You already know how to find areas of simple shapes — rectangles, triangles, circles. But what is the area under the curve y=x2 from 0 to 1? No geometric formula applies directly. The Riemann integral is the answer: it defines “area under a curve” as a limit of increasingly fine rectangular approximations, and does so in a way that is precise enough to prove theorems.
Partitions and Darboux sums
Let f:[a,b]→R be a bounded function. A partition of [a,b] is a finite sequence
P={a=x0<x1<x2<⋯<xn=b}.
On each subinterval [xi−1,xi], f is bounded, so its supremum and infimum exist:
Mi:=x∈[xi−1,xi]supf(x),mi:=x∈[xi−1,xi]inff(x).
The upper Darboux sum and lower Darboux sum of f over P are:
U(f,P):=i=1∑nMi(xi−xi−1),L(f,P):=i=1∑nmi(xi−xi−1).
Geometrically, U(f,P) is the total area of rectangles that lie above the graph, and L(f,P) is the total area of rectangles that lie below it. Since mi≤Mi for every i, you always have L(f,P)≤U(f,P).
Refining a partition
A partition Q is a refinement of P if P⊆Q (every boundary point of P is also in Q). Adding points can only lower the upper sum and raise the lower sum:
P⊆Q⟹L(f,P)≤L(f,Q)≤U(f,Q)≤U(f,P).
A consequence is that every lower sum is at most every upper sum, even for different partitions: if P1 and P2 are any two partitions, then L(f,P1)≤U(f,P2).
Upper and lower integrals
Because the set of all lower sums is bounded above (by any upper sum), its supremum exists in R. Likewise, the infimum of all upper sums exists. These define the lower integral and upper integral of f:
∫abf:=PsupL(f,P),∫abf:=PinfU(f,P).
Since every lower sum is ≤ every upper sum, you always have ∫f≤∫f.
Riemann integrability
A bounded function f:[a,b]→R is Riemann integrable if the upper and lower integrals agree:
∫abf=∫abf.
When they do, their common value is called the Riemann integral of f on [a,b], written
∫abf(x)dx=∫abf.
The Darboux criterion
A practical reformulation: f is Riemann integrable if and only if for every ε>0 there exists a partition P with
U(f,P)−L(f,P)<ε.(1)
This is the working tool for most integrability proofs.
Equivalent Riemann sum definition
You can also define the integral via Riemann sums directly. For a partition P and a choice of sample points ξi∈[xi−1,xi], the Riemann sum is
S(f,P,ξ):=i=1∑nf(ξi)(xi−xi−1).
The mesh of P is ∥P∥:=maxi(xi−xi−1). Then f is Riemann integrable with value I if and only if: for every ε>0 there exists δ>0 such that ∥P∥<δ implies ∣S(f,P,ξ)−I∣<ε for every choice of sample points.
The Darboux and Riemann-sum formulations are equivalent; the Darboux approach tends to be cleaner for proofs.
Every continuous function is integrable
Theorem. If f:[a,b]→R is continuous, then f is Riemann integrable.
Proof. A continuous function on a closed bounded interval is uniformly continuous: for every ε>0 there exists δ>0 such that ∣x−y∣<δ⇒∣f(x)−f(y)∣<ε/(b−a).
Given ε>0, choose δ as above and let P be any partition with mesh ∥P∥<δ. On each subinterval [xi−1,xi], the oscillation satisfies
Mi−mi≤b−aε,
because any two points in [xi−1,xi] are within distance δ of each other. Therefore
U(f,P)−L(f,P)=i=1∑n(Mi−mi)(xi−xi−1)≤b−aε⋅(b−a)=ε.
By the Darboux criterion (1), f is integrable. □
Worked example: ∫01x2dx
Take f(x)=x2 and the uniform partition Pn={0,n1,n2,…,1}. Since f is increasing on [0,1]:
Mi=(ni)2,mi=(ni−1)2,Δxi=n1.
The upper and lower sums are:
U(f,Pn)=n1i=1∑nn2i2=n31⋅6n(n+1)(2n+1)=6n2(n+1)(2n+1),
L(f,Pn)=n1i=0∑n−1n2i2=6n2(n−1)(2n−1).
Their difference is U−L=6n2(n+1)(2n+1)−(n−1)(2n−1)=6n26n=n1→0. So the Darboux criterion is satisfied. Both sums converge to 31, confirming
∫01x2dx=31.
Summary
- A partition of [a,b] divides it into subintervals; Darboux sums L(f,P) and U(f,P) bound the “area under f” from below and above.
- The lower integral ∫f and upper integral ∫f are the supremum and infimum of all lower and upper sums, respectively.
- f is Riemann integrable when ∫f=∫f; the common value is ∫abf.
- Darboux criterion: f is integrable iff for every ε>0 some partition achieves U(f,P)−L(f,P)<ε.
- Every continuous function on [a,b] is integrable, by uniform continuity.