Skip to content

Unit 2: Differentiation: Definition and Fundamental Properties

AP Calc cheatsheet

Open ↗

Loading…

How do we measure the slope of a tangent line? Before talking about tangent lines, start with something we already know how to measure: the slope of a secant line. If a curve y=f(x)y=f(x) passes through the points x=ax=a and x=bx=b, then the secant slope is

f(b)−f(a)b−a.\frac{f(b)-f(a)}{b-a}.

This is the average rate of change on [a,b][a,b]. Geometrically, it is the slope of the line connecting two points on the graph.

12312345secantx=atangentxy

To get the slope of the tangent line at x=ax=a, move the second point closer and closer to aa. Suppose that we want the tangent slope at point P=(a,f(a))P = (a, f(a)) on f(x)f(x). Define a point Q=(x,f(x))Q = (x, f(x)) also on f(x)f(x) and connect the secant line PQPQ. The secant line rotates toward the tangent line, so the tangent slope is the limiting value of those secant slopes:

mtangent=lim⁡Q→PmPQ.m_{\text{tangent}} = \lim_{Q\to P}m_{PQ}.

The slope of the secant line through PP and QQ is

mPQ=f(x)−f(a)x−a.m_{PQ} = \frac{f(x)-f(a)}{x-a}.

When Q→PQ\to P along the graph, its xx-coordinate approaches the xx-coordinate of PP. In other words, x→ax\to a. Therefore,

mtangent at x=a=lim⁡x→af(x)−f(a)x−a.m_{\text{tangent at }x=a} = \lim_{x\to a}\frac{f(x)-f(a)}{x-a}.

The denominator measures horizontal change (the “run” of the slope formula), and the numerator measures vertical change (the “rise” of the slope formula). The limit asks what that ratio becomes when the two points collapse into one point.

Equivalently, instead of naming the nearby input xx, write it as a+ha+h, where hh is the horizontal change from aa. Then the secant slope becomes

f(a+h)−f(a)h.\frac{f(a+h)-f(a)}{h}.

This is known as the difference quotient. As the second point moves toward aa, the horizontal change hh moves toward 00, so the tangent slope is

lim⁡h→0f(a+h)−f(a)h.\lim_{h\to0}\frac{f(a+h)-f(a)}{h}.

The two formulas are equivalent because x=a+hx=a+h, so h=x−ah=x-a.

Example. For f(x)=x2+1f(x)=x^2+1, find the slope of the tangent line at x=3x=3 using the first method.

Use a=3a=3:

lim⁡x→3f(x)−f(3)x−3.\lim_{x\to3}\frac{f(x)-f(3)}{x-3}.

Since f(3)=10f(3)=10,

lim⁡x→3x2+1−10x−3=lim⁡x→3x2−9x−3.\lim_{x\to3}\frac{x^2+1-10}{x-3} = \lim_{x\to3}\frac{x^2-9}{x-3}.

Factor and cancel:

lim⁡x→3(x−3)(x+3)x−3=lim⁡x→3(x+3)=6.\lim_{x\to3}\frac{(x-3)(x+3)}{x-3} = \lim_{x\to3}(x+3) = 6.

So the tangent slope at x=3x=3 is 66.

Example. For f(x)=2x2−xf(x)=2x^2-x, find the slope of the tangent line at x=1x=1 using the second method.

Start with

lim⁡h→0f(1+h)−f(1)h.\lim_{h\to0}\frac{f(1+h)-f(1)}{h}.

Compute the two function values:

f(1+h)=2(1+h)2−(1+h)=1+3h+2h2,f(1+h)=2(1+h)^2-(1+h) =1+3h+2h^2,

and

f(1)=1.f(1)=1.

Then

lim⁡h→0f(1+h)−f(1)h=lim⁡h→01+3h+2h2−1h.\lim_{h\to0}\frac{f(1+h)-f(1)}{h} = \lim_{h\to0}\frac{1+3h+2h^2-1}{h}.

Simplify:

lim⁡h→03h+2h2h=lim⁡h→0(3+2h)=3.\lim_{h\to0}\frac{3h+2h^2}{h} = \lim_{h\to0}(3+2h) = 3.

So the tangent slope at x=1x=1 is 33.

Definition. The derivative of ff at xx is defined as

f′(x)=lim⁡h→0f(x+h)−f(x)hf'(x) = \lim_{h \to 0} \frac{f(x+h)-f(x)}{h}

Equivalently, for a derivative at x=ax=a, the derivative becomes the tangent line at aa:

f′(a)=lim⁡x→af(x)−f(a)x−a.f'(a) = \lim_{x \to a} \frac{f(x)-f(a)}{x-a}.

There are many interpretations of the derivative:

  • instantaneous rate of change,
  • slope of the tangent line,
  • limit of secant slopes,
  • local sensitivity of output to input.

Example. Use the limit definition to find f′(x)f'(x) to find the derivative for f(x)=2xf(x)=\frac{2}{x}.

Start with the hh form:

f′(x)=lim⁡h→0f(x+h)−f(x)h.f'(x) = \lim_{h\to0}\frac{f(x+h)-f(x)}{h}.

Substitute f(x)=2/xf(x)=2/x:

f′(x)=lim⁡h→02x+h−2xh.f'(x) = \lim_{h\to0}\frac{\frac{2}{x+h}-\frac{2}{x}}{h}.

Combine the fractions in the numerator:

2x+h−2x=2x−2(x+h)x(x+h)=−2hx(x+h).\frac{2}{x+h}-\frac{2}{x} = \frac{2x-2(x+h)}{x(x+h)} = \frac{-2h}{x(x+h)}.

So

f′(x)=lim⁡h→0−2hx(x+h)h.f'(x) = \lim_{h\to0}\frac{\frac{-2h}{x(x+h)}}{h}.

Dividing by hh is the same as multiplying by 1/h1/h:

f′(x)=lim⁡h→0−2hx(x+h)⋅1h.f'(x) = \lim_{h\to0}\frac{-2h}{x(x+h)}\cdot\frac{1}{h}.

Cancel hh:

f′(x)=lim⁡h→0−2x(x+h).f'(x) = \lim_{h\to0}\frac{-2}{x(x+h)}.

Now let h→0h\to0:

f′(x)=−2x2.f'(x) = \frac{-2}{x^2}.

Therefore,

ddx(2x)=−2x2.\frac{d}{dx}\left(\frac{2}{x}\right) = -\frac{2}{x^2}.

This formula is valid for x≠0x\ne0, because the original function is not defined at x=0x=0. Note that ddx\frac{d}{dx} is a notation for the derivative.

There exists many different notations for derivatives, but the most common ones are:

f′(x),y′,dydx,ddx[f(x)].f'(x),\qquad y',\qquad \frac{dy}{dx},\qquad \frac{d}{dx}[f(x)].

They all refer to rate of change, but they emphasize different things. The notation f′(a)f'(a) is a number. The notation f′(x)f'(x) is a function (as long as xx is the variable). The notation dy/dxdy/dx emphasizes that the derivative compares a tiny change in yy to a tiny change in xx.

When a problem asks for “the derivative at x=ax=a,” give a value. When it asks for “the derivative of ff,” give a formula.

If f(x)f(x) has units of output and xx has units of input, then f′(x)f'(x)

has units

output unitsinput units.\frac{\text{output units}}{\text{input units}}.

For instance, if position is measured in feet and time in seconds, velocity is measured in feet per second. The sign tells direction; the magnitude tells how fast the position is changing.

The Leibniz Notation (dydx\frac{dy}{dx}) is especially useful for problems that require context. For example, if s(t)s(t) is position, then s′(t)s'(t) and ds/dtds/dt both describe velocity. The notation ds/dtds/dt makes units especially clear because it literally compares a change in position to a change in time.

For a quantity QQ depending on another quantity xx: dQdx\frac{dQ}{dx} has units of QQ-units per xx-unit. This unit check is one of the fastest ways to catch an interpretation error.

Definition. A function is said to be differentiable on an interval [a,b][a,b] if it’s derivative exists in all of its domain.

The domain of f′f' can be smaller than the domain of ff. Even if f(a)f(a) exists, the derivative at aa may fail to exist because the nearby slopes do not settle into one finite value or diverge to infinity.

When finding a derivative formula, always ask where that formula is valid. For example,

f(x)=xf(x)=\sqrt{x}

has domain [0,∞)[0,\infty), but it’s derivative

f′(x)=12xf'(x)=\frac{1}{2\sqrt{x}}

is valid only for x>0x>0. The original function exists at x=0x=0, but the tangent there is vertical, so the ordinary derivative is not finite.

From a graph, differentiability fails at places where the tangent slope is not a single finite number.

Theorem. If ff is differentiable at x=ax=a, then ff is continuous at x=ax=a.

Proof (Differentiability implies continuity). Suppose f′(a)f'(a) exists. For x≠ax\ne a, rewrite the change in the function as

f(x)−f(a)=f(x)−f(a)x−a(x−a).f(x)-f(a) =\frac{f(x)-f(a)}{x-a}(x-a).

As x→ax\to a, the first factor approaches f′(a)f'(a) and the second factor approaches 00. Therefore,

lim⁡x→a(f(x)−f(a))=f′(a)⋅0=0.\lim_{x\to a}\big(f(x)-f(a)\big) =f'(a)\cdot0 =0.

It follows that

lim⁡x→af(x)=f(a),\lim_{x\to a}f(x)=f(a),

which is exactly the definition of continuity at aa.

The contrapositive is often more useful: if ff is not continuous at aa, then it cannot be differentiable there. The converse is false. A function may be continuous at a point without being differentiable at that point.

The most common continuous-but-nondifferentiable behaviors are:

  • Corner: the one-sided derivatives are finite but unequal, as with f(x)=∣x∣f(x)=\lvert x\rvert at x=0x=0.
  • Cusp: the one-sided slopes become infinite with opposite signs, as with f(x)=x2/3f(x)=x^{2/3} at x=0x=0.
  • Vertical tangent: the slopes become infinite with the same sign, as with f(x)=x1/3f(x)=x^{1/3} at x=0x=0.

All three graphs remain unbroken at the point, so continuity alone does not guarantee differentiability.


For constants cc and differentiable functions f,gf,g:

ddx(c)=0\frac{d}{dx}(c) = 0 ddx(xn)=nxn−1\frac{d}{dx}(x^n) = nx^{n-1} ddx[cf(x)]=cf′(x)\frac{d}{dx}[cf(x)] = cf'(x) ddx[f(x)±g(x)]=f′(x)±g′(x)\frac{d}{dx}[f(x) \pm g(x)] = f'(x) \pm g'(x) ddx[f(x)g(x)]=f′(x)g(x)+f(x)g′(x)\frac{d}{dx}[f(x)g(x)] = f'(x)g(x) + f(x)g'(x) ddx[f(x)g(x)]=f′(x)g(x)−f(x)g′(x)[g(x)]2g(x)≠0.\frac{d}{dx}\left[\frac{f(x)}{g(x)}\right] = \frac{f'(x)g(x)-f(x)g'(x)}{[g(x)]^2} \qquad g(x) \ne 0.

The theorems can all be proven by just plugging in the parent function into the derivative definition, so I will only show one example proof below.

Proof (Product Rule). Let H(x)=f(x)⋅g(x)H(x)=f(x) \cdot g(x). Find dHdx\frac{dH}{dx}.

Start from the derivative definition:

ddxH(x)=lim⁡h→0H(x+h)−H(x)h.\frac{d}{dx}H(x) = \lim_{h\to0}\frac{H(x+h)-H(x)}{h}.

Resubstituting:

ddxH(x)=lim⁡h→0f(x+h)g(x+h)−f(x)g(x)h.\frac{d}{dx}H(x) = \lim_{h\to0}\frac{f(x+h)g(x+h)-f(x)g(x)}{h}.

Now, do an algebra trick: We can add and then subtract f(x+h)g(x)f(x+h)g(x) to our expression to factor:

ddxH(x)=lim⁡h→0f(x+h)g(x+h)−f(x+h)g(x)+f(x+h)g(x)−f(x)g(x)h\frac{d}{dx}H(x) = \lim_{h\to0} \frac{ f(x+h)g(x+h)-f(x+h)g(x) +f(x+h)g(x)-f(x)g(x) }{h}

Split the fraction into two limits:

ddxH(x)=lim⁡h→0f(x+h)g(x+h)−g(x)h+lim⁡h→0g(x)f(x+h)−f(x)h.\frac{d}{dx}H(x) = \lim_{h\to0}f(x+h)\frac{g(x+h)-g(x)}{h} + \lim_{h\to0}g(x)\frac{f(x+h)-f(x)}{h}.

Now split off the limits that become derivative definitions:

ddxH(x)=(lim⁡h→0f(x+h))g′(x)+g(x)f′(x).\frac{d}{dx}H(x) = \left(\lim_{h\to0}f(x+h)\right)g'(x) +g(x)f'(x).

This uses the sum rule for limits and the fact that g(x)g(x) is constant with respect to hh.

lim⁡h→0g(x+h)−g(x)h=g′(x),lim⁡h→0f(x+h)−f(x)h=f′(x).\lim_{h\to0}\frac{g(x+h)-g(x)}{h}=g'(x), \qquad \lim_{h\to0}\frac{f(x+h)-f(x)}{h}=f'(x).

Because ff is differentiable at xx, it is also continuous at xx. Therefore

lim⁡h→0f(x+h)=f(x).\lim_{h\to0}f(x+h)=f(x).

Substitute these limits back in:

ddxH(x)=f(x)g′(x)+g(x)f′(x).\frac{d}{dx}H(x) =f(x)g'(x)+g(x)f'(x).

Since multiplication is commutative, this is usually written as

ddx[f(x)g(x)]=f′(x)g(x)+f(x)g′(x).\frac{d}{dx}[f(x)g(x)]=f'(x)g(x)+f(x)g'(x).

Note that if you have more than two functions, the procedure is the same, giving you a result where you sum up the multiplicative terms between one derivative and the other functions for each derivative.

Example. Differentiate y=(x2+1)(x3−4x)y=(x^2+1)(x^3-4x) using the product rule.

Let f(x)=x2+1f(x)=x^2+1 and g(x)=x3−4xg(x)=x^3-4x, so f′(x)=2xf'(x)=2x and g′(x)=3x2−4g'(x)=3x^2-4. The product rule gives

y′=f′(x)g(x)+f(x)g′(x).y' = f'(x)g(x) + f(x)g'(x).

Substitute the pieces:

y′=2x(x3−4x)+(x2+1)(3x2−4).y'=2x(x^3-4x)+(x^2+1)(3x^2-4).

Simplifying gives

y′=5x4−9x2−4.y'=5x^4-9x^2-4.

Example. Differentiate y=x2x+1\displaystyle y = \frac{x^2}{x+1}.

Let f(x)=x2f(x)=x^2 and g(x)=x+1g(x)=x+1, so f′(x)=2xf'(x)=2x and g′(x)=1g'(x)=1. The quotient rule gives

y′=f′(x)g(x)−f(x)g′(x)[g(x)]2.y' = \frac{f'(x)g(x)-f(x)g'(x)}{[g(x)]^2}.

Substitute the pieces:

y′=2x(x+1)−x2(1)(x+1)2.y' = \frac{2x(x+1)-x^2(1)}{(x+1)^2}.

Expand and simplify the numerator:

2x(x+1)−x2=2x2+2x−x2=x2+2x.2x(x+1)-x^2 = 2x^2+2x-x^2 = x^2+2x.

Therefore

y′=x2+2x(x+1)2.y' = \frac{x^2+2x}{(x+1)^2}.

ddx(sin⁡x)=cos⁡x\frac{d}{dx}(\sin x) = \cos x ddx(cos⁡x)=−sin⁡x\frac{d}{dx}(\cos x) = -\sin x ddx(tan⁡x)=sec⁡2x\frac{d}{dx}(\tan x) = \sec^2 x ddx(sec⁡x)=sec⁡xtan⁡x,\frac{d}{dx}(\sec x)=\sec x\tan x, ddx(csc⁡x)=−csc⁡xcot⁡x,\frac{d}{dx}(\csc x)=-\csc x\cot x, ddx(cot⁡x)=−csc⁡2x.\frac{d}{dx}(\cot x)=-\csc^2 x.

All of the trig derivatives can be proved using the derivative definition and the quotient rule, so we will only use a couple examples.

Proof (Derivative of tanxtan x). Start with the identity

tan⁡x=sin⁡xcos⁡x.\tan x=\frac{\sin x}{\cos x}.

Use the quotient rule:

ddx(tan⁡x)=(ddxsin⁡x)(cos⁡x)−(sin⁡x)(ddxcos⁡x)cos⁡2x.\frac{d}{dx}(\tan x) = \frac{(\frac{d}{dx} \sin x)(\cos x)-(\sin x)(\frac{d}{dx} \cos x)}{\cos^2 x}.

For sine, start with the derivative definition:

ddx(sin⁡x)=lim⁡h→0sin⁡(x+h)−sin⁡xh.\frac{d}{dx}(\sin x) = \lim_{h\to0}\frac{\sin(x+h)-\sin x}{h}.

Use sin⁡(x+h)=sin⁡xcos⁡h+cos⁡xsin⁡h\sin(x+h)=\sin x\cos h+\cos x\sin h:

sin⁡(x+h)−sin⁡xh=sin⁡xcos⁡h−1h+cos⁡xsin⁡hh.\frac{\sin(x+h)-\sin x}{h} = \sin x\frac{\cos h-1}{h} +\cos x\frac{\sin h}{h}.

Since

lim⁡h→0sin⁡hh=1,lim⁡h→0cos⁡h−1h=0,\lim_{h\to0}\frac{\sin h}{h}=1, \qquad \lim_{h\to0}\frac{\cos h-1}{h}=0,

we get

ddx(sin⁡x)=cos⁡x.\frac{d}{dx}(\sin x)=\cos x.

For cosine, use cos⁡(x+h)=cos⁡xcos⁡h−sin⁡xsin⁡h\cos(x+h)=\cos x\cos h-\sin x\sin h:

cos⁡(x+h)−cos⁡xh=cos⁡xcos⁡h−1h−sin⁡xsin⁡hh.\frac{\cos(x+h)-\cos x}{h} = \cos x\frac{\cos h-1}{h} -\sin x\frac{\sin h}{h}.

Taking the same two standard trig limits gives

ddx(cos⁡x)=−sin⁡x.\frac{d}{dx}(\cos x)=-\sin x.

Thus,

ddx(tan⁡x)=(cos⁡x)(cos⁡x)−(sin⁡x)(−sin⁡x)cos⁡2x.\frac{d}{dx}(\tan x) = \frac{(\cos x)(\cos x)-(\sin x)(-\sin x)}{\cos^2 x}.

Simplify the numerator:

ddx(tan⁡x)=cos⁡2x+sin⁡2xcos⁡2x=1cos⁡2x=sec⁡2x.\frac{d}{dx}(\tan x) = \frac{\cos^2 x+\sin^2 x}{\cos^2 x} = \frac{1}{\cos^2 x} = \sec^2 x.

Derivatives of exponential and logarithmic functions

Section titled “Derivatives of exponential and logarithmic functions”

Exponential functions are special because their rate of change is proportional to their current value. For axa^x, the derivative is some constant multiple of axa^x:

ddx(ax)=axln⁡a.\frac{d}{dx}(a^x)=a^x\ln a.

The base ee is the unique positive base where that constant is 11. In other words, exe^x is the exponential function whose derivative is exactly itself. Logarithm derivatives come from the fact that logarithms are inverse functions of exponentials.

ddx(ex)=ex\frac{d}{dx}(e^x) = e^x ddx(ax)=axln⁡a\frac{d}{dx}(a^x) = a^x \ln a ddx(ln⁡x)=1x\frac{d}{dx}(\ln x) = \frac{1}{x} ddx(log⁡ax)=1xln⁡a,a>0, a≠1.\frac{d}{dx}(\log_a x)=\frac{1}{x\ln a}, \qquad a>0,\ a\ne 1.

The formulas for general exponential and logarithmic functions can be derived later using the chain rule and derivatives of inverse functions. The most important starting point is the derivative of exe^x.

Proof (Derivative of exe^x). Start from the derivative definition:

ddx(ex)=lim⁡h→0ex+h−exh.\frac{d}{dx}(e^x) =\lim_{h\to0}\frac{e^{x+h}-e^x}{h}.

Using ex+h=exehe^{x+h}=e^xe^h, factor out the term that does not depend on hh:

ddx(ex)=exlim⁡h→0eh−1h.\frac{d}{dx}(e^x) =e^x\lim_{h\to0}\frac{e^h-1}{h}.

The base ee is defined so that

lim⁡h→0eh−1h=1.\lim_{h\to0}\frac{e^h-1}{h}=1.

Therefore,

ddx(ex)=ex.\frac{d}{dx}(e^x)=e^x.

As a reminder, we define

sinh⁡x=ex−e−x2,cosh⁡x=ex+e−x2.\sinh x=\frac{e^x-e^{-x}}{2}, \qquad \cosh x=\frac{e^x+e^{-x}}{2}.

The other hyperbolic functions are defined similarly to trig functions (e.g. tanh⁡x=sinh⁡xcosh⁡x\tanh x = \frac{\sinh x}{\cosh x}). Their main derivatives are:

ddx(sinh⁡x)=cosh⁡x,ddx(cosh⁡x)=sinh⁡x,\frac{d}{dx}(\sinh x)=\cosh x, \qquad \frac{d}{dx}(\cosh x)=\sinh x, ddx(tanh⁡x)=sech⁡2x.\frac{d}{dx}(\tanh x)=\operatorname{sech}^2 x. ddx(sech⁡x)=−sech⁡xtanh⁡x,\frac{d}{dx}(\operatorname{sech} x)=-\operatorname{sech} x\tanh x, ddx(csch⁡x)=−csch⁡xcoth⁡x,\frac{d}{dx}(\operatorname{csch} x)=-\operatorname{csch} x\coth x, ddx(coth⁡x)=−csch⁡2x.\frac{d}{dx}(\coth x)=-\operatorname{csch}^2 x.

Since all of the hyperbolic functions can be defined in terms of exponentials, the derivative formula is just repeated derivatives of exponentials. The proofs are left to the reader as an exercise.

Example. Differentiate y=4ex−3cos⁡x+2x5y = 4e^x - 3\cos x + 2x^5.

Differentiate each term separately using the sum rule. The exponential is its own derivative, the derivative of cos⁡x\cos x is −sin⁡x-\sin x, and the power rule applies to x5x^5:

ddx(4ex)=4ex,ddx(−3cos⁡x)=3sin⁡x,ddx(2x5)=10x4.\frac{d}{dx}(4e^x) = 4e^x,\qquad \frac{d}{dx}(-3\cos x) = 3\sin x,\qquad \frac{d}{dx}(2x^5) = 10x^4.

Combining the terms gives

y′=4ex+3sin⁡x+10x4.y' = 4e^x + 3\sin x + 10x^4.

Example. Differentiate

y=5log⁡2x⋅3sec⁡x+4sinh⁡x.y=5\log_2 x \cdot 3\sec x+4\sinh x.

This is a product plus one hyperbolic term. First rewrite the product with the constant out front:

y=15(log⁡2x)(sec⁡x)+4sinh⁡x.y=15(\log_2 x)(\sec x)+4\sinh x.

Use the product rule on (log⁡2x)(sec⁡x)(\log_2 x)(\sec x):

ddx[(log⁡2x)(sec⁡x)]=1xln⁡2sec⁡x+(log⁡2x)sec⁡xtan⁡x.\frac{d}{dx}\left[(\log_2 x)(\sec x)\right] = \frac{1}{x\ln 2}\sec x + (\log_2 x)\sec x\tan x.

Also,

ddx(4sinh⁡x)=4cosh⁡x.\frac{d}{dx}(4\sinh x)=4\cosh x.

Therefore

y′=15(sec⁡xxln⁡2+(log⁡2x)sec⁡xtan⁡x)+4cosh⁡x.y'=15\left(\frac{\sec x}{x\ln 2}+(\log_2 x)\sec x\tan x\right)+4\cosh x.

Most derivative problems are not about one isolated rule. They are about choosing the order in which rules apply.

Example. Differentiate

y=x2exx+1.y=\frac{x^2e^x}{\sqrt{x+1}}.

This can be treated as a quotient, but rewriting the radical as a power makes the product structure easier:

y=x2ex(x+1)−1/2.y=x^2e^x(x+1)^{-1/2}.

This rewrite uses the exponent rule x+1=(x+1)1/2\sqrt{x+1}=(x+1)^{1/2}, so dividing by x+1\sqrt{x+1} is the same as multiplying by (x+1)−1/2(x+1)^{-1/2}. Now the function is a product of three factors instead of a quotient with a radical.

This is a product of three factors. The expanded product rule says to differentiate one factor at a time and leave the other factors alone. For three factors,

(uvw)′=u′vw+uv′w+uvw′.(uvw)'=u'vw+uv'w+uvw'.

Here,

u=x2,v=ex,w=(x+1)−1/2.u=x^2,\qquad v=e^x,\qquad w=(x+1)^{-1/2}.

Use the product rule in expanded form:

y′=(2x)ex(x+1)−1/2+x2(ex)(x+1)−1/2+x2ex(−12)(x+1)−3/2.y'=(2x)e^x(x+1)^{-1/2} +x^2(e^x)(x+1)^{-1/2} +x^2e^x\left(-\frac12\right)(x+1)^{-3/2}.

Factor the common term xex(x+1)−3/2xe^x(x+1)^{-3/2}:

y′=xex(x+1)−3/2[2(x+1)+x(x+1)−x2].y'=xe^x(x+1)^{-3/2} \left[2(x+1)+x(x+1)-\frac{x}{2}\right].

This form shows the structure clearly. If desired, it can be combined into one rational expression, but the important part is choosing the rule order correctly.


At x=ax=a:

  • tangent slope is f′(a)f'(a),
  • tangent line is
y−f(a)=f′(a)(x−a),y - f(a) = f'(a)(x-a),
  • normal slope is −1/f′(a)-1/f'(a) when f′(a)≠0f'(a) \ne 0.

Example. Find the equation of the tangent line to f(x)=x3+exf(x)=x^3+e^x at x=2x=2.

First find the point on the curve:

f(2)=23+e2=8+e2,f(2)=2^3+e^2=8+e^2,

so the point of tangency is (2,8+e2)(2,8+e^2). Next find the slope from the derivative:

f′(x)=3x2+ex,f′(2)=3(2)2+e2=12+e2.f'(x)=3x^2+e^x,\qquad f'(2)=3(2)^2+e^2=12+e^2.

Use point-slope form with slope 12+e212+e^2 at (2,8+e2)(2,8+e^2):

y−(8+e2)=(12+e2)(x−2).y-(8+e^2) = (12+e^2)(x-2).

Simplifying gives the tangent line

y=(12+e2)x−(16+3e2).y = (12+e^2)x-(16+3e^2).

The second derivative f′′(x)f''(x) measures the rate of change of the first derivative.

The second derivative has many useful interpretations:

  • concavity (whether a graph is opening up or down) in pure math
  • acceleration when ff is position or angular frequency when ff is potential in physics

You may also see f(n)(x)f^{(n)}(x) for the nnth derivative. Basically, higher order derivatives just means take the derivative of a function nn times.

Example. Find f′′(x)f''(x) for f(x)=x4−5x2+3xf(x)=x^4-5x^2+3x.

Differentiate once using the power rule term by term:

f′(x)=4x3−10x+3.f'(x)=4x^3-10x+3.

Differentiate again to get the second derivative:

f′′(x)=12x2−10.f''(x)=12x^2-10.

Alternate notation for higher-order derivatives

Section titled “Alternate notation for higher-order derivatives”

Higher derivatives have several common notations. If y=f(x)y=f(x), then:

OrderPrime notationLeibniz notationFunction notation
Firsty′y'dydx\frac{dy}{dx}f′(x)f'(x)
Secondy′′y''d2ydx2\frac{d^2y}{dx^2}f′′(x)f''(x)
Thirdy′′′y'''d3ydx3\frac{d^3y}{dx^3}f′′′(x)f'''(x)
nnthy(n)y^{(n)}dnydxn\frac{d^ny}{dx^n}f(n)(x)f^{(n)}(x)

The notation d2ydx2\frac{d^2y}{dx^2} means “differentiate yy with respect to xx twice.” It does not mean a fraction where dxdx is squared in the usual algebraic sense.


If you only have values of ff, use the difference quotient for an approximate derivative:

f′(a)≈f(a+h)−f(a)hf'(a) \approx \frac{f(a+h)-f(a)}{h}

or a symmetric estimate:

f′(a)≈f(a+h)−f(a−h)2h.f'(a) \approx \frac{f(a+h)-f(a-h)}{2h}.

Example. A differentiable function ff has the values below. Estimate f′(2)f'(2) using a symmetric difference quotient.

x123f(x)3715\begin{array}{c|ccc} x & 1 & 2 & 3 \\\hline f(x) & 3 & 7 & 15 \end{array}

Use the values one step on each side of x=2x=2, so a=2a=2 and h=1h=1:

f′(2)≈f(3)−f(1)2(1)=15−32.f'(2)\approx\frac{f(3)-f(1)}{2(1)}=\frac{15-3}{2}.

This simplifies to

f′(2)≈122=6.f'(2)\approx\frac{12}{2}=6.

The symmetric estimate uses points on both sides, so it usually gives a more accurate approximation than a one-sided difference quotient.


Most derivative questions are rule-recognition questions with algebra mixed in. The safest way to work is to identify the outer structure before differentiating.


  1. Interpret the limit as a derivative and then evaluate it exactly:

    lim⁡h→0(2+h)7/3−27/3h.\lim_{h\to0}\frac{(2+h)^{7/3}-2^{7/3}}{h}.

    State the function being differentiated and the input at which its derivative is evaluated.

  1. Let

    f(x)={ax2+b,x<1,3ln⁡x+c,x≥1.f(x)= \begin{cases} ax^2+b, & x<1,\\ 3\ln x+c, & x\ge1. \end{cases}

    Given that f(0)=2f(0)=2, find aa, bb, and cc so that ff is differentiable at x=1x=1. Then find f′(1)f'(1).

  1. Define

    f(x)=∣x2−4x+3∣.f(x)=\lvert x^2-4x+3\rvert.

    (A)(A) Find every input where ff is not differentiable and justify each one using one-sided derivatives.

    (B)(B) Find the equation of the tangent line to ff at x=2x=2.

    (C)(C) Determine whether either nondifferentiable point can be repaired by changing only the value of ff at that point. Explain.

  1. Differentiable functions ff and gg satisfy

    xf(x)f′(x)g(x)g′(x)234−15\begin{array}{c|cccc} x & f(x) & f'(x) & g(x) & g'(x) \\\hline 2 & 3 & 4 & -1 & 5 \end{array}

    Let

    H(x)=f(x)g(x)f(x)+g(x).H(x)=\frac{f(x)g(x)}{f(x)+g(x)}.

    Find H(2)H(2) and H′(2)H'(2). Then write an equation of the normal line to the graph of HH at x=2x=2.

  1. Differentiate and simplify enough to identify every input where the derivative does not exist:

    y=(x2+1)(ex+sin⁡x)x3.y=\frac{(x^2+1)(e^x+\sin x)}{x^3}.

    Use the domain of the original function, not only the appearance of your final derivative.

  1. Let

    f(x)=x2+1x−1.f(x)=\frac{x^2+1}{x-1}.

    Find every point on the graph of ff where the tangent line has slope −1-1. Write the equation of each tangent line and determine whether the corresponding normal lines are parallel.

  1. Let f(x)=x2exf(x)=x^2e^x. Find f(12)(0)f^{(12)}(0) without differentiating the function twelve times one line at a time. Develop and justify a pattern for f(n)(x)f^{(n)}(x) that works for every positive integer nn.
  1. Define

    f(x)={xsin⁡(1/x),x≠0,0,x=0,g(x)={x2sin⁡(1/x),x≠0,0,x=0.f(x)= \begin{cases} x\sin(1/x), & x\ne0,\\ 0, & x=0, \end{cases} \qquad g(x)= \begin{cases} x^2\sin(1/x), & x\ne0,\\ 0, & x=0. \end{cases}

    (A)(A) Determine whether each function is continuous at x=0x=0.

    (B)(B) Use the derivative definition to determine whether each function is differentiable at x=0x=0.

    (C)(C) For each derivative that exists, find its value.

  1. For a>0a>0, let f(x)=axf(x)=a^x. The tangent line to the graph of ff at x=0x=0 passes through the point (2,5)(2,5).

    (A)(A) Find aa exactly.

    (B)(B) Write equations of the tangent and normal lines at x=0x=0.

    (C)(C) Find the xx-intercept of the normal line.

  1. Starting only from

    ddx(sinh⁡x)=cosh⁡x,ddx(cosh⁡x)=sinh⁡x,\frac{d}{dx}(\sinh x)=\cosh x, \qquad \frac{d}{dx}(\cosh x)=\sinh x,

    prove that cosh⁡2x−sinh⁡2x\cosh^2x-\sinh^2x is constant. Then determine the value of the constant by evaluating the expression at x=0x=0.

  1. A particle moves along a line with position

    s(t)=tln⁡t,t>0.s(t)=t\ln t, \qquad t>0.

    Find the exact time t∈(1,e)t\in(1,e) at which the instantaneous velocity equals the average velocity on the interval [1,e][1,e]. Verify directly that your answer lies in the required interval.

  1. Suppose ff is differentiable at x=ax=a.

    (A)(A) Rewrite f(x)−f(a)f(x)-f(a) as a product involving the difference quotient f(x)−f(a)x−a\displaystyle\frac{f(x)-f(a)}{x-a}.

    (B)(B) Use limit laws and the derivative definition to prove that ff must be continuous at x=ax=a.

    (C)(C) Give an example showing that the converse is false: a function can be continuous at a point without being differentiable there.