8 topics. Each one opens with its key idea, then the detail you need to use it.
1. The Chain Rule
Key idea: Outer derivative, inside unchanged, times inner derivative. Count the layers and multiply one factor per layer.
If h(x) = f(g(x)), then h′(x) = f′(g(x)) · g′(x). In words, you differentiate the outer function, leave the inner function untouched inside it, and then multiply by the derivative of the inner function. Three things must survive every time: the outer derivative is correct, the inside stays exactly as it was, and the inner derivative appears as a separate factor.
Work through d/dx[(x² + 5)⁷]. The outer function is a seventh power, so the power rule gives 7(x² + 5)⁶, with the inside left alone. Multiplying by the derivative of the inside, 2x, gives the final answer 14x(x² + 5)⁶. The same pattern runs through every basic form, so d/dx[cos(x³)] = −sin(x³) · 3x², d/dx[e^(x²)] = e^(x²) · 2x, and d/dx[ln(x² + 1)] = 2x/(x² + 1).
Watch for hidden layers. The expression sin³x means (sin x)³, so its derivative is 3sin²x · cos x. The function cos²(5x) has three layers, a square, a cosine and the linear inside 5x, so its derivative is 2cos(5x) · (−sin(5x)) · 5, which equals −10 sin(5x)cos(5x). A square root is a one-half power, which means d/dx[√(x² + 9)] simplifies to x/√(x² + 9).
The exam also tests the chain rule with tables of values. Suppose a table gives g(2) = 5, g′(2) = −3 and f′(5) = 4, and you need h′(2) for h(x) = f(g(x)). Then h′(2) = f′(g(2)) · g′(2) = 4(−3) = −12. Using f′(2) instead of f′(5) is the trap answer, because the outer derivative must be evaluated at the inside value, not at the original input.
2. Selecting Differentiation Procedures
Key idea: Name the last operation, rewrite if it removes a rule, then apply rules from the outside in.
Selecting differentiation procedures is a separate skill, and our bank tests it on 51 questions. The method is to identify the outermost operation first and only then choose the rule. A product of two non-constant pieces needs the product rule, so d/dx[x² sin x] = 2x sin x + x² cos x. A quotient needs the quotient rule, (bottom · top′ − top · bottom′)/bottom². For (x² + 1)/(x − 3), that formula gives (2x(x − 3) − (x² + 1))/(x − 3)², which simplifies to (x² − 6x − 1)/(x − 3)².
Often the fastest route is to rewrite the function before you differentiate it. The expression 3/x² is really 3x⁻², so its derivative is −6x⁻³ and the quotient rule is unnecessary. A fraction with a constant denominator, like (x² + 4x)/5, is just a polynomial divided by 5, so its derivative is (2x + 4)/5. Similarly, (x³ + 2x)/x simplifies to x² + 2 for x ≠ 0, ln(x³) becomes 3 ln x with derivative 3/x, and √x · x² combines into x^(5/2).
The rules also nest inside each other. For y = x² e^(3x), the product rule leads, but the second factor needs the chain rule, so y′ = 2x e^(3x) + x² · 3e^(3x). Apply the outer rule first and write each piece in brackets, then differentiate each bracket with its own rule. Most errors here come from simplifying too early, so keep the structure visible until the final line.
3. Derivatives of tan, cot, sec, csc
Key idea: Learn the four formulas, remember the "co" functions carry the minus sign, and expect a chain rule on top.
You already know that the derivative of sin x is cos x and the derivative of cos x is −sin x. The other four trigonometric derivatives come from writing each function as a quotient of sines and cosines. For tan x = sin x/cos x, the quotient rule gives (cos x · cos x − sin x · (−sin x))/cos²x. The numerator is cos²x + sin²x, which equals 1 by the Pythagorean identity, so the derivative is 1/cos²x = sec²x.
The same method produces the complete set of four formulas:
d/dx[tan x] = sec²x
d/dx[cot x] = −csc²x
d/dx[sec x] = sec x tan x
d/dx[csc x] = −csc x cot x
There is a sign pattern worth memorizing. Every function whose name begins with "co" (cosine, cotangent and cosecant) has a minus sign in front of its derivative formula. The pairs also mirror each other, since tangent goes with secant squared and cotangent goes with cosecant squared.
On the exam these formulas rarely appear alone, because they usually come wrapped in a chain rule or evaluated at a special angle. For example, d/dx[sec(2x)] = 2sec(2x)tan(2x), and d/dx[tan²x] = 2tan x · sec²x because tan²x means (tan x)². To find the slope of y = tan x at x = π/4, you compute sec²(π/4) = (√2)² = 2. That means you need your unit-circle values automatic: sec(π/3) = 2, csc(π/6) = 2, and tan(π/3) = √3.
4. Higher-Order Derivatives
Key idea: Differentiate one step at a time, keep every term, and plug in only after you reach the order the question asks for.
A higher-order derivative is what you get when you differentiate a derivative again. The second derivative f″ is the derivative of f′, and the third derivative f‴ is the derivative of f″. In other notation, f⁽⁴⁾(x) is the fourth derivative and d²y/dx² is the second derivative. The method is simple, but every additional step is another opportunity to drop a term.
Take f(x) = x⁵ − 4x³. Its derivatives are f′(x) = 5x⁴ − 12x², then f″(x) = 20x³ − 24x, and finally f‴(x) = 60x² − 24. Notice that −24x becomes the constant −24 rather than disappearing, because a linear term becomes a constant, and that constant only disappears on the next derivative. To find f″(1), you substitute into f″ and not into f′, which gives 20 − 24 = −4.
Negative powers change sign at each step. With y = x⁻², the first derivative is −2x⁻³ and the second derivative is (−2)(−3)x⁻⁴ = 6/x⁴. Two negatives multiply to a positive, and students who skip that line often choose the negative answer.
Trigonometric derivatives repeat every four steps: sin x, cos x, −sin x, −cos x, then sin x again. So the fourth, eighth and hundredth derivatives of sin x are all sin x. For the 27th derivative, divide 27 by 4 to get a remainder of 3, which makes the answer −cos x.
In motion problems, position s(t) gives velocity s′(t) and acceleration s″(t). If s(t) = t³ − 9t² + 15t, then the acceleration function is s″(t) = 6t − 18, so the acceleration at t = 5 is 12.
5. Implicit Differentiation
Key idea: Every y term gets a dy/dx. Collect, factor, divide, and use the original equation to simplify.
Implicit differentiation finds dy/dx when y is tangled inside an equation instead of written as y = f(x). You treat y as an unknown function of x and differentiate every term with respect to x. Any term containing y picks up a dy/dx factor from the chain rule, because y is acting as the inner function.
For the circle x² + y² = 25, differentiating both sides gives 2x + 2y(dy/dx) = 0, and solving for the derivative gives dy/dx = −x/y. At the point (3, 4) the slope is −3/4. The answer depends on both coordinates, which makes sense because the circle passes through x = 3 twice, at (3, 4) and at (3, −4), with opposite slopes.
A product such as xy needs the product rule, since both factors depend on x, so d/dx[xy] = y + x(dy/dx). For the equation x²y + y³ = 9, differentiating gives 2xy + x²(dy/dx) + 3y²(dy/dx) = 0. Collect every dy/dx term on one side, factor it out and divide, which gives dy/dx = −2xy/(x² + 3y²).
The exam also asks for the second derivative implicitly. Differentiate dy/dx = −x/y with the quotient rule to get d²y/dx² = −(y − x(dy/dx))/y², and then substitute −x/y for dy/dx. The expression simplifies to −(x² + y²)/y³, and because x² + y² = 25 everywhere on the circle, it equals −25/y³. At (3, 4) the value is −25/64. Substituting the original equation back in to simplify is a standard final step.
6. Derivatives of Inverse Functions and Inverse Trig Functions
Key idea: (f⁻¹)′(b) = 1/f′(a) where f(a) = b. Memorize the arcsin, arccos and arctan formulas and chain them.
If g is the inverse of f, then f(g(x)) = x for every x in the domain. Differentiating both sides with the chain rule gives f′(g(x)) · g′(x) = 1, so g′(x) = 1/f′(g(x)). Graphically, slopes at matching points are reciprocals, because the inverse is the reflection of f across the line y = x.
The exam usually gives you f and asks for the derivative of its inverse at a value b. The essential move is to find the input a with f(a) = b. As an illustration, let f(x) = x³ + x + 1 and suppose the question asks for (f⁻¹)′(3). Solving f(a) = 3 gives a = 1, and since f′(x) = 3x² + 1, you get f′(1) = 4 and (f⁻¹)′(3) = 1/4. The trap answer is 1/f′(3) = 1/28, which uses the wrong input.
The derivatives of inverse trig functions come from the same idea. If y = arctan x, then tan y = x, and differentiating both sides with respect to x gives sec²y · (dy/dx) = 1, so dy/dx = 1/sec²y = 1/(1 + tan²y) = 1/(1 + x²). These are the three formulas you need:
d/dx[arcsin x] = 1/√(1 − x²)
d/dx[arccos x] = −1/√(1 − x²)
d/dx[arctan x] = 1/(1 + x²)
When a chain rule sits on top, replace x with the inside function and multiply by its derivative. So d/dx[arctan(3x)] = 3/(1 + 9x²), and d/dx[arcsin(x²)] = 2x/√(1 − x⁴). Remember to square the entire inside expression, not just the variable.
7. Linear Approximation & Differentials
Key idea: L(x) = f(a) + f′(a)(x − a). Use the sign of f″ to say whether the estimate is too high or too low.
Close to x = a, a differentiable function looks like its tangent line. So the linearization L(x) = f(a) + f′(a)(x − a) gives a good estimate of f(x) whenever x is near a, and this is called linear approximation. To estimate √16.2, use f(x) = √x at a = 16, where f(16) = 4 and f′(16) = 1/(2√16) = 1/8. Then L(16.2) = 4 + (1/8)(0.2) = 4.025 (the true value is about 4.0249).
The exam almost always follows up by asking whether the estimate is an overestimate or an underestimate. The answer comes from concavity, which tells you which side of the tangent line the curve lies on. If f″ < 0 near a, the curve bends downward below its tangent line, so the tangent line overestimates, and that is why 4.025 is slightly too big for the concave-down square root. If f″ > 0, the tangent line underestimates. You must cite the sign of f″ to earn credit.
Differentials express the same idea in different notation. The differential dy = f′(x) dx estimates the change in y for a small change dx in x. If a cube's side is 10 centimeters and grows by 0.1 centimeters, then V = s³ and dV = 3s² ds = 3(100)(0.1) = 30 cubic centimeters, against an exact change of 30.301, so the estimate is accurate because ds is small.
With a table, if f(2) = 7 and f′(2) = −3, then f(2.1) ≈ 7 + (−3)(0.1) = 6.7. Build the tangent line at the given point closest to your input.
8. Mean Value Theorem
Key idea: Check the conditions, set f′(c) equal to the secant slope, and keep only c values strictly inside (a, b).
The Mean Value Theorem (MVT) says that if f is continuous on [a, b] and differentiable on (a, b), then at least one value c in (a, b) satisfies f′(c) = (f(b) − f(a))/(b − a). The right side is the slope of the secant line through the endpoints, so somewhere inside the interval the tangent line is parallel to that secant line. If you drive 120 miles in 2 hours, your speedometer must read exactly 60 miles per hour at some moment.
To find c, set f′(c) equal to the average rate of change and solve. For f(x) = x³ − x on [0, 2], the average rate is (6 − 0)/2 = 3. Setting 3c² − 1 = 3 gives c² = 4/3, so c = 2/√3, approximately 1.155. Reject any solution outside the open interval, such as −2/√3.
Both hypotheses matter. The function f(x) = |x| on [−1, 1] has an average rate of change of 0, but its slope is only ever −1 or 1. The theorem fails because |x| has a corner at 0, where it is not differentiable.
On free-response questions the theorem often comes with a table. If f(1) = 4 and f(5) = 12, you argue that f′(c) = (12 − 4)/(5 − 1) = 2 for some c in (1, 5). You only earn the point if you state the hypotheses: f is differentiable on [1, 5] (or for all x, as the stem gives it), so it is also continuous on [1, 5]. Differentiability on the open interval (1, 5) alone does not give continuity at the endpoints, so in that case you must state continuity on [1, 5] separately. Rolle's Theorem is the special case where f(a) = f(b), which forces f′(c) = 0.