The Central Idea
Imagine moving along the number line toward a location determined more and more precisely. If every stage of the process stays within a bounded region and the uncertainty keeps shrinking, we expect the process to identify an actual real number.
The completeness of \(\mathbb{R}\) is the principle that makes that expectation mathematically reliable.
One standard form says:
\[ \boxed{\text{Every nonempty subset of }\mathbb{R} \text{ that is bounded above has a least upper bound in }\mathbb{R}.} \]This statement is called the least-upper-bound property or supremum property.
Why Completeness Matters
Many familiar facts from calculus quietly depend on completeness. It is what allows us to justify statements such as:
- a bounded increasing sequence has a real limit;
- nested closed intervals with shrinking lengths determine a real point;
- Cauchy sequences of real numbers converge in \(\mathbb{R}\);
- positive real numbers have real square roots;
- continuous functions on closed intervals attain extrema;
- the real line contains the limits needed for rigorous differentiation and integration.
Completeness is therefore not an isolated technical definition. It is part of the foundation that supports the rest of real analysis.
Order Comes Before Completeness
Completeness is stated using the order relation on the real numbers. We need to know what it means for one number to lie above or below every member of a set.
Let \(S\subseteq\mathbb{R}\). A number \(M\in\mathbb{R}\) is an upper bound for \(S\) if
\[ x\le M \qquad \text{for every }x\in S. \]Similarly, \(m\in\mathbb{R}\) is a lower bound for \(S\) if
\[ m\le x \qquad \text{for every }x\in S. \]Bounded Above and Bounded Below
A set is bounded above if it has at least one upper bound. It is bounded below if it has at least one lower bound.
For example, consider
\[ S=(0,4). \]Every number \(M\ge4\) is an upper bound for \(S\), and every number \(m\le0\) is a lower bound.
Notice that an upper bound does not have to belong to the set.
Maximum Versus Upper Bound
An upper bound and a maximum are different ideas.
A number \(M\) is the maximum of \(S\) if
\[ M\in S \]and
\[ x\le M \qquad \text{for every }x\in S. \]The open interval
\[ (0,4) \]has many upper bounds but no maximum, because \(4\notin(0,4)\) and every point less than \(4\) can be exceeded by another point of the interval.
Minimum Versus Lower Bound
The same distinction appears at the lower end.
A number \(m\) is the minimum of \(S\) if
\[ m\in S \]and
\[ m\le x \qquad \text{for every }x\in S. \]The interval \((0,4)\) has lower bounds but no minimum.
The Least Upper Bound
Suppose \(S\) is nonempty and bounded above. An upper bound \(u\) is the least upper bound if no smaller upper bound exists.
We write
\[ \boxed{u=\sup S} \]and call \(u\) the supremum of \(S\).
Formally, \(u=\sup S\) means:
- \(x\le u\) for every \(x\in S\);
- if \(v\) is any upper bound for \(S\), then \(u\le v\).
The Greatest Lower Bound
Dually, if \(S\) is nonempty and bounded below, its greatest lower bound is called its infimum.
\[ \boxed{\ell=\inf S} \]The statement \(\ell=\inf S\) means:
- \(\ell\le x\) for every \(x\in S\);
- if \(b\) is any lower bound for \(S\), then \(b\le\ell\).
Example: Supremum Without a Maximum
Let
\[ S=\{x\in\mathbb{R}:x<5\}. \]The number \(5\) is an upper bound. It is also the least upper bound:
\[ \boxed{\sup S=5}. \]But \(S\) has no maximum because \(5\notin S\).
This example is important because it shows why the supremum concept is more flexible than the maximum concept.
Example: A Maximum Is Automatically the Supremum
Let
\[ T=[-2,7]. \]The maximum is \(7\), so
\[ \max T=7. \]Since \(7\) is already an upper bound and belongs to \(T\), no smaller number can be an upper bound. Therefore,
\[ \boxed{\sup T=7}. \]Whenever a maximum exists, it equals the supremum. The converse need not be true.
A Useful Characterization of the Supremum
Suppose \(u\) is an upper bound for a nonempty set \(S\). Then \(u=\sup S\) exactly when points of \(S\) can be found arbitrarily close to \(u\) from below.
More precisely, for every \(\varepsilon>0\), there exists \(x\in S\) such that
\[ u-\varepsilonThe Completeness Axiom
We can now state the central property precisely.
\[ \boxed{ S\subseteq\mathbb{R},\quad S\ne\varnothing,\quad S\text{ bounded above} \Longrightarrow \sup S\in\mathbb{R} }. \]In ordinary HTML rendering, the same statement is:
Every nonempty real set that is bounded above has a least upper bound that is itself a real number.
Why the Rational Numbers Are Not Complete
The ordered field \(\mathbb{Q}\) behaves like \(\mathbb{R}\) in many algebraic ways, but it is not complete.
Consider the rational set
\[ S=\{q\in\mathbb{Q}:q>0\text{ and }q^2<2\}. \]This set is nonempty and bounded above in \(\mathbb{Q}\). Intuitively, its least upper bound should be \(\sqrt2\). But
\[ \sqrt2\notin\mathbb{Q}. \]Therefore the rational number system contains a genuine gap relative to this bounded set.
Completeness Fills the Gaps
The real number system is constructed or characterized so that gaps of the rational kind are filled.
When a bounded set of real numbers presses upward toward a boundary, the least possible upper boundary is guaranteed to be another real number.
This is the sense in which the real line is often described informally as having no holes.
Using Completeness to Produce \(\sqrt2\)
Completeness can be used to prove that there is a positive real number whose square is \(2\).
Define
\[ S=\{x\in\mathbb{R}:x\ge0\text{ and }x^2<2\}. \]The set is nonempty because \(1\in S\). It is bounded above; for example, \(2\) is an upper bound.
By completeness, \(S\) has a supremum. Let
\[ a=\sup S. \]Why the Supremum Must Satisfy \(a^2=2\)
Because \(1\in S\), we know \(a\ge1\). We now rule out the two alternatives
\[ a^2<2 \]and
\[ a^2>2. \]First suppose \(a^2<2\). Choose \(h>0\) small enough that
\[ h<1 \qquad\text{and}\qquad h(2a+1)<2-a^2. \]
Since \(h<1\), we have \(h^2
Now suppose \(a^2>2\). Choose \(h>0\) so that
\[ h Then \(a-h>0\) and \[ (a-h)^2 =a^2-2ah+h^2 >a^2-2ah >2. \]
If some \(x\in S\) satisfied \(x\ge a-h\), then \(x\ge0\) would give
\(x^2\ge(a-h)^2>2\), contradicting the definition of \(S\). Hence every
\(x\in S\) satisfies \(x
Both alternatives are impossible. Therefore,
Completeness Is Not the Same as Algebraic Closure
It is useful to separate two different ideas.
The real numbers are complete with respect to their order and metric structure, but they are not algebraically closed.
For example,
\[ x^2+1=0 \]has no real solution.
Completeness guarantees the limits and boundary points required by real analysis. It does not guarantee that every polynomial equation has a real root.
The Archimedean Property
One major consequence of completeness is the Archimedean property:
\[ \boxed{\text{For every }x\in\mathbb{R}, \text{ there exists }n\in\mathbb{N}\text{ such that }n>x.} \]No real number is infinitely large compared with all natural numbers.
Equivalently, for every \(\varepsilon>0\), there exists \(n\in\mathbb{N}\) such that
\[ 0<\frac1n<\varepsilon. \]Why the Archimedean Property Follows from Completeness
Suppose, for contradiction, that \(\mathbb{N}\) were bounded above in \(\mathbb{R}\). Completeness would then give
\[ s=\sup\mathbb{N}. \]
Because \(s-1
Adding \(1\) gives
But \(n+1\in\mathbb{N}\), contradicting the claim that \(s\) is an upper
bound.
Rational Numbers Are Dense in the Reals
Completeness and the Archimedean property help establish that between any two distinct real numbers there is a rational number.
If
\[ aIrrational Numbers Are Also Dense
Between any two distinct real numbers there is also an irrational number.
One way to see this is to begin with an irrational number such as \(\sqrt2\) and combine it with a sufficiently small rational displacement.
Thus every nontrivial interval contains both rational and irrational numbers.
Density describes how points are distributed. Completeness describes whether limiting boundary points are missing. These are related but distinct ideas.
Monotone Sequences
A sequence \((a_n)\) is increasing if
\[ a_n\le a_{n+1} \]for every \(n\).
If the sequence is also bounded above, its set of values
\[ S=\{a_n:n\in\mathbb{N}\} \]has a supremum by completeness.
The Monotone Convergence Theorem
A fundamental consequence of completeness is:
\[ \boxed{\text{Every increasing sequence that is bounded above converges.}} \]More specifically, if \((a_n)\) is increasing and bounded above, then
\[ \lim_{n\to\infty}a_n=\sup\{a_n:n\in\mathbb{N}\}. \]Similarly, every decreasing sequence that is bounded below converges to the infimum of its set of values.
Why the Monotone Convergence Theorem Works
Let
\[ L=\sup\{a_n:n\in\mathbb{N}\}. \]Given \(\varepsilon>0\), the number \(L-\varepsilon\) cannot be an upper bound. Therefore there exists some index \(N\) such that
\[ a_N>L-\varepsilon. \]Since the sequence is increasing,
\[ L-\varepsilonThis is exactly the definition of
\[ \lim_{n\to\infty}a_n=L. \]Nested Intervals
Another completeness principle concerns closed intervals
\[ I_n=[a_n,b_n]. \]Suppose they are nested:
\[ I_1\supseteq I_2\supseteq I_3\supseteq\cdots. \]Under suitable boundedness conditions, completeness guarantees that the intervals share at least one real point.
The Nested Interval Property
If
\[ [a_1,b_1]\supseteq[a_2,b_2]\supseteq[a_3,b_3]\supseteq\cdots \]is a nested sequence of nonempty closed bounded intervals, then
\[ \boxed{\bigcap_{n=1}^{\infty}[a_n,b_n]\ne\varnothing.} \]If, in addition,
\[ b_n-a_n\to0, \]then the intersection contains exactly one real number.
How Nested Intervals Use Suprema
Because the intervals are nested, each left endpoint satisfies
\[ a_n\le b_m \]for all relevant indices \(n\) and \(m\). Thus the set of left endpoints is bounded above.
Let
\[ c=\sup\{a_n:n\in\mathbb{N}\}. \]One then shows
\[ a_n\le c\le b_n \]for every \(n\), so
\[ c\in\bigcap_{n=1}^{\infty}[a_n,b_n]. \]Cauchy Sequences
A sequence may be recognized as convergent even before its limit is known.
A sequence \((a_n)\) is a Cauchy sequence if for every \(\varepsilon>0\), there exists \(N\in\mathbb{N}\) such that
\[ |a_m-a_n|<\varepsilon \qquad \text{whenever }m,n\ge N. \]The terms eventually become arbitrarily close to one another.
Cauchy Completeness of \(\mathbb{R}\)
The real numbers satisfy another fundamental completeness statement:
\[ \boxed{\text{Every Cauchy sequence of real numbers converges to a real number.}} \]This property is especially important because it does not require us to know the candidate limit in advance.
In more advanced settings, this form of completeness becomes the model for complete metric spaces and Banach spaces.
Why Cauchy Completeness Fails in \(\mathbb{Q}\)
A sequence of rational approximations can become arbitrarily close to \(\sqrt2\) while remaining rational at every stage.
Such a sequence can be Cauchy in \(\mathbb{Q}\), but its natural limit is not rational.
Thus the sequence has no limit inside \(\mathbb{Q}\).
In \(\mathbb{R}\), the missing limit has been added, and the Cauchy sequence converges.
Bolzano-Weierstrass
Completeness also underlies the Bolzano-Weierstrass theorem:
\[ \boxed{\text{Every bounded sequence in }\mathbb{R} \text{ has a convergent subsequence.}} \]A bounded sequence need not converge as a whole. For example,
\[ a_n=(-1)^n \]oscillates between \(-1\) and \(1\). But it contains convergent subsequences: one constant subsequence equal to \(1\), and another equal to \(-1\).
Completeness and Compactness
Later in analysis, completeness joins with boundedness to help explain why closed bounded subsets of \(\mathbb{R}\) have especially strong properties.
In \(\mathbb{R}\), the interval
\[ [a,b] \]is compact. This supports theorems about extrema, uniform continuity, and subsequences.
Compactness will receive its own treatment later in this 5.16 sequence.
Several Equivalent Faces of Completeness
For the real numbers, several major statements can be developed as equivalent formulations or consequences of completeness.
- the least-upper-bound property;
- the greatest-lower-bound property;
- the monotone convergence theorem;
- the nested interval property;
- Cauchy completeness;
- the Bolzano-Weierstrass property.
Different textbooks choose different statements as the starting axiom and derive the others.
The Greatest-Lower-Bound Property from the Supremum Property
The infimum property does not need to be assumed separately.
Suppose \(S\) is nonempty and bounded below. Form the reflected set
\[ -S=\{-x:x\in S\}. \]Then \(-S\) is bounded above, so it has a supremum. The infimum of \(S\) is
\[ \boxed{\inf S=-\sup(-S)}. \]Reflection converts lower-bound questions into upper-bound questions.
Supremum Arithmetic: Translation
Suprema interact naturally with translations.
If \(S\) is nonempty and bounded above and \(c\in\mathbb{R}\), define
\[ S+c=\{x+c:x\in S\}. \]Then
\[ \boxed{\sup(S+c)=\sup S+c}. \]Translating every point of a set by the same amount translates its least upper bound by that amount.
Supremum Arithmetic: Positive Scaling
If \(\lambda>0\), then multiplying every element of \(S\) by \(\lambda\) preserves order.
Therefore,
\[ \boxed{\sup(\lambda S)=\lambda\sup S}, \]where
\[ \lambda S=\{\lambda x:x\in S\}. \]If \(\lambda<0\), the order reverses, so supremum and infimum exchange roles.
Finite Bounds Versus Extended Bounds
In real analysis we sometimes use the extended symbols
\[ +\infty \qquad\text{and}\qquad -\infty. \]They are useful notation, but they are not ordinary real numbers.
If a set is not bounded above, one may write
\[ \sup S=+\infty \]in the extended-real sense. This is a convention used in advanced analysis, not an application of the real-number completeness axiom.
Example: Finding a Supremum
Consider
\[ S=\left\{1-\frac1n:n\in\mathbb{N}\right\}. \]The first few terms are
\[ 0,\quad\frac12,\quad\frac23,\quad\frac34,\quad\ldots \]Every term is less than \(1\), so \(1\) is an upper bound.
For every \(\varepsilon>0\), choose \(n\) large enough that
\[ \frac1n<\varepsilon. \]Then
\[ 1-\varepsilon<1-\frac1n<1. \]Hence no number below \(1\) can remain an upper bound, and
\[ \boxed{\sup S=1}. \]Example: Finding an Infimum
Let
\[ T=\left\{\frac1n:n\in\mathbb{N}\right\}. \]Every element is positive, so \(0\) is a lower bound.
The Archimedean property tells us that for every \(\varepsilon>0\), there is an \(n\) such that
\[ 0<\frac1n<\varepsilon. \]Thus no positive number can be a lower bound, and
\[ \boxed{\inf T=0}. \]The set has no minimum because \(0\notin T\).
Example: A Set with Both Extremal Values
Let
\[ A=\{-3,1,4,7\}. \]Since this finite set contains its smallest and largest members,
\[ \inf A=\min A=-3 \]and
\[ \sup A=\max A=7. \]Finite sets provide the simplest cases. Completeness becomes essential when infinite sets approach boundary values that may not belong to the set.
Common Error: Confusing Supremum with Maximum
For
\[ S=(0,1), \]it is incorrect to say that the maximum is \(1\).
The correct statements are
\[ \boxed{\sup S=1} \]and
\[ \boxed{S\text{ has no maximum}.} \]Common Error: Requiring the Supremum to Belong to the Set
The definition of supremum requires the supremum to be a real number, but it does not require the supremum to be a member of the set.
For example,
\[ \sup(2,9)=9 \]even though
\[ 9\notin(2,9). \]Common Error: Calling Any Upper Bound the Supremum
For the set
\[ [0,3], \]the numbers \(3\), \(4\), \(10\), and \(1000\) are all upper bounds.
Only \(3\) is the least upper bound:
\[ \boxed{\sup[0,3]=3}. \]Common Error: Forgetting Nonemptiness
The standard least-upper-bound axiom is stated for nonempty sets.
If \(S=\varnothing\), every real number vacuously satisfies the condition of being an upper bound, so there is no least real upper bound.
Extended-real conventions may assign special values in later contexts, but those conventions are separate from the basic completeness axiom.
Common Error: Treating Infinity as a Real Number
The symbol
\[ \infty \]is not an element of \(\mathbb{R}\).
A set such as
\[ [0,\infty) \]is not bounded above in \(\mathbb{R}\), so the least-upper-bound axiom does not produce a real supremum for it.
Common Error: Assuming Boundedness Implies Convergence
A bounded sequence need not converge.
The sequence
\[ a_n=(-1)^n \]is bounded but not convergent.
Completeness gives stronger conclusions when boundedness is paired with additional structure, such as monotonicity or the Cauchy property.
A Reliable Procedure for Supremum Problems
- Identify the set and confirm that it is nonempty.
- Find one or more upper bounds.
- Propose the smallest plausible upper boundary \(u\).
- Verify that every element of the set satisfies \(x\le u\).
- Show that no smaller number can remain an upper bound.
- Use an \(\varepsilon\)-argument when the set approaches \(u\) without reaching it.
- Distinguish clearly between \(\sup S\) and \(\max S\).
A Reliable Procedure for Infimum Problems
- Confirm that the set is nonempty.
- Find one or more lower bounds.
- Propose the greatest plausible lower boundary \(\ell\).
- Verify that \(\ell\le x\) for every \(x\in S\).
- Show that no larger number can remain a lower bound.
- Use an \(\varepsilon\)-argument when the set approaches \(\ell\) from above.
- Distinguish clearly between \(\inf S\) and \(\min S\).
Practice 1: Upper Bounds
Let
\[ S=(-4,6). \]Give three upper bounds for \(S\), and determine \(\sup S\).
Practice 2: Maximum and Supremum
Let
\[ S=[-1,8). \]Determine whether \(S\) has a maximum, and find \(\sup S\).
Practice 3: Infimum and Minimum
Let
\[ T=(2,10]. \]Determine \(\inf T\) and state whether \(T\) has a minimum.
Practice 4: A Sequence Set
Find the supremum and infimum of
\[ A=\left\{\frac{n}{n+1}:n\in\mathbb{N}\right\}, \]using the convention \(\mathbb{N}=\{1,2,3,\ldots\}\).
Practice 5: A Decreasing Sequence Set
Find the supremum and infimum of
\[ B=\left\{2+\frac1n:n\in\mathbb{N}\right\}. \]Practice 6: Monotone Convergence
Suppose \((a_n)\) is increasing and
\[ a_n\le 12 \]for every \(n\).
What does completeness allow you to conclude?
Practice 7: Cauchy Reasoning
A real sequence \((x_n)\) is known to be Cauchy. Must its limit already be known in order to conclude that the sequence converges?
Practice 8: Identify the Error
A student says:
“The set \((0,1)\) has maximum \(1\) because no element of the set is larger than \(1\).”
Explain the error.
Practice 9: Rational Incompleteness
Explain why the set
\[ \{q\in\mathbb{Q}:q>0,\ q^2<2\} \]illustrates a failure of completeness in \(\mathbb{Q}\).
Practice 10: Nested Intervals
Consider
\[ I_n=\left[2-\frac1n,\,2+\frac1n\right]. \]What is
\[ \bigcap_{n=1}^{\infty}I_n? \]Practice Answers
Practice 1
Examples of upper bounds are \(6\), \(7\), and \(100\). The least upper bound is
\[ \boxed{\sup S=6}. \]Practice 2
The set has no maximum because \(8\notin S\). However,
\[ \boxed{\sup S=8}. \]Practice 3
The greatest lower bound is
\[ \boxed{\inf T=2}. \]There is no minimum because \(2\notin T\).
Practice 4
Since
\[ \frac{n}{n+1}=1-\frac1{n+1}, \]the values increase toward \(1\). With \(n\ge1\), the smallest value is \(1/2\). Thus
\[ \boxed{\inf A=\frac12,\qquad \sup A=1}. \]The infimum is attained, but the supremum is not.
Practice 5
The values begin at \(3\) and decrease toward \(2\). Therefore,
\[ \boxed{\sup B=3,\qquad \inf B=2}. \]The supremum is a maximum, while the infimum is not a minimum.
Practice 6
The set of sequence values is nonempty and bounded above, so it has a supremum \(L\). Because the sequence is increasing,
\[ \boxed{a_n\to L}. \]Practice 7
No. Cauchy completeness of \(\mathbb{R}\) guarantees that every real Cauchy sequence converges to some real number even when the limit has not been identified in advance.
Practice 8
Being an upper bound is not enough to be a maximum. A maximum must belong to the set. Since \(1\notin(0,1)\), the set has no maximum, although
\[ \sup(0,1)=1. \]Practice 9
The set is nonempty and bounded above within the rational numbers, but its natural least upper boundary is \(\sqrt2\), which is not rational. Thus the required least upper bound is missing from \(\mathbb{Q}\).
Practice 10
The intervals are nested and their lengths
\[ \frac2n \]approach zero. The unique common point is
\[ \boxed{\bigcap_{n=1}^{\infty}I_n=\{2\}}. \]What to Remember
The real-number completeness principle can be summarized by the statement
\[ \boxed{\text{nonempty + bounded above} \Longrightarrow \text{a real supremum exists}.} \]- An upper bound need not belong to the set.
- A maximum must belong to the set.
- The supremum is the least upper bound.
- The infimum is the greatest lower bound.
- The real numbers are complete; the rational numbers are not.
- Completeness supports monotone convergence, nested intervals, and Cauchy convergence.
- Many later theorems of calculus and analysis ultimately depend on this structure.
Where We Go Next
Completeness tells us that the real number system contains the boundary points and limits needed for analysis. The next step is to study how sequences of numbers and sequences of functions approach limiting objects.
The next lesson, Sequences and Function Convergence, develops pointwise and uniform ideas of convergence and begins connecting completeness to function spaces.