Given a function f:D→R, we say that x→alimf(x)=L if for every sequence {xn}n⊆D∖{a} such that xn→a, we have n→∞limf(xn)=L.
Compare with the definition of continuity:
f:D→R is continuous at a∈D if for every sequence {xn}n⊆D such that xn→a, we have n→∞limf(xn)=L.
The difference here is that in continuity, the terms of the sequence are allowed to be a, but in the definition of a limit, the elements of your sequence are not allowed to be a. Beyond that, there's not much else to say. Everything is defined in terms of sequences, so as you would expect, all the usual limit laws apply.
Example 1.
Show that x→1lim(x−1x1/3−1)(x2+1)=32.
Solution.
Recall that a3−b3=(a−b)(a2+ab+b2). If we set a=x1/3 and b=1, then
Let I⊆R be an interval and f:I→R be a function. Then f is differentiable at a∈I if
f′(a)=x→alimx−af(x)−f(a)
exists and is finite. In this case, we say f′(a) is the derivative of f at a.
Here are some very useful formulas for calculating derivatives:
Proposition
Suppose f,g are differentiable at a∈R and let c∈R. Then
cf is differentiable at a and (cf)′(a)=cf′(a).
f+g is differentiable at a and (f+g)′(a)=f′(a)+g′(a).
(product rule) fg is differentiable at a and (fg)′(a)=f′(a)g(a)+f(a)g′(a).
(quotient rule) If g(a)=0, then gf is differentiable at a and (gf)′(a)=[g(a)]2f′(a)g(a)−f(a)g′(a).
Theorem (chain rule)
Let a∈R. If f is differentiable at g(a) and g is differentiable at a, then f∘g is differentiable at a and (f∘g)′(a)=f′(g(a))g′(a).
The rest of the notes will be dedicated to examples, since there's not much else to talk about.
Example 2.
(28.8)
Let f(x)=x2 for x rational and f(x)=0 for x irrational. Show that f is differentiable at x=0.
Solution.
First, let's look at a wrong answer:
Notice that f′(x)=2x if x is rational. Thus, f′(0)=0.
The problem with this statement is that f is not continuous if x=0, which means that f′(x) doesn't exist for x=0. As a result, "f′(x) if x is rational" is meaningless. This tells us that our formulas above won't be useful for this problem, and we'll have to do it directly from the definition.
We want to calculate limx→0x−0f(x)−f(0), so it'll be a good idea to see what the function in the limit looks like. If x=0, then
x−0f(x)−f(0)=xf(x)={x0if x is rational,otherwise.
This limit should be 0, so let's try to prove that. Let {xn}n be a sequence such that xn=0 for all n and xn→0. Let ε>0, so by definition, there exists N∈R such that if n>N, then ∣xn∣<ε. Thus,
∣∣xn−0f(xn)−f(0)∣∣={∣xn∣0if xn is rational,otherwise<ε,
since it's less than ε in both cases. Thus, because {xn}n was an arbitrary sequence,
f′(0)=x→0limx−0f(x)−f(0)=0.
Example 3.
(28.14)
Suppose f is differentiable at a. Prove:
h→0limhf(a+h)−f(a)=f′(a)
h→0lim2hf(a+h)−f(a−h)=f′(a)
Solution.
Since f is differentiable at a, we know that for any sequence {xn}n such that xn=a and xn→a, we have
f′(a)=n→∞limxn−af(xn)−f(a).
Let {hn}n be a sequence such that hn=0 and hn→0. Notice that a+hn=a for all n and a+hn→0, so if we set xn=a+hn, we can apply the equation above to get
True or false: If h→0lim2hf(a+h)−f(a−h) exists and is finite, then f is differentiable at a.
Solution.
False. The exercise above tells us that if the derivative exists, then we can calculate f′(a) using the above formula. If the derivative doesn't exist, then the formula may no longer be true. For example, if f(x)=∣x∣, then
h→0lim2hf(a+h)−f(a−h)=h→0lim2h∣h∣−∣−h∣=0,
but f is not differentiable at x=0.
Example 5.
(28.16)
Let f be a function defined on an open interval I containing a. Show that f′(a) exists if and only if there is a function ε(x) defined on I and a real number b such that
f(x)−f(a)=(x−a)(b−ε(x))andx→alimε(x)=0.
Solution.
"⟹"
Suppose f′(a) exists. Let's try to reverse engineer what ε and b would have to be:
f(x)−f(a)=(x−a)(b−ε(x))⟹ε(x)=b−x−af(x)−f(a).
For ε(x) to tend to 0 as x→0, we would need b=f′(a), by the definition of the derivative. Now let's prove that this works:
Technically, ε is defined on I∖{a}, but its value at x=a doesn't matter since we're taking limits, so we can set it to be any number we like to get a function defined on I. Then
f(x)−f(a)=(x−a)(f′(a)−ε(x))
just by definition of ε(x), and by our choice of b, we get x→alimε(x)=0 automatically.
"⟸"
Suppose that ε(x) and b exist as in the problem statement. By rearranging the equation in the problem statement and applying limit laws,