Solution to Problem 72-10*: Conjectured monotonicity of a
matrix
Citation for published version (APA):
Lossers, O. P. (1973). Solution to Problem 72-10*: Conjectured monotonicity of a matrix. SIAM Review, 15(2), 387-390. https://doi.org/10.1137/1015045
DOI:
10.1137/1015045
Document status and date: Published: 01/01/1973
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
providing details and we will investigate your claim.
Thus for
a< t thereis a unique minimum,y(x) 0;
a > r thereis nominimum,i.e. infJ(y)
-while for
a 7 the functions y(x)= b sin
(rex/a)
satisfy J(y)= 0 (for all b,- < b< oe)and0 infJ(y).
Alsosolvedinessentiallythe samewaybyK.PARI,:(NASA,LangleyResearch
Center), O. RtJEHR(Michigan Technological University) and the proposer. How-ever, only the proposer showed that J has no maxima, by observing that if
(I),(x)
sin(nnx/a),
thenJ((I),)
(a/2)((nn/a)
2 1)-,o as n
,
for alla > 0.II. SolutionbyR. T. SHIELD(University ofIllinois).
The substitution y u sin
(nx/a) (cf.
Jacobi’s method of multiplicative vari-ation[1])
is permissiblein viewofthe properties of y(x). The integral becomesand theextrema fora < n, a nanda > ncanbedeterminedby inspection.
REFERENCE
[1] R. COtJRANa"ANDD.HILBERT,MethodsofMathematicalPhysics,vol.I,Interscience,NewYork,
1953,p.458.
Problem72-10", ConjecturedMonotonicity
of
aMatrix,by R.A.
USMANI (Univer-sity of Manitoba, Manitoba, Winnipeg, Canada).Consider A
(aij),
afive-band matrix of order N suchthat5 +
i, i=j=l,N,6 +/;i, =j 2,3,
-4,
li-Jl-
1, 1, li-jl-- 2, 0, otherwise,wherei 0. Itis conjectured that the matrix Ais monotone,i.e., if
A-1
(aj),
then
a/>
0for all and j.Remark. Itisknown thatthe matrix
A0
obtainedfrom A by setting:i 0 is monotone. ForA0
jz,
where J(Jmn)
such thatm n,
Jrnn
1,Im
nl
1,and if
J-x
(j,), then n(N+
m)(u+ 1)
m(N
+
n) (N+ 1) > O, m_>-
n, >0,m<n.
ThusJandconsequently
A0
are monotonematrices.ThematrixA arisesincom-putation of thenorm oftheinverseofafourth orderdifferentialoperator.
Solutionby O. P.LOSSERS (Technological University, Eindhoven, the Nether-lands).
The conjecture is false. Let el 2 N 2,2 > 0, and let A A 2I; then
A-X
2-(I
+
2-A)
-
2-(I
2-XA
+
R), where Rij---O(,-2),
for all and j(2- o).
Thena.*.,,,+2
2-1(6i,i+2
,,,-
_it_O(2-2))
/-2+
O(/-3),
which is less than 0 for sufficiently large 2.Solution by A. A.JAGERS(TwenteUniversity ofTechnology, Enschede, the Netherlands).
A
straightforward calculation ofA-
for N__<
4 shows (i) that A is alwaysmonotoneifN
__<
2,(ii) that,for N 3,Ais monotone ifand onlyif;2-
10 and(iii) that, for N 4, A is monotone if and only if -2
--
;3 5. For larger N the conditions becomemore complicated. Anyhow weshall see later that, ifN>=
3, thena3
< 0,as soon as ei 0 forall 4= 2 and ;2 islarge enough. So, ifN>=
3,the conjecture does not even hold if we restrict ourselves to the casewhere only
one ofthe :i ispositive. Itis worthwhile to consider thisphenomenon inamore
generalsetting"
Let B
(bij)
be a matrix of order N with det(B)4= 0. Choose a fixedk, 1
__<
k__<
N. LetB(e.k)=
(bi(Zk))
be the matrix given bybkk(k)=
bkk +
k andbij(e,k)
bij
if 4= kor j 4: k. Denotethe ijth entry ofthe inverse matrix of byb(e
k)
andb(0)
byb.
Supposethatbk*
> 0. Then we have the followingtwopropositions.
An
unconventional but easy way of provingthe second part ofthe first propositionis by successively solving the N2 differential equationsb(,
)b(:).
*" The second propositionis aconsequence of thefirst.PROPOSITION 1.
(i) det
(BO:k))
(1+ bk*"
ok)
det(B)
4= O.(ii)
+ bk*.c
k* * h,b,
with
Aikj
bkkbij
ik kj"PROPOSITION 2.
If
Bismonotone,thenB(ek)
ismonotone]br
all[;k 0if
andonlyNotethat
Aikk
Aj
0.Now adirectcalculation usingthegivenJm,"* shows thatin theoriginal prob-lemthe quantity correspondingto
A3
is equal to1 N
’j---’3
(N
+
1 k)2(N
+
1)2which isless than0. Thisproves theannouncedstatementconcerning
a’3.
On the other hand,it follows at once fromthe corollary to the proposition below,that atleast all diagonalentriesa/are
positive ifei >0 for all i.PROPOSITION 3.
If
B is symmetricandpositivedefinite, thenb*,i(ek)
> 0for
all andr.>=
O.Proof
Since B is symmetric and positive definite, a symmetric matrix Mexists such that B
m
2.
Then the conditionAik/
_>_
0 is equivalent to Schwarz’s inequality appliedtothe ithand kth columnvectorofM-1.
Now
B(ek)
is again symmetric and positive definite. (Note that det(B(e.k)
4:0forall :k0.)
SowemayapplyProposition3another time,this timechanging another diagonal entry,etc. Wefinallyobtainthe followingcorollary.COROLLARY.
If
B is symmetric and positivedefinite, then the diagonalentriesof
the inverse C-1of
the matrix C obtainedfrom
B by adding l,i(O) tO the ithdiagonal entry(i 1,2,...,N) are all positive.
Solutionby A. N. WILLSON, JR.(University of California, Los Angeles). The stated conjectureis false. This is easily verified by considering the case in which N 3, with r. 11, i= 1,2,3. In this case the elements
(A-1)13
(A-X)3
< 0. Also,it isclearthat the conjecture cannotbe trueforany matrixof the type specified having order N > 2, as it would contradict the following theorem (which is proved in
[1]).
THEOREM. ThematrixA issuch that
(A
+
D)-
>__
0for
everydiagonalmatrixD having positive elements on the main diagonal only
if
the off-diagonal elementsof
A are all nonpositive.REFERENCE
Ill A. N.WILLSOY, JR.,Pairsofmatricesandmatricesofmonotonekind,SIAMJ.Numer.Anal., 10
(1973).
Also solvedby L.ELSNER (InstitutftirAngewandteMathematik
I,
Erlangen-Nurnberg), I.FARICAS (University ofToronto,
Canada), M. J. MARSDEN (Univer-sity of Pittsburgh), C. C. ROUSSEAU (Memphis StateUniversity), O. G. RUEHR (Michigan Technological University),J.N. SHOOSIITH (NASA, LangleyResearch Center), R.A.
SWEET (National CenterforAtmosphericResearch),J. J.VANDER-CRAFT(UniversityofMaryland), L. RUBENFELD(RensselaerPolytechnic Institute) submittedasolutionco-authoredbysixNSF undergraduateresearch participants: M.
FRAME,
D.HALE,
C.KELLER,
D. RUTLEDGE, S. SENTOFFandD. WAGNER.By [1],
it is sufficient to construct a 2m+x 2m+1(__
1,1) matrix W for which det W- 2(’/1)2". Let A denote the 2 x 2 matrix and defineW= (R)"
+IA,
the (m+ 1)th Kronecker power of A. W satisfies detW(det
A)(’/x)2- 2(./2-which provides a desired matrix.
Problem 72-11,Limit
of
anIntegral,by N. MULLINEUXandJ. R. REED(University ofAston,
Birmingham,England).Determine
! lim log
Isin
t-13/21
dt-o sin
13/2
sinThe problem arose in the numerical inversion of singular integral equations
such asthose which occur inaerodynamics and problems associated with plane
transmission lines.
Editorial note. Most solvers
(see
laterdiscussion) noted that the integral asgiven above does not exist since for fixed e, 0, the integrand behaves like
-
log{(13/2)sin
13/2}
as 0. The error is a typographical one since in the original proposal the termIsin
t/2l
appears asIsin
(t
13/2)1.
In the following solution ofO. P. LOSSERS (Technological University of Eindhoven, Eindhoven,the
Netherlands)
the latter expression is used. Replacing 13 by 213 we getlog Isin(t 13)1 dt
,2
sin13 sin 0
dt log
Icos
cot13.sintl
sin
logcost---
+
10gll cot13.tantl
sn sn
Since (logcost)/sin
t/2
+
O(t3)
(t
0), we have that the first term isO(132),
(13
0).By
substitution ofv cot13.tan we getfcote.tan2e
logl1
vl
I lim O(132 --e--,0 d0(1
+
/)2tan213)-1/2
21ogll_/)ldv=
v 4In the solutions by N. M. BLACHMAN (Sylvania Electronics Systems), L. KHATCHATOORIANTZ (California State College, Los Angeles), S. L. PAVERI-FONTANA(University of California,Berkeley), Z. C. MOTTELERandO. G.RUEHR (Michigan Technological University),J. W. RIESE(Kimberly-ClarkCorporation), J.D.TALMAN(University of
Western
Ontario,London,Canada)andP. TH. L. M.VANWOERKOM (National
Aerospace
Laboratories,Amsterdam,theNetherlands),it was remarked that the integral as originally stated does not exist. Various
alternative statementsof the problem weremade. We donot listthem all.