Calculus › Section F · Logarithmic, exponential, and other transcendental functions › Topic F.1
The natural logarithm: derivatives
One integrand escaped every rule of Section E: 1/x. The natural logarithm is the function invented to fill that gap, and its definition by an integral hands us its derivative for free.
The missing antiderivative
When E.1 reversed the power rule, one case had to be skipped: no power of \(x\) differentiates to \(\frac{1}{x}\), because differentiation lowers the exponent by one, and the only candidate, \(x^0 = 1\), has derivative 0. Rather than leave the gap, we define the missing function directly. For \(x > 0\), let
the signed area under the hyperbola \(y = \frac{1}{t}\) from \(1\) to \(x\). Since \(\frac{1}{t}\) is continuous and positive there, every such integral exists. The natural logarithm is literally an accumulation function, built by Section E's machinery to answer Section E's own missing case.
First properties for free
The definition answers the basic questions without any further theory. At the anchor, \(\ln 1 = 0\), an integral over zero width. Because \(\frac{1}{t} > 0\), the function is always increasing, with \(\ln x > 0\) for \(x > 1\) and \(\ln x < 0\) when \(0 < x < 1\), where integration runs backwards. Far out, the area accumulates slowly but without bound, so \(\ln x \to \infty\) as \(x \to \infty\); as \(x \to 0^+\), \(\ln x \to -\infty\). The number \(e\) is defined as the value whose logarithm is \(1\); numerically \(e \approx 2.71828\).
The logarithm laws
The celebrated identities are consequences of the integral, and the cleanest route uses the derivative (proved next card). Consider \(\ln(ax)\) for a constant \(a > 0\): its derivative is \(\frac{a}{ax} = \frac{1}{x}\), the same as \(\ln x\)'s, so by D.2's corollary the two differ by a constant, evaluated at \(x = 1\) as \(\ln a\). Hence \(\ln(ax) = \ln a + \ln x\): the product law, \(\ln(ab) = \ln a + \ln b\). For the quotient law, \(\ln\frac{1}{b} + \ln b = \ln 1 = 0\), so \(\ln\frac{1}{b} = -\ln b\) and \(\ln\frac{a}{b} = \ln a - \ln b\). The power law, \(\ln(a^r) = r\ln a\), uses the same derivative trick: \(\ln(x^r)\) and \(r\ln x\) both have derivative \(\frac{r}{x}\) and both equal 0 at \(x = 1\). Logarithms turn multiplication into addition.
The derivative
Now the payoff, and it is one line of the fundamental theorem. Since \(\ln x\) is an accumulation function of the continuous integrand \(\frac{1}{t}\), FTC Part 1 (E.4) says its derivative is the integrand evaluated at the moving edge:
For negative inputs, \(\ln|x|\) works: when \(x < 0\), \(\ln(-x)\) differentiates to \(\frac{-1}{-x} = \frac{1}{x}\) by the chain rule. So \(\frac{d}{dx}\ln|x| = \frac{1}{x}\) for all \(x \ne 0\), which is why antiderivatives of \(\frac{1}{x}\) were always going to need absolute values.
Worked example: the chain form
Differentiate \(g(x) = \ln(x^2 + 1)\). By the chain rule with inner \(u = x^2 + 1\): \(g'(x) = \frac{1}{u} \cdot u' = \frac{2x}{x^2 + 1}\). The pattern \(\frac{d}{dx}\ln u = \frac{u'}{u}\) is worth memorising as its own rule. A second flavour: differentiate \(h(x) = \ln|\sin x|\). The inner derivative is \(\cos x\), so \(h'(x) = \frac{\cos x}{\sin x} = \cot x\) wherever the sine is nonzero. In both, the logarithm differentiates to "derivative of the inside over the inside", no matter how tangled the inside.
Simplify first with the log laws
A quotient or product inside a logarithm is begging to be split before you differentiate. To differentiate \(f(x) = \ln\!\left(\frac{x^2 + 1}{3x}\right)\), first apply the laws: \(f(x) = \ln(x^2 + 1) - \ln(3x)\). Now differentiate term by term: \(f'(x) = \frac{2x}{x^2 + 1} - \frac{3}{3x} = \frac{2x}{x^2 + 1} - \frac{1}{x}\). The chain rule with the quotient rule gives the same answer, but only after much more algebra. Same for powers: \(\ln\sqrt{x} = \frac{1}{2}\ln x\), whose derivative is \(\frac{1}{2x}\) immediately. The log laws exist to be used before the calculus, not after.
Worked example: where is ln defined?
What is the domain of \(k(x) = \ln(3x - 6)\)? The logarithm's definition only ever integrates over positive territory, so the argument must satisfy \(3x - 6 > 0\), giving \(x > 2\). At \(x = 3\) we can sanity-check: \(k(3) = \ln 3 \approx 1.0986\), since the area under \(\frac{1}{t}\) from \(1\) to \(3\) must be a positive number around \(1\). Always find the domain first: a derivative formula is only valid where the original logarithm is defined.
The graph, and why logs rule science
The graph of \(\ln\) rises through \((1, 0)\), always increasing (slope \(\frac{1}{x} > 0\)) and always concave down (\(\frac{d}{dx}\frac{1}{x} = -\frac{1}{x^2} < 0\)), growing ever more slowly. That slow, decelerating growth is exactly what logarithmic scales exploit. The sound intensities we hear span about thirteen powers of ten, and the decibel scale, a logarithm, compresses that span to roughly 0 to 130; the Richter magnitude does the same for earthquake ground-shaking amplitude, and pH for acidity. (These scales use base-10 logarithms, met in F.5, but the shape is the same.) When a quantity is naturally multiplicative, doublings and ten-foldings, a logarithm turns it into tidy equal steps, which is why so many scientific dials read in logs.
Common mistakes to avoid
The recurring slips: differentiating \(\ln g(x)\) as \(\frac{1}{g(x)}\) and dropping the \(g'\) chain factor, the rule is \(\frac{g'}{g}\); believing \(\frac{d}{dx}\ln x = \ln\) of something, it is \(\frac{1}{x}\), full stop; applying the log laws to sums, \(\ln(a + b)\) is NOT \(\ln a + \ln b\), the laws cover products, quotients and powers only; writing \((\ln x)^2\) as \(\ln x^2\), the first is a power of a log, the second is \(2\ln x\); and forgetting the domain, so log-of-negative expressions sneak through. Each law should be checkable on numbers in one second, for instance \(\ln 8 = 3\ln 2\).
Key takeaways
\(\ln x = \int_1^x \frac{1}{t}\,dt\) fills the last gap in the reverse power rule, and the fundamental theorem immediately gives \(\frac{d}{dx}\ln x = \frac{1}{x}\), so the table closes: the antiderivative of \(\frac{1}{x}\) is \(\ln|x| + C\). The laws, \(\ln(ab) = \ln a + \ln b\), \(\ln(a^r) = r\ln a\), quotients to differences, tame any expression before differentiating via \(\frac{d}{dx}\ln u = \frac{u'}{u}\). The domain is positive arguments only, the graph rises while flattening for ever, and the scale-compressing property explains decibels, Richter and pH. The exponential is waiting as the inverse of all this, in F.4.
Flashcards
Flashcard
∫ from 1 to x of (1/t) dt, for x > 0
Test your understanding
Question 1 of 5
What is the derivative of ln x?
Question 2 of 5
What is the derivative of ln(x² + 1)?
Question 3 of 5
What is ln(e³)?
Question 4 of 5
Which expression equals ln(x⁵) − ln(x²) for x > 0?
Question 5 of 5
What is the domain of ln(3x − 6)?
Standards: 3
