A Geometric Interpretation of the Riemann-Stieltjes Integral
Author(s): Gregory L. Bullock
Reviewed work(s):
Source: The American Mathematical Monthly, Vol. 95, No. 5 (May, 1988), pp. 448-455
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2322483 .
Accessed: 27/05/2012 23:07
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to
The American Mathematical Monthly.
http://www.jstor.org
448 GREGORY L. BULLOCK [May
Lemma 2.2 we can eliminateall of theodd qi's from(2.5). This leaves us with
a2 + nb2 = 2ap
If n = 1 or 2, we can also apply Lemma 2.2 to 2 = 12 + 12 = 02 + 2- 12 to
eliminatefactorsof 2, showingthat p = a2 + nb2,a contradiction.If n = 3, the
case q = 4 of Lemma 2.2 shows that we can reduce to eitherp = a2 + 3b2 or
2p = a2 + 3b2. It remainsto showthatthelattercase cannotoccur.But pla2 + 3b2
implies (- 3/p) = 1, whichby quadraticreciprocitymeans p 1 mod3, so that
2p--2mod3. Yet 2p = a2 + 3b2 implies2p a2 lmod3, and thuswe have a
contradiction.
This completestheproofof Theorem2.1.
This finishesour proofof Fermat'sthreetheorems(0.3), and theonlygap is that
we haven'tprovedquadraticreciprocity. So once a coursein numbertheorycovers.
reciprocity,therereallyare nice applicationswaitingto be proved,and some of
themcan even help motivatethestatementof the theoremitself.
REFERENCES
1. L. Euler, Opera Omnia,seriesI, volume2, Teubner,Leipzig and Berlin,1915.
2. K. Ireland and M. Rosen, A Classical Introductionto Modern NumberTheory,Springer-Verlag,
New York-Berlin-Heidelberg, 1982.
3. T. Nagell, Introductionto NumberTheory,Chelsea,New York, 1981.
4. A. Weil, Number Theory:An ApproachthroughHistory,Birkhauser,Boston-Basel-Stuttgart,
1984.
A GeometricInterpretation
of theRiemann-Stieltjes
Integral
GREGORY L. BULLOCK,
MissionResearchCorporation,
Santa Barbara,CA 93102
1. Introduction.For manystudentsof mathematics, the momentof truthcomes
in a rigorousstudyof real analysis.A promisingstudentmay become discouraged
duringa real analysiscourseand subsequentlyabandon serioushopes of pursuinga
degreein mathematicsand may even dissuadeotherstudentsfromdoing the same.
A graphicalexplanationof some theoremsof analysiscan simplifythe task of an
instructorand may save some studentsfromunwarranted distress.
Herein are presentedgeometricinterpretations of the Riemann-Stieltjes
integral
(or, simply,theStieltjesintegral)and a fewof theassociatedtheoremsas givenin [1,
chapter6]. For convenience,the definitionof the integralis reproduced,and the
theoremsthat are interpreted hereare statedwithoutproof.
2. Definitionof theIntegral.Let [a, b] be--agiveninterval.Define a partitionP
of [a, b] to be a set of pointsxo,xl x. , wherea = xo < X1 < ... < xn =b.
Let a be a monotonicallyincreasingfunctionon [a, b]. For each partitionP of
[a, bI writeLAak = a(Xk)- a(Xk-l)-
For any real functionf thatis bounded on [a, b] let Mk = max{f(x), Xk-l <
x < Xk}, mk = min{f(x), Xkl <x <xk}, and set
n
U(P, f, a)= MkAak,
k=1
1988] THE TEACHING OF MATHEMATICS 449
and
n
L(P, f, a)= E mklak.
k=1
If there is a unique numberI that satisfiesthe inequalityL(P, f, a) ? I <
U(P, f, a) for all partitionsP of [a, b], then I is called the Stieltjesintegralof f
froma to b and is denotedby
fbfda or fb(x) dae(x).
THEOREM 1. Assume a increasesmonotonically and a' is Riemannintegrableon
[a, b]. Let f be a bounded,real, Riemannintegrable
function on [a, b]. Then
fbfda = ff(x)aa(x) dx.
THEOREM 2. If a < s < b, f is boundedon [a, b], f is continuous
at s, and
a(x) = ? x < 5
thenfbfda = f (s).
3. In 3-D. When dependentsets are graphed,each set is assigneda dimension
with a directionperpendicularto thatof the dimensionsof the othersets. So f(x)
and a(x) are graphedas in FIGURES 1 and 2. For this discussion,let a(x) be
differentiable and observethat a(x) is monotonically increasingin agreementwith
the definitionof the Stieltjesintegral.If f(x), a(x), and x are to be considered
simultaneously, theneach is assignedits own dimensionand directionperpendicular
to thatof theothertwo.A keyto findinga geometricinterpretation of theintegralis
to notice that sincef(x) is independentof a(x), thenfora givenx, f(x) mustbe
constantwith respectto a(x). So f(x) is a cylinder(or moreintuitively, a sheet)
thatis straightin the a-direction.If one looks along thesheetin thex-directionone
may see hills and valleys,but one will see only flatterrainin the a-direction(see
FIG. 3). The same resultholdsfora(x), whichis independentof f(x).
f(x)
x
FIG. 1.
450 GREGORY L. BULLOCK [May
a(x)
FIG. 2.
f xx)
a(x)
FIG. 3.
4. The IntegratedFence and its Illuminating Shadow. Assume still that a(x) is
differentiable. If the a-x plane is thoughtof as horizontaland the f-directionas
pointingstraightup, thenthe surfaceto be consideredis like a curvedfence.The
fencefollowsthe curvetracedby a(x), and theheightof thefenceis givenby f(x)
(see FIG. 4). The fenceis reallythesectionof the a-sheetthatis bounded between
the a-x plane and the f-sheet.The Stieltjesintegralintegratesalong this fence,
summingthe productsof heightsand infinitesimal widths.It uses f as theheightin
each interval,but forthedifferential widthit considersonlyLak, thelengthof the
infinitesimal intervalin the a-direction.As a result,thearea givenby theintegralis
reallythe area of theprojectionof the fenceonto thef-a plane (in the same sense
thatthe "area undera curve"is definedas thatgivenby theRiemannintegral).If a
spotlightwith a beam parallel to the x-axis is positionedso as to aim towardthe
fence,thenthe area givenby theintegralis thearea of theshadowof thefenceon a
wall builtin thef-a plane (see FIG. 5).
1988] THE TEACHING OF MATHEMATICS 451
a(x) a(x)
FIG. 4. FIG. 5.
f(x)
"shadow" 4 f n e
a(x) = x
FIG. 6.
to notethatin thisconfiguration,
It is interesting the Riemannintegralis simply
the area of the projectionof the fenceonto the f-x plane. It is easy to see thatif
a(x) is definedby a(x) = x, thenthe fenceallows a straightline fromthe origin
thatmakes an angleof 450 withboththe a-axis and thex-axis.In thiscase thearea
of the projectiononto thef-a plane is equal to the area of the projectiononto the
f-x plane (see FIG. 6). Thus, the Stieltjesintegralreducesto the Riemannintegral.
This can be verifiedby a simpleapplicationof Theorem1.
5. Throwinga Curve. Considerthe case wherea is not a straightline. Define f
as in FIGURE 7 (in threedimensionsthisis a plane), and let a1 be givenby FIGURE
8a and a2 by FIGURE 9a. Let a1 and a2 be differentiable.
The corresponding figures
show the fencesand theirprojectionsonto the f-a plane. A carefulinspectionof
these shadows will convincethe readerthatthe curvein a weightsthe area of the
shadow.The values of x forwhicha(x) has thesteepestslope correspondto regions
452 GREGORY L. BULLOCK [May
f(x)
FIG. 7.
a1(-x)
02(x)
~~~~~~~~~~~~x
FIG. 8a. FIG. 8b.
~~~~~~~~~~~~~~~~f(- (f(A)
"shadow" ,-( a >sX
FIG. 9a. FIG. 9b.
1988] THE TEACHING OF MATHEMATICS 453
of the fencethat cast the most shadow and therebycarrythe most weightin the
integral.While tryingto picturethis,it maybe usefulto observethatregionsof the
fence in which a(x) has zero slope cast no shadow at all. This is the essence of
theidentityfromTheorem1
fid= fa(x)a'(x)dx (1)
where a'(x) = da/dx is simplytheweighting
function.
6. A Gate in theFence. Using thedefinition
a(x) = , x< 5
wheref is continuousat s, thefencehas a hole-or a gate-in it (see FIGS. 10 and
11). The widthof thegateis unity,and theheightof thefenceat thegateis f(s), so
the area of the gate is f(s). That the fencehas no projection(otherthan two line
segments)in the f-a plane only slightlycomplicatesthe analogy to the geometric
interpretation of theintegral.The complicationcan be resolvedby simplyfillingin
the fenceat the hole, in otherwords,closingthe gate door. The projectionof the
gate door has area f(s) whichis what the Stieltjesintegralgives.This agreeswith
Theorem2 above. This analogyis applicablewhenevera(x) is monotonicand has
at least one jump discontinuity.
a(x) .x)
"shadow"
s x a(x)
FIG. 10. FIG. 11.
7. A Sidewise Glance. In the general definitionof the Stieltjesintegral,the
restriction
that a(x) be monotonically increasingis not imposed.For the following
discussion,let the restrictionbe removed,and consider the case where a(x) is
decreasing on some intervals.On intervalswhere a(x) is decreasing,a'(x) is
negative,and the area of thecorresponding projectedregioncontributesnegatively
to theintegralas suggestedby equation(1). The analogyto thegeometricinterpreta-
454 GREGORY L. BULLOCK [May
tion of the integralmay be applied directlyby dividingthe originalintervalinto
subintervalson whicha(x) is monotonic.The shadow of each subintervalmay be
consideredseparately,and thearea of each shadowmaybe added or subtracted(as
determinedby the sign of a'(x) on the subinterval)to determinethe value of the
integralas illustratedin FIGURES 12a-c. It is amusing to observe that in this
configuration, the net regionwhose area is givenby the Stieltjesintegraltypically
assumes unusual shapes (see FIG. 13). In fact,theintegralcan give the sum of the
areas of two (or more) regionsconnectedonly by a point as shown in FIGURES
14a-b.
f(x)
"shadow"
a(x) fence"
FIG. 12a.
f(x) f(x)
a( x
FIG. 12b. FIG. 12c.
1988] THE TEACHING OF MATHEMATICS 455
f(x)~~~~~
a(x)
FIG. 13.
f(x)
f( x)
"fence"
shadow"
a( x) a( x)-
FIG. 14a. FIG. 14b.
8. Conclusion. Geometricinterpretations of the Riemann-Stieltjes
integraland
severalof its associatedtheoremshave been presented.The authorhopes thatthese
in presenting
will aid mathematicsinstructors conceptsassociatedwiththe
difficult
integral.
I am indebtedto Dick Epstein,mymathematics professorat the Universityof California,Berkeley,
I am gratefulto RobertDruckerforhis ever-illuminating
forhis helpfulsuggestionsand encouragement.
observations.
REFERENCE
1. WalterRudin, Principlesof MathematicalAnalysis,thirdedition,McGraw-Hill,New York, 1976.
