise 322

This is the linear-exponent rule with k equal to ln a. Divide by the slope, then convert back from base e to base a, and you have the second line of rule 5.21.

The definite integral from Exercise 322 is a good first use of it. Two to the x from 0 to 3 is two cubed minus two to the zero, which is 7, divided by ln 2, about 10.099. A numerical estimate lands on the same number.

Two bases have to be excluded. A negative or zero base does not give a sensible exponential function at all. And base 1 gives the constant function 1, whose logarithm is zero, so the formula would divide by zero; its integral is just x.

20. Reading rule (5.21) piece by piece

Notation

Annotate

On: \( \int a^{x}\,dx = \frac{a^{x}}{\ln a} + C \)

  • The variable is in the exponent and the base is a fixed positive number. This is not x to a power.
  • The answer keeps the same exponential: exponentials integrate to themselves, up to a constant.
  • The correction factor. It cancels the ln a that differentiating a to the x produces.
  • Then ln a is 1 and the rule collapses to the first line of (5.21): e to the x integrates to itself.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 516 — rule (5.21)

Step through the notes one at a time. The first is the distinction the whole section rests on: a to the x has the variable in the exponent. It is not x to a power, and the power rule has nothing to say about it.

The second and third notes together describe the answer: the same exponential, divided by ln a. The division is there because differentiating a to the x produces a factor of ln a, and the integral has to cancel it.

The last note ties the two lines of rule 5.21 together. When the base is e, ln a is one, the division disappears, and you are back to the statement that e to the x is its own antiderivative.

21. Why e is the natural base

Intuition

Every exponential integrates to itself divided by the logarithm of its base. The base whose logarithm is 1 needs no division at all.

base aln aintegral of a to the x
20.6932ˣ / 0.693 ≈ 1.443 · 2ˣ
e1eˣ
31.0993ˣ / 1.099 ≈ 0.910 · 3ˣ
102.30310ˣ / 2.303 ≈ 0.434 · 10ˣ

That is what natural means here: e is the base for which calculus adds no constant, in either direction.

The table shows the correction factor for four bases. For base 2 you must multiply by about 1.443; for base 10, by about 0.434. For base e, by exactly one.

That is the real reason calculus prefers base e. You met the same fact in differentiation: every exponential differentiates to itself times ln a, and only for base e is that factor invisible. Integration inherits the same property in reverse.

In practice this means you can always convert an unfamiliar base to base e, do the calculus there, and convert back. The factor ln a will appear on its own, in the right place.

22. Find the error: the logarithm on the wrong floor

Error analysis

Annotate

On: \( \int 3^{x}\,dx = 3^{x}\ln 3 + C \)

  • This is the DERIVATIVE of 3 to the x, not its antiderivative. The rule was applied in the wrong direction.
  • Differentiate the proposed answer: you get 3 to the x times ln 3 times ln 3, which is about 1.21 times the integrand, not the integrand.
  • Divide instead of multiplying: 3 to the x over ln 3, plus C. Differentiating that gives exactly 3 to the x.

\[ \int 3^{x}\,dx = \frac{3^{x}}{\ln 3} + C \]

Read the claim and decide what is wrong before revealing the notes. The logarithm of 3 does belong in the answer; the question is where.

The claimed answer is the derivative of 3 to the x, not its antiderivative. Whoever wrote it remembered that a logarithm appears and multiplied by it, when integration has to divide. Differentiating the claimed answer shows the damage: you get the integrand times ln 3 squared, about 1.21 times too big.

A reliable way to keep the direction straight: differentiation makes the natural log appear as a multiplier, so integration, which undoes it, must divide by it.

23. Rates in, totals out

Section

Part 3

24. Integrating a rate gives a total

Concept

A derivative is a rate of change. Integrating a rate recovers the quantity, up to a constant, and one known value pins the constant down.

\[ Q'(t) = q(t) \quad\Longrightarrow\quad Q(t) = \int q(t)\,dt \]

Over an interval, the net change theorem of Section 5.4 says the same thing with limits:

\[ Q(b) = Q(a) + \int_a^{b} q(t)\,dt \]

Growth models are exponential rates, so the rules of Parts 1 and 2 are exactly what these problems need.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 517 — exponential rates and total growth

The applications in this part all have the same shape. You are told how fast something changes, a rate, and asked how much of it there is. Integrating the rate gives the quantity up to a constant, and one known value fixes the constant.

The second form, with limits, is the net change theorem from Section 5.4. The amount at time b is the amount at time a plus the integral of the rate from a to b. It saves you from solving for C when you only need a total at one time.

Growth models are usually exponential, because many things grow at a rate proportional to their own size. That is why the rules from the first two parts are exactly the tools these problems need.

25. Example 5.40: a price-demand equation

Worked example

Demand for a toothpaste is 50 tubes a week at 2.35 dollars a tube. The marginal price-demand function is given. Find the price-demand equation, and the price at 100 tubes a week.

\[ p'(x) = -0.015e^{-0.01x}, \qquad p(50) = 2.35 \]

Integrate the marginal function

Why: The constant factor comes outside.

\[ p(x) = \int -0.015e^{-0.01x}\,dx = -0.015\int e^{-0.01x}\,dx \]

Substitute for the exponent

Why: Its derivative is the constant minus 0.01.

\[ u = -0.01x, \qquad du = -0.01\,dx, \qquad dx = \frac{du}{-0.01} \]

Combine the constants

Why: Minus 0.015 over minus 0.01 is 1.5.

\[ p(x) = \frac{-0.015}{-0.01}\int e^{u}\,du = 1.5\int e^{u}\,du \]

Integrate and substitute back

Why: Keep the constant; the data will fix it.

\[ p(x) = 1.5e^{-0.01x} + C \]

Use the known point

Why: At 50 tubes the price is 2.35.

\[ p(50) = 1.5e^{-0.5} + C = 2.35 \]

Solve for C

Why: 1.5 times e to the minus one half is about 0.910.

\[ C = 2.35 - 1.5e^{-0.5} \approx 2.35 - 0.910 = 1.44 \]

Evaluate at 100 tubes

Why: 1.5 over e is about 0.552.

\[ p(100) = 1.5e^{-1} + 1.44 \approx 0.552 + 1.44 \approx 1.99 \]

Figure (svg): The price-demand curve p of x equals 1.5 e to the minus 0.01 x plus 1.44 for x from 0 to 300 tubes a week, falling from 2.94 toward a dashed floor at 1.44. Dots mark the given point (50, 2.35) and the answer (100, 1.99).

Integration gives the shape of the curve; the one known point, 50 tubes at $2.35, fixes its height. The constant C turns out to be the price the curve levels off at.

Check both conditions

Why: Differentiating returns the marginal function, and the equation reproduces the given point.

\[ p'(x) = 1.5(-0.01)e^{-0.01x} = -0.015e^{-0.01x}, \quad p(50) \approx 0.910 + 1.44 = 2.35 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 518 — Example 5.40

The marginal price-demand function tells you how fast the price changes as weekly demand rises. It is negative, because more tubes sell at lower prices. Integrating it gives the price-demand equation itself, up to a constant.

The integration is the linear-exponent rule: the constant minus 0.015 divided by the slope minus 0.01 gives 1.5. The constant C is then fixed by the one data point, 50 tubes at 2.35 dollars. Since 1.5 times e to the minus one half is about 0.910, C is about 1.44.

Look at the figure: the curve levels off at 1.44 as demand grows, because the exponential term dies away. So C is not just a bookkeeping constant; it is the lowest price this model ever predicts. At 100 tubes the price is about 1.99 dollars, and the check confirms both the derivative and the data point.

26. Example 5.42: bacteria in a Petri dish

Worked example

Bacteria grow at the rate below, in thousands per hour, and the culture starts with 10,000 bacteria. Find the number at any time, and after 2 hours.

\[ q(t) = 3^{t}, \qquad Q(0) = 10 \text{ (thousand)} \]

Integrate the rate

Why: Rule (5.21) with base 3.

\[ Q(t) = \int 3^{t}\,dt = \frac{3^{t}}{\ln 3} + C \]

Use the starting value

Why: At t equal to 0, 3 to the 0 is 1.

\[ Q(0) = \frac{1}{\ln 3} + C = 10 \]

Solve for C

Why: One over ln 3 is about 0.910.

\[ C = 10 - \frac{1}{\ln 3} \approx 9.090 \]

Write the model

Why: Both pieces together.

\[ Q(t) = \frac{3^{t}}{\ln 3} + 9.090 \]

Evaluate at 2 hours

Why: Nine over ln 3 is about 8.192.

\[ Q(2) = \frac{9}{\ln 3} + 9.090 \approx 8.192 + 9.090 = 17.282 \]

Check with the net change theorem

Why: The growth over two hours is the definite integral of the rate, added to the 10 thousand at the start.

\[ 10 + \int_0^{2}3^{t}\,dt = 10 + \frac{9 - 1}{\ln 3} \approx 10 + 7.282 = 17.282 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 519-520 — Example 5.42

Watch the units first. The rate is in thousands of bacteria per hour, so the starting value of 10,000 bacteria becomes 10 in the model. Mixing up thousands and ones is the easiest way to get this wrong.

Integrating 3 to the t uses rule 5.21 with base 3: divide by ln 3. At time zero, 3 to the zero is 1, so the model gives one over ln 3 plus C, and setting that equal to 10 makes C about 9.090. After two hours the model gives 17.282 thousand, which is 17,282 bacteria.

The check uses the net change theorem instead of the constant. The bacteria added in two hours are the integral of the rate from 0 to 2, which is 8 over ln 3, about 7.282 thousand. Ten thousand plus that is the same 17,282. Two routes, one answer.

27. What the bacteria integral looks like

Picture it

Figure (svg): Two panels. Left: the growth rate q of t equals 3 to the t for t from 0 to 2.5 hours, with the area under it from 0 to 2 shaded and labelled 7.28. Right: the population Q of t climbing from 10 at t equals 0 to 17.28 at t equals 2, with the rise of 7.28 marked by a vertical arrow.

The shaded area on the left is exactly the rise on the right. Integrating a rate over an interval gives the change in the total, and the starting value supplies the rest.

On the left, the rate: its area over the first two hours is 7.28 thousand. On the right, the population: it rises by exactly that amount, from 10 to 17.28 thousand. The constant C is what makes the right-hand curve start at 10.

The two panels are the same information seen twice. On the left is the rate: how many thousand bacteria per hour are being added at each moment. The shaded area under it from 0 to 2 hours is the total added in that time, 7.28 thousand.

On the right is the population itself. It starts at 10 thousand and rises, and the green arrow marks its rise over the first two hours: exactly 7.28 thousand, the area on the left.

This is the Fundamental Theorem in a practical form. The area under a rate is the change in the total. The integral tells you the change; only the starting value tells you where the total began.

28. Checkpoint 5.35: base 2 instead of base 3

Comparison

Now the rate is 2 to the t thousand bacteria per hour, still starting from 10 thousand. Fill in the Checkpoint column.

Comparison matrix

quantityrate 3ᵗ (Example 5.42)rate 2ᵗ (Checkpoint 5.35)
divide byln 3ln 2
Q(t)3ᵗ/ln 3 + C2ᵗ/ln 2 + C
C, from Q(0) = 109.0908.557
time asked2 hours3 hours
Q at that time (thousands)17.28220.099

\[ Q(3) = \frac{2^3}{\ln 2} + 10 - \frac{1}{\ln 2} \approx 11.542 + 8.557 = 20.099 \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 520 — Checkpoint 5.35

Work down the empty column using the filled-in column as a model. The only real change is the base, so the correction factor becomes ln 2 instead of ln 3.

The constant comes from the starting value again: one over ln 2 is about 1.443, so C is 10 minus that, about 8.557. At three hours, 2 cubed is 8, and 8 over ln 2 is about 11.542. Adding the constant gives 20.099 thousand, so about 20,099 bacteria.

Notice that the smaller base grows more slowly, so the checkpoint needs three hours to reach roughly what base 3 reached in a little over two.

29. Example 5.43: a fruit fly population

Worked example

A population of fruit flies grows at the rate below, in flies per day, starting from 100 flies. How many are there after 10 days?

\[ g(t) = 2e^{0.02t}, \qquad G(0) = 100 \]

Apply the net change theorem

Why: The flies after 10 days are the starting flies plus the flies added.

\[ G(10) = G(0) + \int_0^{10} 2e^{0.02t}\,dt \]

Find the antiderivative

Why: Linear exponent: divide by 0.02.

\[ \int 2e^{0.02t}\,dt = \frac{2}{0.02}e^{0.02t} = 100e^{0.02t} \]

Evaluate between 0 and 10

Why: At 0 the exponential is 1.

\[ G(10) = 100 + \left[100e^{0.02t}\right]_0^{10} = 100 + 100e^{0.2} - 100 \]

Compute

Why: e to the 0.2 is about 1.2214.

\[ G(10) = 100e^{0.2} \approx 122.14 \approx 122 \text{ flies} \]

Figure (svg): The growth rate g of t equals 2 e to the 0.02 t for t from 0 to 12 days, a gently rising curve from 2 to about 2.5, with the area under it from 0 to 10 shaded and labelled 22.14 flies.

The rate barely changes, from 2 to 2.44 flies a day, so the area is close to a rectangle of height about 2.2 and width 10. The integral makes that estimate exact.

Check against the rate's range

Why: The rate stays between 2 and about 2.44 flies a day, so ten days add between 20 and 24.4 flies. The integral says 22.14.

\[ 10(2) = 20 < 22.14 < 24.43 \approx 10(2e^{0.2}) \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 520 — Example 5.43

This time the book skips the constant altogether and goes straight to the net change theorem: flies after 10 days equal flies at the start plus the integral of the growth rate over those ten days.

The antiderivative of 2e to the 0.02t divides by 0.02, which turns the 2 into 100. Evaluating from 0 to 10 gives 100 e to the 0.2 minus 100, and adding the starting 100 flies leaves 100 e to the 0.2, about 122 flies.

The figure makes the check intuitive. The rate barely changes over ten days, from 2 to about 2.44 flies a day, so the added flies must lie between 20 and about 24.4. The integral says 22.14, right in the middle, and a count of flies is rounded to 122.

30. Checkpoint 5.36: fifteen days of slower growth

Estimation

\[ g(t) = e^{0.01t}, \qquad G(0) = 100, \qquad G(15) = ? \]

Predict first

Estimate first from the rate: it stays close to 1 fly a day. Roughly how many flies are there after 15 days?

  • About 101
  • About 116
  • About 150
  • About 215

Correct: About 116

Why: The rate starts at one fly a day and only reaches about 1.16 by day 15, so fifteen days add a little more than fifteen flies. The integral gives 100 times the quantity e to the 0.15 minus 1, which is 16.18, for a total of about 116. The answer 215 comes from forgetting to divide by 0.01 correctly, and 101 from forgetting the fifteen days.

\[ G(15) = 100 + \left[100e^{0.01t}\right]_0^{15} = 100e^{0.15} \approx 116.18 \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 520 — Checkpoint 5.36

Estimate before choosing. The rate starts at one fly per day and only climbs to about 1.16 by day fifteen. So fifteen days should add a little more than fifteen flies.

The integral confirms it. The antiderivative of e to the 0.01t is 100 e to the 0.01t, and evaluating from 0 to 15 gives 100 times the quantity e to the 0.15 minus 1, about 16.18 flies. With the starting 100, that is about 116 flies.

An estimate from the rate is a cheap and powerful check on growth problems. If your integral had given 215, the estimate would have told you at once that a constant went wrong.

31. Decay: a negative exponent and a half-life

Concept

A quantity that decays at a rate proportional to itself follows an exponential with a negative exponent. The half-life T is the time for it to halve.

\[ A(t) = A_0 e^{-kt}, \qquad A(T) = \frac{A_0}{2} \]

\[ e^{-kT} = \frac12 \quad\Longrightarrow\quad -kT = -\ln 2 \quad\Longrightarrow\quad k = \frac{\ln 2}{T} \]

Figure (svg): A decay curve A of t equals 100 e to the minus kt with a half-life of 5 hours, for t from 0 to 22. Dots at t equals 0, 5, 10, 15 and 20 sit at heights 100, 50, 25, 12.5 and 6.25, each dashed step dropping to half the previous height.

The half-life is the same at every stage: 100 to 50 takes 5 hours, and so does 12.5 to 6.25. A constant ratio in equal times is the signature of the exponential.

The starting amount cancels, so the half-life belongs to the substance, not to the sample. A 5-hour half-life gives a rate constant of about 0.139 per hour.

\[ k = \frac{\ln 2}{5} \approx 0.139 \]

Decay is growth run backwards: an exponential with a negative exponent. The half-life is the time it takes for the quantity to fall to half. Setting A of T equal to half the starting amount and solving gives k equal to ln 2 over T.

The key step is that the starting amount cancels from both sides before you take logarithms. That is why the half-life belongs to the substance and not to the sample: a gram and a tonne of the same isotope halve in the same time.

Look at the dots in the figure: 100, 50, 25, 12.5, 6.25, at equal steps of five hours. Each dashed step is the same length. A fixed fraction lost in a fixed time is what makes a process exponential rather than linear.

32. A decaying source: the total over ten hours

Worked example

A radioactive source emits particles at a rate that decays exponentially, in thousands per hour. How much does it emit in the first 10 hours?

\[ r(t) = 50e^{-0.2t}, \qquad \text{total} = \int_0^{10} 50e^{-0.2t}\,dt \]

Find the antiderivative

Why: Linear exponent: divide by minus 0.2.

\[ \int 50e^{-0.2t}\,dt = \frac{50}{-0.2}e^{-0.2t} = -250e^{-0.2t} \]

Evaluate at the limits

Why: Upper limit first.

\[ \left[-250e^{-0.2t}\right]_0^{10} = -250e^{-2} - \left(-250\right) \]

Simplify

Why: Factor out 250.

\[ = 250\left(1 - e^{-2}\right) \approx 250(0.8647) \approx 216.2 \]

Check against two bounds

Why: The rate falls from 50 to about 6.77, so ten hours emit more than 67.7; and the total over all time is 250, which ten hours cannot exceed.

\[ 67.7 < 216.2 < 250 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 517 — applications of exponential integrals

Here the rate itself is decaying, and the question is how much accumulates. That is a definite integral of the rate. The linear-exponent rule divides by minus 0.2, which turns 50 into minus 250.

Evaluating from 0 to 10 gives 250 times the quantity one minus e to the minus 2, about 216 thousand particles. The factor one minus e to the minus 2 is about 0.865: in ten hours the source has emitted about 86 percent of everything it will ever emit.

The check traps the answer between two numbers you can find without integrating. The rate never drops below its final value of about 6.77, so ten hours emit at least 67.7. And the total over all time is 250, which a finite stretch cannot exceed. The answer 216 sits between them.

33. Slide the decay rate

Tweak it

Parameter explorer

The curve is the total emitted from time 0 to time x by a source with rate e^(−kt). Slide k. Where does the total level off, and how fast does it get there?

\[ \int_0^{x} e^{-{k}t}\,dt = \frac{1 - e^{-{k}x}}{{k}} \]

  • k — from 0.1 to 2: decay rate k

Start with k at one half and watch the curve rise and flatten. The total emitted up to time x is one minus e to the minus kx, all divided by k, and as x grows it levels off at one over k.

Now slide k down toward 0.1. The ceiling rises to ten, but the curve takes much longer to get there. Slide k up to 2 and the ceiling drops to one half, reached almost at once.

This is the squeeze picture from Part 1 again. A slow decay means a wide, gentle curve with a large total area; a fast decay means a narrow curve with a small one. The one over k is the same one over k.

34. Caffeine and its half-life

Real world

Discussion prompt

Caffeine has a half-life of about 5 hours in an adult. You drink 200 mg at 8 am. Find the rate constant, and how much caffeine is left at 8 pm.

Work it out before revealing. The rate constant comes straight from the half-life formula: ln 2 over 5, about 0.139 per hour.

Twelve hours is two and two fifths half-lives, so the amount left is 200 milligrams times one half raised to the power 2.4, about 37.9 milligrams. You can get the same number from e to the minus 12k, since the two forms are the same function.

Almost a fifth of the dose is still there twelve hours later. Half-lives vary a lot between people, so treat this as a model, but the shape of the calculation is exactly the one used for real drug dosing.

35. Definite integrals: change the limits

Section

Part 4

36. Carry the limits through the substitution

Concept

In a definite integral, substitution changes the variable, so the limits must change with it. The x limits become u limits through the same formula.

\[ \int_a^{b} f\big(g(x)\big)g'(x)\,dx = \int_{g(a)}^{g(b)} f(u)\,du \]

Then there is no need to go back to x at all: evaluate the u antiderivative at the u limits.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 519 — Example 5.41, changing the limits

When a definite integral needs a substitution, you have two choices. You can find the antiderivative in u, convert it back to x, and use the original limits. Or you can convert the limits to u values and never go back. The book uses the second way throughout this section.

The rule is simply to put each x limit into the substitution formula. If u is g of x, the limit a becomes g of a and the limit b becomes g of b. The lower limit stays the lower limit, even if its u value turns out larger.

Mixing the two methods is the error to avoid: evaluating a u antiderivative at x limits gives a wrong number with no warning.

37. Example 5.41: e to the 1 minus x from 1 to 2

Worked example

Evaluate the definite integral.

\[ \int_1^{2} e^{1-x}\,dx \]

Let u be the exponent

Why: Its derivative is minus one.

\[ u = 1 - x, \qquad du = -dx \]

Change the limits

Why: Put each x limit into u equal to 1 minus x.

\[ x = 1 \Rightarrow u = 0, \qquad x = 2 \Rightarrow u = -1 \]

Rewrite the integral

Why: The minus sign from du comes outside.

\[ \int_1^{2} e^{1-x}\,dx = -\int_0^{-1} e^{u}\,du \]

Swap the limits to remove the minus

Why: Reversing the limits changes the sign.

\[ -\int_0^{-1} e^{u}\,du = \int_{-1}^{0} e^{u}\,du \]

Evaluate

Why: The antiderivative is e to the u.

\[ \int_{-1}^{0} e^{u}\,du = e^{0} - e^{-1} = 1 - \frac1e \approx 0.632 \]

Figure (svg): The curve y equals e to the 1 minus x for x from minus 1 to 3, falling from e squared toward zero, with the region under it from x equals 1 to x equals 2 shaded.

The integrand falls from 1 to 0.368 across the strip, so the area must lie between those two numbers. The exact value, 0.632, does.

Check without substituting

Why: The antiderivative of e to the 1 minus x is minus e to the 1 minus x; evaluate it in x directly.

\[ \left[-e^{1-x}\right]_1^{2} = -e^{-1} + e^{0} = 1 - \frac1e \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 518-519 — Example 5.41 and Figure 5.38

Let u be the exponent, 1 minus x, so du is minus dx. Putting the limits through the formula, x equal to 1 gives u equal to 0 and x equal to 2 gives u equal to minus 1. The new limits run downward, from 0 to minus 1.

The minus sign from du and the backwards limits cancel each other: swapping the limits changes the sign once more, leaving the integral of e to the u from minus 1 to 0. That is 1 minus one over e, about 0.632.

The figure gives a quick plausibility check: the integrand falls from 1 to 0.368 across a strip of width 1, so the area must lie between those numbers. The final step checks exactly, by integrating directly in x with the antiderivative minus e to the 1 minus x.

38. Checkpoint 5.34: how big is the area under e to the 2x?

Estimation

\[ \int_0^{2} e^{2x}\,dx \]

Predict first

The integrand climbs from 1 to e⁴ ≈ 54.6 across the interval. Which is closest to the integral?

  • About 7
  • About 27
  • About 55
  • About 109

Correct: About 27

Why: With u equal to 2x, the limits become 0 and 4, and dx is du over 2, so the integral is half of e to the 4 minus 1, about 26.8. The answer 55 forgets to divide by 2, and 109 multiplies by 2 instead. A curve that rises steeply puts most of its area near the right end, so it is far less than the rectangle of height 54.6 and width 2.

\[ \frac12\int_0^{4} e^{u}\,du = \frac{e^4 - 1}{2} \approx 26.80 \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 519 — Checkpoint 5.34

Pick an option before revealing. The integrand goes from 1 up to about 54.6, so a rectangle of full height would give about 109, and the true area is much less, because the curve only gets tall near the right end.

With u equal to 2x, the limits become 0 and 4, and dx is half of du. The integral is half of e to the 4 minus 1, about 26.8.

The two tempting wrong answers each mishandle the constant: 55 forgets to divide by 2, and 109 multiplies by it. Estimating the size first catches both.

39. Example 5.44: e to the 1 over x, over x squared

Worked example

Evaluate the definite integral using substitution.

\[ \int_1^{2} \frac{e^{1/x}}{x^2}\,dx \]

Rewrite with negative powers

Why: The exponent and the denominator are both powers of x.

\[ \int_1^{2} \frac{e^{1/x}}{x^2}\,dx = \int_1^{2} e^{x^{-1}}x^{-2}\,dx \]

Let u be the exponent

Why: Its derivative is minus x to the minus 2, which is present up to sign.

\[ u = x^{-1}, \qquad du = -x^{-2}\,dx \]

Change the limits

Why: One over 1 is 1, one over 2 is one half.

\[ x = 1 \Rightarrow u = 1, \qquad x = 2 \Rightarrow u = \tfrac12 \]

Rewrite in u

Why: Minus du replaces x to the minus 2 dx.

\[ \int_1^{2} e^{x^{-1}}x^{-2}\,dx = -\int_1^{1/2} e^{u}\,du \]

Swap the limits

Why: The limits run downward; reversing them absorbs the minus sign.

\[ -\int_1^{1/2} e^{u}\,du = \int_{1/2}^{1} e^{u}\,du \]

Evaluate

Why: e to the one half is the square root of e.

\[ \int_{1/2}^{1} e^{u}\,du = e - e^{1/2} = e - \sqrt e \approx 1.070 \]

Figure (svg): Two panels. Left: y equals e to the 1 over x, over x squared, for x from 0.8 to 2.3, shaded from 1 to 2. Right: y equals e to the u for u from 0 to 1.3, shaded from one half to 1. Both shaded areas are labelled 1.070.

Two different regions with the same area. The substitution reshapes the region, and changing the limits moves it to where it now lives.

Check numerically

Why: A Simpson's rule estimate of the original integral in x, with 2000 strips, agrees to every digit shown.

\[ \int_1^{2} \frac{e^{1/x}}{x^2}\,dx \approx 1.06956, \qquad e - \sqrt e \approx 1.06956 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 520-521 — Example 5.44

The first move is to rewrite everything as powers of x. The exponent is x to the minus one, and dividing by x squared is multiplying by x to the minus two. Now you can see that x to the minus two is, up to sign, the derivative of the exponent.

So let u be one over x, with du equal to minus x to the minus two dx. The limits become 1 and one half, a decreasing pair, because one over x decreases. The minus sign and the backwards limits cancel, leaving the integral of e to the u from one half to 1, which is e minus the square root of e, about 1.070.

The two panels of the figure show what the substitution did. The region in x and the region in u have completely different shapes, but the same area, 1.070. A numerical estimate of the original integral agrees to five decimal places.

40. When the new limits run backwards

Concept

A decreasing substitution, such as one over x or 1 minus x, turns the lower limit into the larger u value. That is not an error.

\[ \int_{c}^{d} f(u)\,du = -\int_{d}^{c} f(u)\,du \]

You can evaluate the backwards integral directly, or swap the limits and change the sign, which the book does. Both give the same number; what you must never do is swap the limits silently.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 521 — the remark in Example 5.44

Examples 5.41 and 5.44 both produced limits where the lower one is bigger than the upper one. That happens whenever the substitution is decreasing, and it is perfectly legal.

You have two honest options. Evaluate the backwards integral as it stands, upper limit minus lower limit, or swap the limits and put a minus sign in front. The book does the second, usually because a minus sign from du is waiting to be cancelled.

The dishonest option, and a common one, is to swap the limits quietly to make them look normal and forget the sign. The answer then comes out as the negative of the truth, and nothing in the arithmetic warns you.

41. The reciprocal and the logarithm

Section

Part 5

42. The one power the power rule misses

Concept

The power rule divides by n plus 1. At n equal to minus 1 that is division by zero, so the rule has a hole exactly at one over x.

\[ \int x^{n}\,dx = \frac{x^{n+1}}{n+1} + C \quad (n \ne -1) \]

Figure (svg): Four curves for x from 0.3 to 4: the antiderivatives x to the h minus 1, over h, for h equal to 1, one half and one fifth, and the curve ln x beneath them. As h shrinks the curves close in on ln x.

Each curve is the power-rule antiderivative of a power just above minus 1, arranged to be zero at x equals 1. They crowd down onto ln x: the logarithm is where the power rule was heading all along.

The logarithm fills the hole, and the picture shows it is not an accident: the power-rule antiderivatives of powers just above minus one, started at 1, close in on ln x.

\[ \int_1^{x} t^{h-1}\,dt = \frac{x^{h} - 1}{h} \;\longrightarrow\; \ln x \quad (h \to 0) \]

The power rule works for every exponent except one. At n equal to minus one it would divide by zero, and that is exactly the integrand one over x. The logarithm is the function that fills the hole.

The figure shows that the logarithm is not an arbitrary patch. Each colored curve is the power-rule antiderivative of x to a power just above minus one, shifted to be zero at x equal to 1. As the power creeps toward minus one, the curves settle down onto ln x.

So the logarithm is the natural continuation of the power rule to the one exponent the rule cannot handle. That explains why ln x grows so slowly: it sits at the boundary between the powers that grow and the powers that level off.

43. Why the absolute value

Concept

One over x is defined for negative x too, but ln x is not. The fix is ln of the absolute value, and its derivative is one over x on both sides.

\[ x < 0: \quad \frac{d}{dx}\ln(-x) = \frac{1}{-x}\cdot(-1) = \frac1x \]

Figure (svg): The graph of y equals ln of the absolute value of x: two mirror-image branches, one for x negative and one for x positive, each dropping toward minus infinity near the y-axis. Tangent segments at x equals 2 and x equals minus 2 have slopes one half and minus one half.

The left branch is the mirror image of the right one, so its slopes are the negatives: minus one half at minus 2. And minus one half is exactly 1/x at x equals minus 2. One formula, both branches.

\[ \int \frac{dx}{x} = \ln|x| + C \qquad (x \ne 0) \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 521 — rule (5.22), first line

One over x is defined for negative x, so it needs an antiderivative there too. Plain ln x will not do, because the logarithm of a negative number does not exist.

For negative x, try ln of minus x, which is defined. The chain rule gives one over minus x times minus one, which is one over x. So ln of minus x works on the left and ln x works on the right, and the absolute value combines them into one formula.

The figure shows the geometry. The left branch is the mirror image of the right, so its slopes are negated: at minus 2 the slope is minus one half, which is exactly one over minus 2. The two branches never meet, because the point zero is missing from both.

44. Reading the reciprocal rule piece by piece

Notation

Annotate

On: \( \int \frac{du}{u} = \ln|u| + C, \quad u \ne 0 \)

  • The reciprocal of the variable, the power minus one. Any integrand that can be written in this shape integrates to a logarithm.
  • Makes the answer defined for negative u as well. Its derivative is one over u on both branches.
  • Strictly, a separate constant is allowed on each branch, because the two branches are not connected.
  • The rule never crosses zero. A definite integral of one over u must stay on one side.

Step through the notes. The first is about recognition: anything that can be put in the shape du over u integrates to a logarithm, and Part 6 is devoted to spotting that shape.

The second note is the absolute value you just derived. The third is a fine point: since the two branches are separate pieces, you are allowed a different constant on each. For a definite integral on one branch this never matters, but it is the honest statement.

The last note is the one that costs marks. The rule never crosses zero. A definite integral of one over u has to live entirely on one side of zero, a restriction the absolute value does nothing to remove.

45. Two branches, mirror areas

Picture it

Figure (svg): The hyperbola y equals 1 over x for x from minus 4 to 4. The region between the curve and the axis from x equals 1 to 3 is shaded above the axis; the mirror-image region from minus 3 to minus 1 is shaded below the axis.

The hyperbola is symmetric through the origin, so the two shaded regions are congruent. Below the axis the integral counts the area as negative: minus ln 3.

The hyperbola is symmetric through the origin. The region from minus 3 to minus 1 is the mirror image of the region from 1 to 3, but it lies below the axis, so its integral is the negative of the other one.

The hyperbola is symmetric through the origin: turn the right branch half a turn about the origin and it lands on the left branch. So the shaded region from minus 3 to minus 1 is congruent to the one from 1 to 3.

The areas are the same size, but the left region lies below the axis, so a definite integral counts it as negative. That predicts the answer to the next example before you do any calculus: minus ln 3.

46. A negative interval: one over x from minus 3 to minus 1

Worked example

Evaluate a definite integral entirely on the negative side.

\[ \int_{-3}^{-1} \frac{dx}{x} \]

Choose the antiderivative with the absolute value

Why: Plain ln x is undefined at negative numbers.

\[ \int \frac{dx}{x} = \ln|x| + C \]

Evaluate at the limits

Why: Upper minus lower.

\[ \ln|-1| - \ln|-3| = \ln 1 - \ln 3 \]

Simplify

Why: The logarithm of 1 is 0.

\[ \int_{-3}^{-1} \frac{dx}{x} = -\ln 3 \approx -1.099 \]

Check by symmetry

Why: The integrand is odd, so the region is the mirror image of the one from 1 to 3, below the axis.

\[ \int_{-3}^{-1} \frac{dx}{x} = -\int_{1}^{3} \frac{dx}{x} = -\ln 3 \;\checkmark \]

This example exists to show the absolute value doing its job. The interval lies entirely on the negative side, where one over x is continuous, so the integral exists and the Fundamental Theorem applies.

With the absolute value, the upper limit gives ln 1, which is zero, and the lower limit gives ln 3. Upper minus lower is minus ln 3, about minus 1.099.

The check is the symmetry from the previous slide. One over x is an odd function, so its integral over a negative interval is the negative of its integral over the mirror-image positive interval. The sign agrees with the picture too: the integrand is negative there, so the integral must be.

47. Trap: dropping the absolute value

Trap

The trap

The antiderivative is remembered without the bars:

\[ \int \frac{dx}{x} = \ln x + C \]

\[ \int_{-3}^{-1}\frac{dx}{x} = \ln(-1) - \ln(-3) \]

Wrong. Neither logarithm exists.

The fix

ln x is the antiderivative only for positive x. The absolute value is what makes the formula work on the negative branch.

\[ \ln|-1| - \ln|-3| = -\ln 3 \]

Drop the bars only when you know the inside is positive, such as x squared plus 1.

This slip is invisible as long as every x is positive, which is why it survives so long. The moment you evaluate on a negative interval, it produces logarithms of negative numbers, which do not exist.

The repair is to write the absolute value every time you integrate one over something. You may drop the bars afterwards when you know the inside is positive, as with x squared plus 1, or with x on an interval to the right of zero.

48. Break the rule: integrating across zero

Counterexample

Figure (svg): The hyperbola y equals 1 over x for x from minus 1 to 2, with the region between the curve and the axis shaded on both sides of zero. Near x equals 0 the curve shoots off to plus and minus infinity, and the shading runs off the top and bottom of the picture.

The region is not a finite shape with an area to compute. The Fundamental Theorem needs a continuous integrand on the whole interval, and this one is not even defined at 0.

Discussion prompt

Someone writes the line below. The antiderivative is right, and each logarithm exists. What is wrong, and what is the integral?

Decide what is wrong before revealing. The antiderivative is correct, the absolute values are there, and every number is defined. The problem is not in the arithmetic.

The Fundamental Theorem needs the integrand to be continuous on the whole interval, and one over x is not even defined at zero. The figure shows what that means: near zero the region shoots off to infinity on both sides. There is no finite area to compute.

So the line on the slide is a meaningless calculation that happens to produce a tidy number. The integral from minus 1 to 2 does not exist. Whenever you integrate one over something, check that the denominator never vanishes on your interval.

49. Example 5.45: three over x minus 10

Worked example

Find the antiderivative.

\[ \int \frac{3}{x - 10}\,dx \]

Move the constant outside

Why: The 3 is a constant factor.

\[ \int \frac{3}{x-10}\,dx = 3\int \frac{1}{x-10}\,dx \]

Let u be the denominator

Why: Its derivative is 1.

\[ u = x - 10, \qquad du = dx \]

Rewrite in u

Why: Exactly the reciprocal rule.

\[ 3\int \frac{du}{u} = 3\ln|u| + C \]

Substitute back

Why: The domain excludes 10.

\[ \int \frac{3}{x-10}\,dx = 3\ln|x - 10| + C, \quad x \ne 10 \]

Figure (svg): The graph of y equals 3 over x minus 10 for x from minus 2 to 32, with a dashed vertical asymptote at x equals 10: the left branch is below the axis and plunges downward near 10, the right branch is above and shoots upward near 10.

The antiderivative 3 ln|x − 10| works on each branch separately. A definite integral must stay on one side of the asymptote.

Check on both branches

Why: The absolute value rule differentiates on either side of 10; and a definite integral on the right branch matches a numerical estimate.

\[ \frac{d}{dx}3\ln|x-10| = \frac{3}{x-10}, \qquad \int_{11}^{13}\frac{3\,dx}{x-10} = 3\ln 3 \approx 3.296 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 521-522 — Example 5.45 and Figure 5.39

The constant 3 comes out first. Then the denominator is x minus 10, whose derivative is 1, so the substitution u equal to x minus 10 changes nothing but the name, and the reciprocal rule gives 3 ln of the absolute value of u.

The book records the domain restriction with the answer, x not equal to 10. The figure, the book's Figure 5.39, shows why: there is a vertical asymptote at 10, with one branch on each side. The antiderivative works on each branch separately.

The check differentiates on both sides using the absolute-value rule, and then evaluates a definite integral that stays on the right branch, from 11 to 13. The answer, 3 ln 3, matches a numerical estimate.

50. The logarithmic form

Section

Part 6

51. Numerator equals the derivative of the denominator

Concept

When the top of a fraction is the derivative of the bottom, substitute for the whole denominator. What is left is one over u.

\[ u = g(x), \quad du = g'(x)\,dx \]

\[ \int \frac{g'(x)}{g(x)}\,dx = \int \frac{du}{u} = \ln|g(x)| + C \]

If the top is only a constant multiple of the derivative, adjust the constant, exactly as in Part 1. If it is something else, this rule does not apply.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, pp. 521-522 — the u to the minus one rule

This is the most useful shape in the second half of the section. When the top of a fraction is the derivative of the bottom, substitute u for the whole bottom. Then du is exactly the top times dx, and the integral collapses to du over u.

So the answer is always the logarithm of the absolute value of the denominator. You do not even need to write out the substitution once you are confident: look at the fraction, check that the top is the derivative of the bottom, and write down ln of the bottom.

As with exponentials, a constant multiple is fine. If the top is half the derivative of the bottom, the answer is half the logarithm. If the top is a different function altogether, stop: this rule does not apply.

52. Example 5.46: a rational function with a hidden derivative

Worked example

Find the antiderivative.

\[ \int \frac{2x^3 + 3x}{x^4 + 3x^2}\,dx \]

Let u be the denominator

Why: Differentiate it.

\[ u = x^4 + 3x^2, \qquad du = \left(4x^3 + 6x\right)dx \]

Factor the differential

Why: The bracket is twice the numerator.

\[ du = 2\left(2x^3 + 3x\right)dx \]

Solve for the numerator times dx

Why: Divide by 2.

\[ \left(2x^3 + 3x\right)dx = \frac12\,du \]

Rewrite in u

Why: The half comes outside.

\[ \int \frac{2x^3 + 3x}{x^4 + 3x^2}\,dx = \frac12\int \frac{du}{u} \]

Integrate and substitute back

Why: The reciprocal rule.

\[ \frac12\ln|u| + C = \frac12\ln\left|x^4 + 3x^2\right| + C \]

Check by differentiating

Why: The derivative of the logarithm is the derivative of the inside over the inside, halved.

\[ \frac12\cdot\frac{4x^3 + 6x}{x^4 + 3x^2} = \frac{2x^3 + 3x}{x^4 + 3x^2} \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 522 — Example 5.46

At first sight this looks like a hard rational function. The trick is to differentiate the denominator and compare: the derivative of x to the fourth plus 3x squared is 4x cubed plus 6x, and that is exactly twice the numerator.

So let u be the denominator, factor the 2 out of du, and the numerator times dx becomes half of du. The integral is half the integral of du over u, which is half the logarithm of the absolute value of the denominator.

The check is the chain rule for logarithms: the derivative of ln of something is the derivative of the something over the something. Halve it and you are back at the integrand. At x equal to 1, the derivative of the answer and the integrand are both 1.25.

53. Exercise 332: the integral of tangent

Worked example

Tangent does not look like a fraction until you write it as one.

\[ \int \tan\theta\,d\theta \]

Write tangent as a quotient

Why: Sine over cosine.

\[ \int \tan\theta\,d\theta = \int \frac{\sin\theta}{\cos\theta}\,d\theta \]

Let u be the denominator

Why: The numerator is its derivative, up to sign.

\[ u = \cos\theta, \qquad du = -\sin\theta\,d\theta \]

Rewrite in u

Why: Minus du replaces sine d theta.

\[ \int \frac{\sin\theta}{\cos\theta}\,d\theta = -\int \frac{du}{u} \]

Integrate and substitute back

Why: The reciprocal rule.

\[ -\ln|u| + C = -\ln|\cos\theta| + C = \ln|\sec\theta| + C \]

Check by differentiating, then on a definite integral

Why: The chain rule returns the tangent; and Exercise 356's integral from 0 to a quarter turn matches a numerical estimate.

\[ \frac{d}{d\theta}\left(-\ln|\cos\theta|\right) = \frac{\sin\theta}{\cos\theta}, \quad \int_0^{\pi/4}\tan\theta\,d\theta = \ln\sqrt2 \approx 0.3466 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 524 — Exercises 332 and 356

Tangent does not look like a fraction, which is exactly why this integral catches people. Write it as sine over cosine and the shape appears: the derivative of cosine is minus sine, so the top is the derivative of the bottom up to a sign.

Let u be cosine. Then sine d theta is minus du, and the integral becomes minus the integral of du over u. So the integral of tangent is minus ln of the absolute value of cosine, which you can also write as ln of the absolute value of secant.

The check differentiates back to tangent, then uses the formula for Exercise 356: tangent from 0 to a quarter turn gives ln of the square root of 2, about 0.3466, matching a numerical estimate. This is a standard result worth remembering.

54. Example 5.48: sine over 1 plus cosine

Worked example

Find the definite integral.

\[ \int_0^{\pi/2} \frac{\sin x}{1 + \cos x}\,dx \]

Let u be the denominator

Why: Its derivative is minus sine x.

\[ u = 1 + \cos x, \qquad du = -\sin x\,dx \]

Change the limits

Why: Cosine is 1 at 0 and 0 at a quarter turn.

\[ x = 0 \Rightarrow u = 2, \qquad x = \tfrac{\pi}{2} \Rightarrow u = 1 \]

Rewrite in u

Why: Minus du replaces sine x dx.

\[ \int_0^{\pi/2} \frac{\sin x}{1 + \cos x}\,dx = -\int_2^{1} \frac{du}{u} \]

Swap the limits

Why: This removes the minus sign.

\[ -\int_2^{1} \frac{du}{u} = \int_1^{2} \frac{du}{u} \]

Evaluate

Why: The logarithm of 1 is 0.

\[ \ln|u|\Big|_1^{2} = \ln 2 - \ln 1 = \ln 2 \approx 0.693 \]

Figure (svg): The curve y equals sine x over 1 plus cosine x for x from 0 to about 1.9, rising from 0 through 1 at x equals pi over 2; the region under it from 0 to pi over 2 is shaded.

The integrand climbs from 0 to 1 across the interval, so the area must be less than pi over 2 times 1, about 1.57, and ln 2 comfortably is.

Check numerically

Why: A Simpson's rule estimate of the original integral agrees.

\[ \int_0^{\pi/2} \frac{\sin x}{1 + \cos x}\,dx \approx 0.693147 = \ln 2 \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 523 — Example 5.48

The denominator is 1 plus cosine, and its derivative is minus sine, which is the numerator up to a sign. So u is the whole denominator. The limits convert to 2 and 1, running backwards, and the minus sign from du cancels against swapping them.

What remains is the integral of du over u from 1 to 2, which is ln 2 minus ln 1, or ln 2, about 0.693. The book introduces this as a trigonometric integral, but the only idea used is the logarithmic form.

The figure shows a curve that climbs from 0 to 1 across the interval, so the area has to be less than the rectangle, about 1.57. The value ln 2 fits, and a numerical estimate agrees to six decimal places.

55. Trap: forcing the logarithm

Trap

The trap

Every fraction starts to look like a logarithm:

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

Wrong. The numerator is not the derivative of the denominator.

The fix

Differentiate the claim: you get 2x over x squared plus 1, not 1 over it. The 2x is missing and is not a constant, so it cannot be fixed.

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

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

Once the logarithmic form clicks, every fraction starts to look like one. But the rule needs the numerator to be the derivative of the denominator, and here the numerator is 1 while the derivative of the denominator is 2x.

The missing 2x is a variable, not a constant, so you cannot manufacture it. Differentiating the claimed answer exposes the error at once: you get 2x over x squared plus 1, not 1 over it.

Put the two integrals side by side and remember them as a pair. With 2x on top, a logarithm. With 1 on top, an arctangent, which Section 5.7 is about.

56. Match each integrand to its antiderivative

Matching

Match the pairs

  • a. 1/(x + 2)
  • b. 2x/(x² + 1)
  • c. x/(x² + 1)
  • d. 1/(x² + 1)
  • e. cos x/sin x
  • p. ln|x + 2| + C
  • q. ln(x² + 1) + C
  • r. (1/2) ln(x² + 1) + C
  • s. arctan x + C
  • t. ln|sin x| + C

Why: Four of the five are the logarithmic form: the numerator is the derivative of the denominator, exactly or up to a constant, as with x over x squared plus 1, which needs a half. The fraction one over x squared plus 1 is the odd one out: its numerator is not the derivative, and its antiderivative is the arctangent. The first line is Checkpoint 5.38.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 522 — Checkpoint 5.38

Match all five before checking. For each fraction, differentiate the denominator and compare it with the numerator; that decides everything.

Four of them are the logarithmic form. One over x plus 2 is Checkpoint 5.38, with a denominator whose derivative is 1. 2x over x squared plus 1 is exact, x over x squared plus 1 needs a half, and cosine over sine gives the logarithm of sine.

One over x squared plus 1 is the odd one out, the fraction from the trap slide. Its antiderivative is the arctangent, not a logarithm.

57. The integral of ln x

Concept

Rule (5.22) also gives the integral of the logarithm itself. Deriving it needs integration by parts, a Calculus II technique, but checking it needs only the product rule.

\[ \int \ln x\,dx = x\ln x - x + C = x(\ln x - 1) + C \]

\[ \frac{d}{dx}\left(x\ln x - x\right) = \ln x + x\cdot\frac1x - 1 = \ln x \]

Figure (svg): The curve y equals ln x for x from 0.3 to 3.6, crossing the axis at x equals 1, with the region under it from x equals 1 to x equals e shaded; a dashed vertical line marks x equals e.

The antiderivative x ln x − x is 0 at e and −1 at 1, so the shaded area is exactly 1: the same number that defines e.

\[ \int_1^{e}\ln x\,dx = (e - e) - (0 - 1) = 1 \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 521 — rule (5.22), second line

Rule 5.22 includes the integral of the logarithm itself. You will derive it in Calculus II with integration by parts, but you can check it now with nothing more than the product rule.

Differentiate x ln x: the product rule gives ln x plus x times one over x, which is ln x plus 1. Subtracting the derivative of x removes that 1 and leaves ln x. So x ln x minus x is an antiderivative, and the book writes it as x times the quantity ln x minus 1.

The figure shows a lovely consequence. The area under ln x from 1 to e is exactly 1, since the antiderivative is zero at e and minus one at 1. It is a check you can keep in your head.

58. Reading the logarithm rule for any base

Notation

Annotate

On: \( \int \log_a x\,dx = \frac{x}{\ln a}\left(\ln x - 1\right) + C \)

  • Change of base: the log base a of x is ln x divided by ln a. So this is the ln x rule with a constant factor.
  • The constant one over ln a, pulled out front, times the x from the ln x rule.
  • The same bracket as the natural-log rule, x times (ln x minus 1).
  • Then ln a is 1 and the rule is exactly the integral of ln x.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 521 — rule (5.22), third line

Step through the notes. The whole rule comes from the change of base formula: the log base a of x is the natural log of x divided by ln a. That turns any logarithm into a constant times the natural one.

The constant one over ln a moves outside the integral, and the ln x rule does the rest. So the answer is x over ln a times the quantity ln x minus 1.

As with rule 5.21, base e is the special case where ln a is one and the correction disappears. Every base but e carries a factor of one over ln a, in both the exponential rule and the logarithm rule.

59. Example 5.47: the integral of log base 2

Worked example

Find the antiderivative of the log function.

\[ \int \log_2 x\,dx \]

Change the base

Why: Divide the natural logarithm by ln 2.

\[ \log_2 x = \frac{\ln x}{\ln 2} \]

Pull the constant out

Why: One over ln 2 is a number.

\[ \int \log_2 x\,dx = \frac{1}{\ln 2}\int \ln x\,dx \]

Use the ln x rule

Why: Rule (5.22).

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

Write it in the book's form

Why: Factor out x.

\[ \int \log_2 x\,dx = \frac{x}{\ln 2}\left(\ln x - 1\right) + C \]

Check by differentiating

Why: The product rule gives ln x plus 1 minus 1, divided by ln 2, which is log base 2 of x.

\[ \frac{1}{\ln 2}\left(\ln x + 1 - 1\right) = \frac{\ln x}{\ln 2} = \log_2 x \;\checkmark \]

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 523 — Example 5.47

The book simply reads the answer off the formula, but it is worth seeing where the formula comes from. Change the base first: log base 2 of x is ln x over ln 2.

Then the one over ln 2 is a constant factor, and the integral of ln x is x ln x minus x. Factoring out x gives the form in the book, x over ln 2 times the quantity ln x minus 1.

Differentiate to check: the bracket differentiates to ln x plus 1 minus 1, which is ln x, and dividing by ln 2 gives log base 2 of x. As a further numerical check, the formula gives about 3.672 for the integral from 1 to 4, and so does a direct estimate.

60. Checkpoint 5.39: log base 3

Fill the middle

Fill in the blanks

\int \log_3 x\,dx = \frac\ln 3___}\left(\ln x - 1\right) + C

Why: The log base 3 of x is ln x over ln 3, so the constant one over ln 3 comes out front of the ln x rule. Differentiating gives ln x over ln 3, which is the log base 3 of x again.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 523 — Checkpoint 5.39

Fill the blank before checking. It is the same procedure as Example 5.47 with 3 in place of 2.

Log base 3 of x is ln x over ln 3, so the constant one over ln 3 goes out front, and the answer is x over ln 3 times the quantity ln x minus 1. If you wrote ln 2, look again at which base the question uses; if you wrote 3, remember that the correction is always the natural logarithm of the base, not the base itself.

61. Putting it together

Section

Part 7

62. Pattern: integrating exponentials and logarithms

Pattern

Figure (svg): A flow diagram with three branches from the question: what shape is the integrand? An exponential leads to: let u be the exponent. A quotient leads to: let u be the denominator, giving ln of the absolute value of u. Another base leads to: rewrite as e to the x ln a. All three lead to a final box: check by differentiating.

Name the shape first; the shape names the substitution. Every branch ends with the same check.
  1. e to a power: let u be the exponent. Its derivative must be present, up to a constant.
  2. Any other base: rewrite as e to the x ln a, or use a to the x over ln a directly.
  3. A quotient: if the numerator is the derivative of the denominator (up to a constant), the answer is ln of the absolute value of the denominator.
  4. A definite integral: change the limits with u; if they come out backwards, keep the sign honest.
  5. Never integrate one over x across zero, and always differentiate your answer to check.

This is the whole lesson in one routine. Start by naming the shape of the integrand, because the shape tells you what u should be.

An exponential with e as the base: u is the exponent, provided its derivative is present up to a constant. Another base: rewrite in base e or use the formula with ln a. A fraction: check whether the top is the derivative of the bottom; if it is, the answer is a logarithm of the bottom's absolute value.

For a definite integral, convert the limits with the substitution and watch the sign when they come out backwards. Never integrate one over something across a point where the something is zero. And finish every problem the same way, by differentiating your answer.

63. Put the substitution steps in order

Ranking

Put in order

Order the steps for evaluating the integral of sin x/(1 + cos x) from 0 to π/2.

  1. Recognise the numerator as the derivative of the denominator, up to sign
  2. Set u = 1 + cos x and du = −sin x dx
  3. Convert the limits: x = 0 gives u = 2, x = π/2 gives u = 1
  4. Rewrite as −∫ from 2 to 1 of du/u, then swap the limits
  5. Evaluate ln|u| from 1 to 2 to get ln 2

Why: The shape comes first because it names the substitution. The limits must be converted before the integral is rewritten in u, and the sign is dealt with before evaluating.

Arrange the steps before checking. This is Example 5.48 taken apart, so you can see the order of decisions rather than just the order of lines.

Recognition comes first, because it tells you to substitute for the denominator. The limits have to be converted before you write the integral in u, since the u integral needs u limits. The sign from du and the backwards limits are dealt with together, and only then do you evaluate.

64. Check: an exponential with a composition

Check

Check your understanding

Which is the integral of x e^(−x²)?

  • A. −(1/2) e^(−x²) + C (correct)
  • B. −2 e^(−x²) + C
  • C. (x²/2) e^(−x²) + C
  • D. e^(−x² + 1)/(−x² + 1) + C

Answer: A

Why: Let u be minus x squared, so du is minus two x dx and x dx is minus one half du. The integral is minus one half of e to the u. Differentiating the answer gives minus one half times e to the minus x squared times minus two x, which is the integrand.

Why B tempts people
This multiplies by the derivative of the exponent instead of dividing by it; differentiating gives 4x times the exponential.
Why C tempts people
The x was used up by du; it should not be integrated separately and multiplied back in.
Why D tempts people
This applies the power rule to the exponent, which is never valid for e to a variable power.

Choose an answer, then differentiate it to confirm; that is the fastest way to settle any antiderivative question.

The exponent is minus x squared, with derivative minus two x, so x dx is minus a half du. The answer is minus one half of e to the minus x squared. The distractors are the three mistakes from Part 1: multiplying instead of dividing, integrating a factor that du already used, and the power rule on the exponent.

65. Check: the logarithmic form

Check

Check your understanding

Which integral equals ln|x³ − 4| + C?

  • A. ∫ 1/(x³ − 4) dx
  • B. ∫ 3x²/(x³ − 4) dx (correct)
  • C. ∫ x²/(x³ − 4) dx
  • D. ∫ 3x/(x³ − 4) dx

Answer: B

Why: The derivative of x cubed minus 4 is three x squared, and the logarithmic form needs exactly that on top. Differentiating ln of the absolute value of x cubed minus 4 gives three x squared over x cubed minus 4.

Why A tempts people
The numerator 1 is not the derivative of the denominator, and the missing x squared is not a constant, so this is not a logarithmic form.
Why C tempts people
Close: this integrates to one third of ln|x³ − 4|, because the numerator is a third of the derivative.
Why D tempts people
The numerator 3x is not a constant multiple of 3x², so the substitution does not clear it.

For each option, compare the numerator with the derivative of the denominator, which is three x squared.

Only one option has exactly three x squared on top. The option with x squared is close: it is a third of the derivative, so it integrates to a third of the logarithm, not the logarithm itself. The other two have numerators that are not constant multiples of three x squared at all.

66. Check: a definite integral with the reciprocal

Check

Check your understanding

What is the value of the integral of 1/x from −e to −1?

  • A. −1 (correct)
  • B. 1
  • C. It does not exist
  • D. 0

Answer: A

Why: The interval lies entirely on the negative side, so the integral exists. With the absolute value, ln of 1 minus ln of e is 0 minus 1, which is minus 1. The integrand is negative there, so a negative answer is expected.

Why B tempts people
That is the value from 1 to e; the mirror-image region below the axis gives the negative.
Why C tempts people
The integral fails only when the interval contains zero; this one does not.
Why D tempts people
Zero would need equal areas above and below the axis, but the whole region is below it.

Before calculating, ask whether the integral exists. The interval from minus e to minus 1 does not contain zero, so it does.

With the absolute value, the answer is ln 1 minus ln e, which is minus one. The sign makes sense: on the negative side one over x is negative, so the region lies below the axis. The mirror-image integral from 1 to e would be plus one.

67. Explain the absolute value

Explain it to yourself

Discussion prompt

In two or three sentences: why does the integral of one over x need an absolute value, and why does that absolute value still not let you integrate from minus 1 to 1?

Write your answer before revealing. Being able to explain this is the best evidence that the absolute value is not just a symbol you remember to write.

There are really two separate points. The absolute value is needed because one over x lives on both sides of zero while ln x lives only on the right; ln of the absolute value is the antiderivative on both branches. But the branches are separated by zero, where one over x blows up, so no single antiderivative can carry you across. Existence of a formula on each side is not the same as continuity across the gap.

68. Exit ticket

Exit ticket

\[ \int_1^{2} \frac{1}{x^3}e^{4x^{-2}}\,dx \]

Discussion prompt

Checkpoint 5.37. Choose u, convert the limits, and evaluate. Give the exact value and a decimal.

OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions §5.6, p. 521 — Checkpoint 5.37

This is Checkpoint 5.37, and it combines almost every idea in the lesson: a composition in the exponent, a constant to fix, limits to convert, and limits that run backwards.

The exponent is 4 over x squared, whose derivative is minus 8 over x cubed, so one over x cubed dx is minus an eighth of du. The limits become 4 and 1. The minus sign and the backwards limits cancel, leaving one eighth of the integral of e to the u from 1 to 4.

The exact value is e to the 4 minus e, all over 8, about 6.485, and a numerical estimate of the original integral agrees. If your answer was negative, look for a sign that was lost when the limits were swapped.

69. Recap

Recap

integrandantiderivativewatch for
eˣeˣ + Cits own antiderivative
e^(kx)(1/k) e^(kx) + Cdivide by the slope
aˣaˣ / ln a + Cdivide, never multiply, by ln a
1/xln|x| + Cabsolute value; never across 0
g′(x)/g(x)ln|g(x)| + Cnumerator must be the derivative
ln xx ln x − x + Ccheck with the product rule

\[ Q(b) = Q(a) + \int_a^{b} q(t)\,dt \]

Next, Section 5.7 finishes the chapter with the integrals that produce inverse trigonometric functions, the other place where one over x squared plus 1 turned up.

Stewart, Calculus: Early Transcendentals 8e, §5.5 The Substitution Rule §5.5, pp. 412-420 — the same material in Stewart

Six integrals, all of them derivatives read backwards. The exponential is its own antiderivative; a linear exponent divides by its slope; any other base divides by the logarithm of the base; one over x gives the logarithm of the absolute value; a fraction whose top is the derivative of its bottom gives the logarithm of the bottom; and ln x integrates to x ln x minus x.

The applications all rest on one idea, the net change theorem: the integral of a rate over an interval is the change in the quantity. Growth, decay, prices and populations are all that idea with an exponential rate.

Next, Section 5.7 handles the integrals that produce inverse trigonometric functions, including the one over x squared plus 1 that refused to be a logarithm in this lesson.

Sources

  1. OpenStax Calculus Volume 1, §5.6 Integrals Involving Exponential and Logarithmic Functions — Strang & Herman, OpenStax / Rice University, CC BY-NC-SA 4.0, pp. 516-525
  2. Stewart, Calculus: Early Transcendentals 8e, §5.5 The Substitution Rule — James Stewart, Cengage Learning, 2016, pp. 412-420
  3. Stewart, Calculus: Early Transcendentals 8e, §3.6 Derivatives of Logarithmic Functions — James Stewart, Cengage Learning, 2016, pp. 218-223

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

Book on Wyzant · Text (657) 465-8108