Where a function is headed near a point — even when the point itself cannot be trusted, or plugged in at all.
Figure 2.2.1: f(x)=x−2x2−4 — the line y=x+2 with the point (2,4) punched out.
Dear reader,
This section is about a single, patient question: where is a function headed as its input creeps toward a point? Sometimes the point is a smooth part of the curve. Sometimes there is a hole, a jump, or a wall. The limit is the language that describes all four at once.
Read for the destination the graph is approaching, not the value it may or may not hold at the point itself.
In this section you will
For f(x)=x−2x2−4, the point x=2 has no value — you would divide by zero. So where is the graph headed as x draws close? Pick x-values just below and just above 2, and watch f(x).
From both sides the outputs close in on 4, even though f(2)=00 has no value. The graph heads toward 4 without ever reaching it.
Table of values — both sides of 2
| from left | f(x) | from right | f(x) |
|---|---|---|---|
| 1 | 3 | 3 | 5 |
| 1.5 | 3.5 | 2.5 | 4.5 |
| 1.9 | 3.9 | 2.1 | 4.1 |
| 1.99 | 3.99 | 2.01 | 4.01 |
| 1.999 | 3.999 | 2.001 | 4.001 |
| 1.9999 | 3.9999 | 2.0001 | 4.0001 |
You found where f was heading as x→2 — without ever using x=2 itself. The point of mathematics is to write down exactly what you did, so it works for the next function too:
The method, named
The Closing-In Method
To find the value a function heads toward as x nears a point, plug in x-values closer and closer from both sides and watch what number the outputs settle on.
How can we describe where a function is headed near a point — even when we can't, or shouldn't, plug the point in?
The hole at x=2 is only the first way a point can't be trusted. By the end you will have a name for every way a function can behave as x approaches — and every way that behavior can fail.
Let f(x) be defined for all x in an open interval containing a, with the possible exception of a itself, and let L be a real number. If all values of f(x) approach L as the values of x (=a) approach a, then the limit of f(x) as x approaches a is L.
The limit reports what the neighborhood is doing, not the value at the point. For f(x)=x−2x2−4, the value f(2) does not exist — yet x→2limf(x)=4.
The notation
x→alimf(x)=L.
Figure 2.2.2: the shaded band around a is the neighborhood the limit reads — the point a itself may be excluded.
Factor and cancel:
x−2x2−4=x−2(x−2)(x+2)=x+2for x=2.So f is the line y=x+2 with the single point (2,4) punched out. The cancellation is legal only for x=2 — which is why we approach the point instead of substituting it. (Evaluating limits exactly by factoring is the work of §2.3.)
A limit reads where a function is headed, not where it is — like reading which way a friend is heading through a crowd, even when you cannot see exactly where they stand.
Evaluate x→0limxsinx with a table. The function is even, so the left and right columns match:
| x | sinx/x |
|---|---|
| ±0.1 | 0.99833 |
| ±0.01 | 0.99998 |
| ±0.001 | 0.9999998 |
Both sides march toward 1: x→0limxsinx=1.
Figure 2.2.3: xsinx heads to 1 at x=0, where the function itself is undefined.
Estimate x→1limx−1x1−1 using a table of values.
Simplify first: x−1x1−1=x(x−1)−(x−1)=−x1 for x=1.
Both columns march toward −1, so x→1limx−1x1−1=−1. Answer: −1.
Table of values — both sides of 1
| x | f(x) | x | f(x) |
|---|---|---|---|
| 0.9 | −1.1111 | 1.1 | −0.9091 |
| 0.99 | −1.0101 | 1.01 | −0.9901 |
| 0.999 | −1.0010 | 1.001 | −0.9990 |
For the function g(x) shown, evaluate x→−1limg(x).
The marked point shows g(−1)=4, but the curve approaches height 3 from both sides: x→−1limg(x)=3.
The limit and the function value are allowed to disagree — the limit reports the neighborhood, not the isolated point.
Figure 2.2.4: the curve heads to 3 at x=−1, while the plotted point sits at 4.
Tables and graphs are quick, but they rest on guesswork — you are reading a trend, not proving a value. The next section builds algebraic limit laws that compute limits exactly.
Before we leave the informal stage, two limits are so basic that every one of those laws will lean on them.
The identity function f(x)=x marches along with x; a constant function never moves. These two limits are the atomic building blocks inside every limit law we prove from here on.
For any real a and constant c
x→alimx=ax→alimc=c.For a two-sided limit to exist, f(x) must approach a single real number from both sides. It can miss in two very different ways — and the rest of the section gives each its own precise vocabulary.
Way one
Oscillation
f bounces around forever without settling.
Way two
Blow-up
f shoots off toward ±∞ — a vertical wall.
Both get labeled DNE for now — but each is about to earn its own precise name.
Evaluate x→0limsin(x1). As x→0, the input 1/x blows up, so sin(1/x) oscillates faster and faster — never settling.
Along x=π2,3π2,5π2,… the value cycles 1,−1,1,−1,… forever. Different paths to 0 give different values, so x→0limsin(x1) DNE.
Figure 2.2.5: sin(1/x) oscillates ever faster near 0 — no single value to head toward.
Saying "the limit does not exist" sometimes throws away useful information. sin(1/x) genuinely wandered — nothing more to say.
But recall g(x)=x−2∣x−2∣: it settles cleanly on −1 from the left and +1 from the right.
The two-sided limit fails, yet each side is perfectly well-behaved. That deserves its own notation.
Figure 2.2.6: g(x)=x−2∣x−2∣ — clean on each side, a jump at x=2.
If f(x) approaches L as x (x<a) approaches a from the left, we write x→a−limf(x)=L. If it approaches L as x (x>a) approaches a from the right, we write x→a+limf(x)=L.
For g(x)=x−2∣x−2∣: the two-sided limit DNE, but the side limits are clean — x→2−limg(x)=−1 and x→2+limg(x)=1.
From the left · from the right
x→a−limf(x)=L x→a+limf(x)=LSuperscript sign: a− reads "just below a"; a+ reads "just above a".
Take the piecewise function f(x)={x+1x2−4x<2,x≥2.
Use the correct branch on each side: x→2−limf(x)=3 and x→2+limf(x)=0.
The side limits disagree, so x→2limf(x) DNE — the graph jumps at x=2.
Figure 2.2.7: the branches meet at different heights — a jump discontinuity at x=2.
Estimate the one-sided limits of x−2∣x2−4∣ at x=2.
Factor: ∣x2−4∣=∣x−2∣∣x+2∣. For x>2, the ratio →∣x+2∣=4; for x<2, it →−∣x+2∣=−4.
The two side limits
x→2−limx−2∣x2−4∣=−4 x→2+limx−2∣x2−4∣=4Answer: left side −4; right side +4; the two-sided limit DNE.
The relationship
x→alimf(x)=L⟺x→a−limf(x)=L and x→a+limf(x)=L.The two-sided limit exists exactly when the two side limits shake hands on the same value. If they disagree, the two-sided limit does not exist.
Every function so far stayed bounded near its trouble point. But the third opening function never settles:
h(x)=(x−2)21 ⟶ +∞as x→2.This is the blow-up failure we flagged earlier. Rather than write "DNE", we record how it fails — with the symbol ∞.
Figure 2.2.8: h(x)=(x−2)21 climbs without bound on both sides of x=2.
If f(x) increases without bound as x→a, we write x→alimf(x)=+∞; if it decreases without bound, x→alimf(x)=−∞. The same applies to the one-sided versions x→a− and x→a+. Writing +∞ records how the limit fails — by blowing up — instead of just labeling it "DNE". (Infinity is a description of behavior, not a real-number value.)
Increase / decrease without bound
x→alimf(x)=+∞ x→alimf(x)=−∞The line x=a is a vertical asymptote of f(x) if any one of the following holds:
Any one is enough
x→a−limf(x)=±∞,x→a+limf(x)=±∞,x→alimf(x)=±∞.Example: h(x)=(x−2)21 has x→2limh(x)=+∞, so the line x=2 is a vertical asymptote.
Evaluate the one- and two-sided limits of x1 at x=0.
| x | 1/x | x | 1/x |
|---|---|---|---|
| −0.01 | −100 | 0.01 | 100 |
| −0.001 | −1000 | 0.001 | 1000 |
The sides blow up in opposite directions, so x→0−limx1=−∞, x→0+limx1=+∞, and the two-sided limit DNE.
Figure 2.2.9: x1 dives to −∞ on the left and climbs to +∞ on the right of 0.
If n is even
x→alim(x−a)n1=+∞Both sides climb together to +∞.
If n is odd
x→a+lim(x−a)n1=+∞,x→a−lim(x−a)n1=−∞The sides split to opposite infinities.
For a real number a and a positive integer n — either way, the line x=a is a vertical asymptote.
Evaluate the limits of f(x)=(x+3)41 at x=−3, and identify any vertical asymptote.
Match the template (x−a)n1: here a=−3 and n=4 (even).
Even n means both sides shoot to +∞, and because the limit blows up at x=−3, the line x=−3 is a vertical asymptote.
Both sides climb
x→−3−limf(x)=x→−3+limf(x)=+∞ x→−3limf(x)=+∞Vertical asymptote: x=−3.
Evaluate the limits of f(x)=(x−2)31 at x=2 and identify any vertical asymptote.
n=3 is odd, so the side limits go to opposite infinities. The sides disagree, so the two-sided limit DNE — but one side blowing up is enough.
Opposite infinities
x→2−lim(x−2)31=−∞ x→2+lim(x−2)31=+∞Answer: left −∞; right +∞; two-sided DNE; vertical asymptote at x=2.
Read each behavior off the graph of f:
x=−4 (smooth): x→−4limf(x)=0, and f(−4)=0.
x=−2 (hole): x→−2limf(x)=3, but f(−2) is undefined.
x=1 (jump): lim−=6, lim+=3, two-sided DNE, f(1)=6.
x=3 (asymptote): both sides →−∞, f(3) undefined.
Figure 2.2.10: smooth at −4, a hole at −2, a jump at 1, an asymptote at 3.
Where is a function headed near a point — even when the point itself can't be trusted? One graph holds all four answers.
The limit always reads the neighborhood, never the point. When the neighborhood agrees, the limit exists; when it splits or blows up, we don't stop at "DNE" — we name how.
The four behaviors
| Point on the graph | Limit verdict |
|---|---|
| smooth (x=−4) | exists; lim=f(a) |
| hole (x=−2) | exists; lim=f(a) |
| jump (x=1) | two-sided DNE; sides clean |
| asymptote (x=3) | infinite; vertical asymptote |
Table · graph · one-sided · infinite — every way a limit can hold or fail.
Yours in the limit,
— the Calculus Volume 1 reader
Next: §2.3 The Limit Laws — computing limits exactly, so the tables and graphs become proofs.