Scribd
Upload
8 views3 pages

An Algorithm-Inspired Proof of the Spectral Theorem in En

The document presents an algorithm-inspired proof of the spectral theorem for real symmetric matrices by Herbert S. Wilf, published in The American Mathematical Monthly. It discusses the method of Jacobi's numerical approach to find eigenvalues and eigenvectors, emphasizing the minimization of the sum of squares of off-diagonal elements. Additionally, it touches on the efficient computation of mathematical constants like √2 and π, comparing their iterative sequences and convergence rates.

Uploaded by

tentimarc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views3 pages

An Algorithm-Inspired Proof of the Spectral Theorem in En

The document presents an algorithm-inspired proof of the spectral theorem for real symmetric matrices by Herbert S. Wilf, published in The American Mathematical Monthly. It discusses the method of Jacobi's numerical approach to find eigenvalues and eigenvectors, emphasizing the minimization of the sum of squares of off-diagonal elements. Additionally, it touches on the efficient computation of mathematical constants like √2 and π, comparing their iterative sequences and convergence rates.

Uploaded by

tentimarc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Mathematical Association of America

An Algorithm-Inspired Proof of the Spectral Theorem in E n


Author(s): Herbert S. Wilf
Source: The American Mathematical Monthly, Vol. 88, No. 1 (Jan., 1981), pp. 49-50
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2320714
Accessed: 20-10-2015 04:08 UTC

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 support@jstor.org.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to The American
Mathematical Monthly.

http://www.jstor.org

This content downloaded from 129.97.58.73 on Tue, 20 Oct 2015 04:08:56 UTC
All use subject to JSTOR Terms and Conditions
1981] MATHEMATICAL NOTES 49

THEOREM 4. fE( 6, f'= O a.e.


For mostfunctions
Lipschitzmaps.We considerthosefunctions
Analogousphenomenahappenforfirst-order
f EiC([O,1]) forwhich
a < f(y)- f(x) <A
y-x
forall pairs of distinctpoints x,y in [0,1], a and /8
being fixedreal numbers(a <,8). Let ctr, ,B be
the family of all such functions; obviously c.V, c :T (max{lal, ,/3 I). T, is of the second
We getanalogously:
category.
fEE
5. For mostfunctions
THEOREM we have,at eachpointx E (0,1],
ft-(x) = a or f,-(x)
and,at eachpointx EE[0,1),
f +(x)=a or f,+(x)=/3.

f E 'W.,, theset
6. For mostfunctions
THEOREM
f'-'(a) Uf' '(/3)
has measure1.
References
des fonctions
et la recherche
Le9onssurl'integration
1. H. Lebesgue, primitives, Paris,1904.
Gauthier-Villars,
2. H. Minkowski, Zur Geometrie der Zahlen,Verhandl.III Intern.Math.-Kongresses (1904)
Heidelberg
164-173.
singular
continuous
3. L. Takacs,An increasing thisMONTHLY,85 (1978)35-37.
function,

AN ALGORI TM-INSPIRED PROOF OF THE SPECTRAL THEOREM IN ET

HERBERT S. WILF

ofPennsylvania,
ofMathematics,University
Department PA 19104
Philadelphia,

THEOREM. matrix,
If A is a realsymmetric matrixQ suchthatQ TAQ
thereis a realorthogonal
is diagonal.
Of course,thisis thespectraltheorem. It impliesthattheeigenvaluesare real,thatthereis a
pairwiseorthogonalcompleteset of eigenvectors-namely, the columnsof Q-and thatthe
dimensionof an eigenspaceis equal to thealgebraic of theeigenvalue.
multiplicity
Many proofsgrapplewiththe questionof findingenoughindependent eigenvectorsfora
multipleeigenvalue, usuallyone at a time.We shallfindthewholematrixQ at oncebyusingthe
main idea of Jacobi'snumericalmethodforcalculatingthe eigenvaluesand vectors,together
witha littlecompactness.
For an n x n realmatrixA we shalluse Od(A) forthesumof thesquaresof theoff-diagonal
elementsofA, and 0(n) willdenotetheset (group)of n x n orthogonal matrices.
Supposewe can provethefollowing.
LEMMA. real symmetric
If A is a nondiagonal matrixJ
matrix,thenthereis a real orthogonal
suchthatOd(JTAJ)< Od(A).
Thenthetheorem wouldfollowquickly,forletA be realand symmetric. Considerthemappingf
thatsendsan orthogonalmatrixP intof(P) = P TAP. For fixedA thisis a continuousmapping
of 0(n), a compactset, and so f(O(n)) is compact.Let f(Q)= D be a point at whichthe
continuousfunctionOd attainsits minimumvalue on theimageset off. This value mustbe

This content downloaded from 129.97.58.73 on Tue, 20 Oct 2015 04:08:56 UTC
All use subject to JSTOR Terms and Conditions
50 L. BAXTER [January

zero,else D could play therole of A in theabove lemma,and so it would not minimizeOd,


concludingtheproof.
It remainsto provethelemma.This was done by Jacobi,in his celebratedmethodof plane
rotationsforcomputing eigenvaluesand eigenvectors,as follows:Supposeapq#0, p #q. Then
take forJ the matrixthatagreeswiththe nX n identitymatrixexceptin the fourpositions
(p,p),(p, q),(q,p),(q, q), wherethe entriesare cos9, sinG,- sinG,cos9, and the real angle 9 is
chosento make (JTAJ)pq =-. It is easy to check thatOd(JTAJ)= Od(A) - 2akq,and we are
finished.
The proofreadilygeneralizesto thecomplexHermitiancase.
bytheNationalScienceFoundation.
Researchforthispaperwas supported

UNSOLVED PROBLEMS
EDITED BY RICHARD GuY

In this departmentthe MONTHLY presentseasily stated unsolvedproblemsdealing withnotionsordinarily


encountered mathematics.
in undergraduate Each problemshouldbe acconmaniedby relevantreferences (ifany
of knownpartial results.Manuscriptsshouldbe sent to
are knownto the author)and by a briefdescription
RichardGuy,Department of Calgary,Calgary,Alberta,Canada,
ofMathematicsand Statistics,The University
T2N 1N4.

ARE -r,e, AND V2 EQUALLY DIFFICULT TO COMPUTE?

L. BAXTER
Science,YorkUniversity,
ofComputer
Department Ontario,Canada
Downsview,

The generalproblemis to findefficient methodsforcalculatingfundamental mathematical


constants.The simplestof methodsis chosen,namely,iterativesequencesexpressedusingthe
operationsof +, -, X, ?, and \/.Whereasiterative sequencesexistwhichshowthatthetime
to computen digitsof V2 and 'r are essentiallythe same, O(n logn loglogn), no equally
efficientmethodof computing e is known.
The efficient computation of V/2and 'r comparedto thatofe dependson twofactors:rateof
convergence and the complexity of multiplication.The iterativesequencesconvergequadrati-
cally,i.e., thenumber of significant doubles
digits with each iteration.
Consequently thetimeto
computen digitsof theconstantis essentially thetimeto perform Since this
thelast iteration.
involvesmultiplications, theSchonhage-Strassen algorithm is used whichmultiplies two n digit
numbersin 0(n logn loglogn)time[1, Section7.5]. (Note thatdivisionor extraction of square
roots also has this complexity.)If eitherthe convergenceis linear or the classical 0(n2)
multiplication algorithmis used, thenthe timeto computen digitsis at least 0(n2), which
theoretically is as bad as summing theappropriate series.
Specifically, Newton's sequence =
xi '(xi_1 2/xe 1) or the more convenientxi =
+
xi-1(6-x x2_)/4 converge quadratically to V2. These methodshave essentiallythe same
complexity as thatbased on continuedfractions [2]. Vieta'sformulafor7r,
2 2~~~ I ..
A/I* + I /2 /2+ 2 /2+ A\/2))

can be definedby thetwoiterative


sequences
xi = IY
Illl+}=2 . x.=n' Y=1,

This content downloaded from 129.97.58.73 on Tue, 20 Oct 2015 04:08:56 UTC
All use subject to JSTOR Terms and Conditions

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy