Limits & One-Sided Limits
What a limit really means — the value a function heads toward — plus left vs right limits, limits at infinity, and why a limit can exist where the function isn't even defined.
What you'll learn
- A limit is the value f(x) approaches as x → a — not necessarily f(a) itself
- Left-hand and right-hand limits, and that the two-sided limit exists only when they agree
- Limits at infinity describe end behavior and horizontal asymptotes
- A limit can exist even where f(a) is undefined (a hole in the graph)
Before you start
Last lesson closed on a puzzle. The function 1/x is not defined at x = 0, yet as you
creep toward zero along its curve it plainly does something — it climbs without bound.
To describe that honestly, you have to stop asking “what is the height at the point?” and start asking “where is the height headed as I approach?” Picture walking along a graph toward some x-value, watching only where you are going — the destination, not the arrival.
That “where am I headed” has a name: the limit. The gap between heading-toward and actually-arriving sounds tiny, but it is the engine behind everything ahead — derivatives, continuity, the whole Calculus block — and GATE leans on it as a quiet first step in nearly every limit and continuity question.
The same question drives real machine learning: a training loss flattening toward a floor, or a learning rate decaying toward zero, are both limits in disguise.
The intuition
Walk along the graph toward x = a and keep your eye on the height. If that height settles
toward a single number L no matter how you approach, we write
lim_{x → a} f(x) = L
and read it “the limit of f as x approaches a is L.” The key word is approaches:
L is decided by the neighbourhood of a — the values of f at points arbitrarily
close to a on either side — and never by the single point a itself.
That last clause is the one to say out loud, because it is the confusion the whole topic rests on: the limit asks where the function is going, not where it is.
Those are separate questions with separate answers. A graph can have a hole at a and still
have a perfectly good limit there; it can be defined at a to some unrelated value and the
limit ignores it entirely. Whenever a limit question feels wrong, check whether you have
quietly answered “what is f(a)?” instead of “where is f headed near a?”
A concrete example
Concretely: let f(x) = x + 2 everywhere except at x = 2, where somebody has declared
f(2) = 7. Creep in from the left and the heights run 3.9, 3.99, 3.999; creep in from the
right and they run 4.1, 4.01, 4.001.
Both sides head to 4, so lim_{x → 2} f(x) = 4 — while f(2) = 7 sits there, off the
trend line, changing nothing.
To watch a limit actually happen, drag the slider that controls h in the widget below. The
secant line — the straight line cutting the curve at two points — rotates as h shrinks,
and as h → 0 it settles onto one fixed line.
That settling, that approach to a single position, is exactly what “the limit exists” looks like with your own eyes.
Drag the point — read the slope off the tangent line
The secant line touches the curve at two points a distance h apart. Shrink h toward 0 and watch it rotate into the tangent: that limiting slope is the derivative f′(x).
Left-hand vs right-hand limits
You can sneak up on a from two sides, and which side you choose can matter:
- Left-hand limit
lim_{x → a⁻} f(x)— approach withxslightly less thana(from the left). - Right-hand limit
lim_{x → a⁺} f(x)— approach withxslightly greater thana(from the right).
The two-sided limit lim_{x → a} f(x) exists only when both one-sided limits exist and are
equal. If the left says 2 and the right says 3, the curve is heading to two different
places depending on direction, so there is no single destination — the limit does not
exist.
Limits at infinity
Instead of approaching a finite point, we can ask where f(x) heads as x runs off to
+∞ or −∞. That captures the end behaviour of the graph and pins down any
horizontal asymptote.
A horizontal asymptote is a horizontal line the curve flattens toward forever without ever reaching. The classic one is
lim_{x → ∞} 1/x = 0
As x grows huge, 1/x shrinks toward 0 without ever touching it, so y = 0 is a
horizontal asymptote.
The other examples are:
- Polynomials behave the opposite way, running off to
±∞. eˣblows up to+∞on the right yet decays to0on the left.
Each of these is just the heading-toward question pointed at the far ends of the axis.
How GATE asks this
This material rides embedded inside limit and continuity questions.
A question may:
- hand you a piecewise function — one defined by different formulas on different stretches of the x-axis — and ask whether the limit at the seam between two stretches exists;
- ask for one one-sided value; or
- ask for an at-infinity limit (typed as a NAT) and the matching asymptote (an MCQ).
You answer by checking both one-sided limits and comparing. The reflex never changes: approach from each relevant side, then compare where the sides point.
Worked example
A function is defined piecewise as
f(x) = x + 1forx < 1andf(x) = x + 2forx ≥ 1. Doeslim_{x → 1} f(x)exist? Separately, evaluatelim_{x → ∞} 1/x.
Start with the seam at x = 1, taking each side on its own.
left limit: lim_{x → 1⁻} (x + 1) = 1 + 1 = 2
right limit: lim_{x → 1⁺} (x + 2) = 1 + 2 = 3
The left-hand limit is 2 and the right-hand limit is 3. They disagree, so the curve heads to two different heights depending on the direction of approach.
Therefore the two-sided limit lim_{x → 1} f(x) does not exist — even though f(1) = 3
is perfectly well defined. Existence of the limit and the value of f(1) are simply separate
questions, and the worked answer shows them parting ways.
Now the at-infinity part.
lim_{x → ∞} 1/x = 0
As x grows without bound, 1/x collapses toward 0. The value 0 is never actually
attained, but that is exactly what a limit at infinity reports — where the function is
headed, not where it lands.
A question to carry forward
We now know what a limit means and how to read one off a graph or sneak up on it from each side. But reading is not yet computing.
Try to evaluate lim_{x → 1} (x² − 1)/(x − 1) by the obvious move: substitute x = 1.
You get 0/0, which is nonsense, not an answer. Yet the curve clearly heads somewhere
definite.
Here is the thread onward: when plugging in produces a meaningless 0/0, what honest algebra
recovers the value the function is really approaching?
In one breath
- A limit
lim_{x → a} f(x) = Lis the valuefheads toward neara— decided by the neighbourhood, not byf(a). - One-sided limits: approach from the left (
a⁻) or right (a⁺). The two-sided limit exists only when both agree; disagree (2 vs 3) ⇒ does not exist. - A limit can exist even where
f(a)is undefined (a hole), and can differ from a definedf(a).lim = f(a)is the special case called continuity. - Limits at infinity give end behaviour and horizontal asymptotes:
lim_{x→∞} 1/x = 0, polynomials →±∞,eˣ→+∞right and0left. - Reflex: approach from each relevant side and compare before declaring a limit.
Practice
Quick check
Practice this in an interview
All questionsThe Law of Large Numbers (LLN) says the sample mean converges to the true mean as sample size grows — it is a statement about where the mean lands. The Central Limit Theorem says the sampling distribution of the mean is approximately normal — it is a statement about the shape of that distribution. LLN guarantees convergence; CLT characterises the rate and shape of that convergence.
The CLT states that the sampling distribution of the sample mean converges to a normal distribution as sample size grows, regardless of the shape of the underlying population distribution. It is the theoretical foundation for confidence intervals, hypothesis tests, and many machine-learning approximations — but it applies to the distribution of the mean, not to the raw data.