Textbooks write \sup S and move on. What they rarely show is what the supremum actually does when you disturb the set. This satellite gives you that disturbance. Grab the set. Shove it around. Squeeze it, stretch it, flip it from closed to open. The sup follows like a rigid ceiling that always sits on top of the set — as low as it can go without falling inside.
Supremum in one sentence
The supremum of a set of real numbers is the smallest real number that is still \ge every element of the set. Equivalently: it is the tightest ceiling you can place over the set without pushing any element through the roof.
The word "tightest" is carrying all the weight here. Any large enough number is an upper bound — 1000 is an upper bound of the set (0, 1), and so is 10, and so is 1.5, and so is 1.01. The supremum is the one you cannot shrink any further without losing the upper-bound property. For (0, 1) that one number is exactly 1.
Drag the widget below. Whatever interval you build, the red tick at the top of the canvas marks the supremum. Slide the right endpoint b and watch the tick move with it, one-for-one. Slide a, watch b stay put and the tick stay put. The sup is locked to the right end of the set, not the left, not the middle.
The widget
Three things to try.
First, slide b while leaving a alone. The supremum triangle tracks b exactly. That is the content of the statement "the sup is b" for an interval — nothing subtle about it, just a rigid lock.
Second, slide a past b (the widget silently swaps them so a \le b always). The sup keeps sitting on the larger endpoint. The sup does not care about the left end of the set.
Third, toggle the "open at b" checkbox. When unticked, b itself belongs to the set — the set is [a, b] — and the dot on the triangle is filled, meaning \sup S is also \max S. When ticked, b is excluded — the set is [a, b) — and the dot goes hollow, meaning \sup S still equals b but the set has no maximum. Same supremum; different status as an element.
Why the supremum exists at all
It is not obvious that every bounded non-empty set should have a least upper bound. On the rationals, this property fails. Consider S = \{x \in \mathbb{Q} : x > 0,\ x^2 < 2\}: a set of rationals bounded above by 2, but with no least rational upper bound. Any rational r you propose as the tightest ceiling can be nudged down a hair and still sit above every element of S, because the "true" ceiling \sqrt{2} is not rational and there is no nearest rational to it.
The real numbers were built specifically to fix this. The key axiom is the completeness axiom, also called the least upper bound property:
Every non-empty subset of \mathbb{R} that is bounded above has a least upper bound in \mathbb{R}.
"Bounded above" means there exists some M with x \le M for all x \in S. "Least upper bound" means the smallest such M. The axiom says: if the first exists, the second does too. That is exactly the guarantee that the widget's red triangle always has somewhere real to land, no matter how you wiggle the sliders.
Why this is an axiom and not a theorem: on the rationals the analogous statement is false, so the property is not something you can prove from arithmetic alone. It is baked into the definition of \mathbb{R} — either by Dedekind cuts, by equivalence classes of Cauchy sequences, or by decimal expansions. Every construction of the real numbers has to produce this property, and each does it differently. See Real Numbers — Properties for the axiom's role in the full picture.
Sup versus max — a distinction that matters
The supremum is the smallest upper bound. The maximum is the largest element of the set. They can coincide, or they can disagree.
- Coincide: for S = [0, 1], both \sup S = 1 and \max S = 1. The ceiling 1 belongs to the set, so 1 is also the biggest element.
- Disagree: for S = [0, 1), we still have \sup S = 1, but \max S does not exist. No element of [0, 1) is biggest — whatever you propose, say 0.999, there is a larger element 0.9995 in the set, and so on forever.
In the widget, this is the open/closed toggle. The supremum is stubborn: b is b. What changes is whether the set contains its supremum. When it does, the set has a maximum and the dot is filled. When it does not, the set has no maximum — only a supremum that it approaches without ever reaching.
JEE problems exploit this distinction. Whenever a question asks for the "maximum value of f(x) on an open interval," the honest answer is sometimes "the maximum does not exist, but the supremum is..." Teachers accept both as long as you name the situation correctly.
Worked example — a set with a surprising sup
Consider
Every element is strictly less than 1. So 1 is an upper bound. Is it the least upper bound?
Claim. \sup S = 1, and \max S does not exist.
Proof that 1 is an upper bound. For every n \ge 1, \frac{1}{n} > 0, so 1 - \frac{1}{n} < 1. Every element of S is below 1.
Proof that nothing smaller works as an upper bound. Take any candidate M < 1. You need to exhibit an element of S that exceeds M. Choose n large enough that \frac{1}{n} < 1 - M — such n exists by the Archimedean property. Then 1 - \frac{1}{n} > 1 - (1 - M) = M, so that element of S is larger than M. The candidate M is not an upper bound after all.
Combining the two, 1 is an upper bound and nothing less than 1 is. So \sup S = 1.
No maximum. If S had a maximum, it would be some element 1 - \frac{1}{N}. But then 1 - \frac{1}{N+1} is also in S and is larger (since \frac{1}{N+1} < \frac{1}{N}). So the "maximum" is not actually the maximum — contradiction. No element of S is biggest.
Set a = 0 and b = 1 with the "open at b" box ticked in the widget. You are looking at [0, 1), a cousin set with the same supremum and same no-maximum status. The dots of \{1 - 1/n\} would sit inside that shaded band, piling up against the hollow dot at 1 — approaching the ceiling, never touching it.
What this generalises to
Every time you see a phrase like "\sup_{x \in S} f(x)" in calculus, you are asking the same question the widget asks: over some set S, what is the tightest ceiling the values of f can be pushed against? Sometimes the ceiling is achieved — the function attains it — and you can write \max instead. Often the ceiling is approached but never reached, and you must write \sup to be honest. The widget teaches your eye to distinguish these two cases by shape: filled dot versus hollow dot, same triangle position either way.
Related: Real Numbers — Properties · The Rational Line Has Holes, the Real Line Does Not · Intervals and Inequalities Preview · Number Systems