Beruflich Dokumente
Kultur Dokumente
StructuralAnalysis
DevdasMenon
Professor
IITMadras
(dmenon@iitm.ac.in)
NationalProgrammeonTechnology
EnhancedLearning(NPTEL)
www.nptel.ac.in
Lecture17 Module3:
BasicMatrixConcepts
AdvancedStructuralAnalysis
Modules
1. Reviewofbasicstructuralanalysis 1(6lectures)
2. Reviewofbasicstructuralanalysis 2(10lectures)
3. Basicmatrixconcepts
4. Matrixanalysisofstructureswithaxialelements
5. Matrixanalysisofbeamsandgrids
6. Matrixanalysisofplaneandspaceframes
7. Analysisofelasticinstabilityandsecondordereffects
Module3:BasicMatrixConcepts
Reviewofmatrixalgebra
Introductiontomatrixstructuralanalysis(forceand
displacementtransformations;stiffnessand
flexibilitymatrices;basicformulations;equivalent
jointloads).
Reviewof
Basic
Conceptsin
Structural
Analysis
Matrix
Concepts
and
Methods
Structures
withAxial
Elements
Beams
andGrids
Planeand
Space
Frames
Elastic
Instability
andSecond
order
Analysis A
d
v
a
n
c
e
d
S
t
r
u
c
t
u
r
a
l
A
n
a
l
y
s
i
s
1INTRODUCTION
2MATRIX
3VECTOR
4ELEMENTARYMATRIXOPERATIONS
5MATRIXMUTLITPLICATION
6TRANSPOSEOFAMATRIX
MatrixConceptsandAlgebra
7RANKOFAMATRIX
8LINEARSIMULTANEOUSEQUATIONS
9MATRIXINVERSION
10EIGENVALUESANDEIGENVECTORS
MatricesinStructuralAnalysis
LOADS
(input)
STRUCTURE
(system)
RESPONSE?
(output)
( ( (
=
( ( (
( ( (
A AA AR A
R RA RR R
F k k D
F k k D
LoadVectorF
A
SupportDisplacementsD
R
InitialDeformationsD
*in
DisplacementVectorD
A
SupportReactionsD
R
InternalForcesF
*
MemberDeformationsD
*
Analysissoftwarepackages
Introduction
Thereisanincreasingtendencyamongmodernstructural
engineerstoleanheavilyonsoftwarepackagesfor
everything.
Thisinducesafalsesenseofknowledge,securityandpower.
Thecomputerisindeedapowerfultoolandanassetforany
structuralengineer.Itisdangerous,however,tomakethe
toolonesmaster,andtomakeitaconvenientsubstitutefor
humanknowledge,experienceandcreativethinking.
Ref.: Preface to Advanced Structural Analysis
By definition,amatrix isrectangulararrayofelementsarrangedinhorizontal
rowsandverticalcolumns.
Theentriesofamatrix,calledelements,arescalarquantities commonly
numbers,butmayalsobefunctions,operatorsorevenmatrices(calledsub
matrices)themselves.
A= A
= A
mn
= a
ij
= a
ij
mn
=
a
11
a
12
a
13
L a
1n
a
21
a
22
a
23
L a
2n
a
31
a
32
a
33
L a
3n
M M M M
a
m1
a
m2
a
m3
L a
mm
(
(
(
(
(
(
(
order
m= n
square matrix
a
ij
=0i , j
null matrix, O
identity matrix, I
a
ij
=0: i = j
a
ij
=1: i = j
diagonal elements
1 0 0
0 1 0
0 0 1
(
(
(
Matrix?
square matrix
symmetric matrix (m = n = 6)
a
ij
=a
ji
( )
banded matrix
sparse matrix
a 0 0
d b 0
e f c
(
(
(
a d e
0 b f
0 0 c
(
(
(
lower triangular matrix, L upper triangular matrix, U
sub-matrices
Type of matrix?
4 -1
-1
-2
-2
9
9
5
7
4 -1
-2
-2
-1
1
1
0
0
0
0
0
0
0
0
0
0 0
0
0
0
0
0
0
0 0
0
A= a
ij
mn
=
b
ij
i p
c
ij
l np
( )
d
ij
ml
( )
p
e
ij
ml
( )
np
( )
(
(
(
(
mn
=
B C
D E
(
Partitioning:
Avectorisasimplearrayofscalarquantities,typicallyarrangedinavertical
column.
Hence,thevectorcanbevisualisedasamatrixoforderm1,wherethe
numbermiscalledthedimensionofthevector.Thescalarentriesofavector
arecalledcomponentsofthevector.
V= V
{ }
m
= v
i
{ }
= V
{}
m1
= v
ij
{ }
m1
=
v
1
v
2
v
3
M
v
m
Whatisavector?Isitatypeofmatrix?
a
1
a
2
a
3
L a
n
(
row vector
Wecanvisualizeamultidimensionallinearvectorspace,
m
,whose
dimensionmisgivenbytheminimumnumberoflinearlyindependent
vectors(withrealcomponents)requiredtospanthespace.
Vectorsaresaidtospanavectorspace,ifthespaceconsistsofallpossible
linearcombinations ofthosevectors.Anysetofvectorsthatarelinearly
independentandalsospanthevectorspaceiscalledabasisofthevector
space.
0
x
y
z
V
V=
2
1
3
(
(
(
canbevisualisedin
3
vectorspaceas
V=2
j +3
k
unitvectors
1
0
0
(
(
(
'
0
1
0
(
(
(
and
0
0
1
(
(
(
provideanorthogonalbasisin
3
vectorspace.
havingamagnitudeorlength,
V = v
i
2
i =1
m
= 14
Asetofvectors,{V
1
,V
2
,,V
n
},havingthesamedimensionm,issaidtobe
linearlyindependentifnolinearcombinationofthem(otherthanthezero
combination)resultsinazerovector;i.e.,c
1
= c
2
= = c
n
= 0,if.
c
i
V
i
i =1
n
=0
ELEMENTARYMATRIXOPERATIONS
ScalarMultiplication:
A= a
ij
mn
= a
ij
mn
= A
MatrixAddition:
A+B= a
ij
mn
+ b
ij
mn
= a
ij
+b
ij
mn
AB= a
ij
mn
+ 1
( )
b
ij
mn
= a
ij
b
ij
mn
A+B=B+A
A+ B+C
( )
= A+B
( )
+C
MatrixMultiplication:
a
ij
mn
b
ij
np
= c
ij
mp
AB =C;i.e.,
L L L L L
L L L L L
a
i 1
a
i 2
a
i 3
L a
in
L L L L L
L L L L L
(
(
(
(
(
(
mn
M M b
1 j
M M
M M b
2 j
M M
M M M M M
M M b
nj
M M
(
(
(
(
(
np
=
M
M
L L c
ij
L
M
M
(
(
(
(
(
(
mp
i
th
row
j
th
column j
th
column
i
th
row
A B+C
( )
= AB+AC
A BC
( )
= AB
( )
C
B+C
( )
A=BA+CA
AO= O
AI= A
2 1
3 4
1 2
(
(
(
32
1 4
2 0
(
22
=
0 8
11 12
5 4
(
(
(
32
mn
B
np
C
mp
2
3
1
(
(
(
1
+
1
4
2
(
(
(
2
=
EverycolumnvectorofC isalinear
combinationofthecolumnvectorsof
thepremultiplyingmatrixA!
= 3
1 4
+ 4
2 0
EveryrowvectorofC isalinear
combinationoftherowvectorsofthe
postmultiplyingmatrixB!
Matrix multiplication operation does not
possess the property of commutativity;
i.e, in general, AB BA
TRANSPOSEOFAMATRIX
Transposition is an operation in which the rectangular array of the matrix is rearranged
(transposed) such that the order of the matrix changes fromm n to n m, with the
rows changed into columns, preserving the order. If the original matrix is A =[a
ij
]
mn
,
then the transpose of A, which is denoted as A
T
, is given by A
T
= [a
ij
]
T
= [a
ji
]
nm
.
Transposition is an operation in which the rectangular array of the matrix is rearranged
(transposed) such that the order of the matrix changes fromm n to n m, with the
rows changed into columns, preserving the order. If the original matrix is A =[a
ij
]
mn
,
then the transpose of A, which is denoted as A
T
, is given by A
T
= [a
ij
]
T
= [a
ji
]
nm
.
(A
T
)
T
=A
(A)
T
=A
T
(A+B)
T
=A
T
+B
T
(AB)
T
=B
T
A
T
S
T
=A
T
(A
T
)
T
nm
T
A
mn
= S
nn
Square matrix
=A
T
A=S
1 2 3
2 0 1
(
(
1 2
2 0
3 1
(
(
(
=
14 1
1 5
(
(
Symmetric matrix
(s
ji
= s
ij
)
A
T
=A
T
(i.e., a
ji
= a
ji
) Skew - Symmetric
Theproduct{F}
T
{D}= resultsinamatrixoforder1 1,whichisnothing
butascalar.Suchaproduct,whichissometimesdenotedas,is
calledinnerproduct (dotproductinvectoralgebrainvolvingtwo and
threedimensionalvectors).Thisproductessentiallyreflectstheprojected
lengthofonevectoralongthedirectionoftheothervector.The
magnitudeofanyvectorV,definedasmaybeviewedasthe
innerproductofVwithitself,i.e.,.
Theproduct{F}
T
{D}= resultsinamatrixoforder1 1,whichisnothing
butascalar.Suchaproduct,whichissometimesdenotedas,is
calledinnerproduct (dotproductinvectoralgebrainvolvingtwo and
threedimensionalvectors).Thisproductessentiallyreflectstheprojected
lengthofonevectoralongthedirectionoftheothervector.The
magnitudeofanyvectorV,definedasmaybeviewedasthe
innerproductofVwithitself,i.e.,.
F,D
V = v
i
2
i =1
m
V,V = V
T
V
orthonormal
vectors
X
i
T
X
j = o
ij
=
1 if i = j
0 if i = j
D,F = D
{ }
T
k
D
{ }
=
D,F
T
= D
{ }
T
k
D
{ }
|
\
|
.
T
= D
{ }
T
k
T
D
{ }
=
[k] is symmetric !
Commutative property
of inner product:
F
{}
T
D
{ }
= D
{ }
T
F
{}
=
F
{}
m1
= k
mm
D
{ }
m1
Therankofthematrix[A]isequaltothenumberoflinearly
independentcolumnvectorsofthematrix,andthisnumberisidentical
tothenumberoflinearlyindependentrowvectors.
Themaximumvalueoftherankr ofanymatrixofordermn isgiven
bymorn(whicheverislower),andtheminimumvalueis1.
Linearsimultaneousequations:
a
ij
X
j
j =1
n
= c
i
i = 1,2,.....,m
( )
mn
X
{ }
n1
= C
{ }
m1
coefficientmatrix
RANKOFAMATRIX
Thesubspaceinthevectorspace
m
containingalllinearcombinations
oftheindependentcolumn(orrow)vectorsiscalledthecolumnspace
(orrowspace)ofA,andthissubspacehasadimensi0nequaltorankr.
A=
1 2 3
2 1 3
3 1 4
4 2 6
(
(
(
(
(
Rank of A = 3 or 2 or 1 ?
Sum of first two columns! Rank = 2.
C=O
(homogeneous
equations)
nullspaceofA
(allpossible
solutionsofX)
RowReducedEchelonForm
mn
=
I
r r
F
r nr
( )
O
mr
( )
r
O
mr
( )
nr
( )
(
(
(
Arelativelyeasyandcertainwayofdeterminingtherankofamatrixisby
reducingthematrixtoarowreducedechelonformR throughaprocessof
elimination(transformingA ascloselyaspossibletoanidentifymatrixI inthe
upperleftcorner).
Free variable
coefficient matrix
A =
2 4 6
2 1 3
3 1 4
4 2 6
(
(
(
(
(
1 2 3
0 1 1
0 0 0
0 0 0
(
(
(
(
(
pivot
1 2 3
0 3 3
0 5 5
0 6 6
(
(
(
(
(
1 0 1
0 1 1
0 0 0
0 0 0
(
(
(
(
(
=R =
I F
O O
(
Rank r = 2
2 4 6
2 1 3
3 1 4
4 2 6
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
c
1
c
2
c
3
c
4
(
(
(
(
(
AX=C:
LINEARSIMULTANEOUSEQUATIONS
AX=C
r r
F
r nr
( )
O
mr
( )
r
O
mr
( )
nr
( )
(
(
mn
X
pivot
r 1
X
free
nr
( )
1
(
(
(
n1
=
D
pivot
r 1
D
zero
mr
( )
1
(
(
m1
pivotvariables pivotrowconstants
freevariables zerorowconstants
X
pivot
{ }
= D
pivot
{ }
F
X
free
{ }
X = X
p
+ X
n
D
zero
{ }
= O
{ }
2 4 6
2 1 3
3 1 4
4 2 6
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
c
1
c
2
c
3
c
4
(
(
(
(
(
AX=C
RX=D
1 0 1
0 1 1
0 0 0
0 0 0
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
c
1
+4c
2
( )
6
c
1
c
2
( )
3
c
3
+ c
1
10c
2
( )
6
2c
2
+c
4
( )
(
(
(
(
(
(
RX=D
zero
zero
c
3
=
10c
2
c
1
6
c
4
= 2c
2
particular + null space solution
2 4 6
2 1 3
3 1 4
4 2 6
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
a
b
(10ba) / 6
2b
(
(
(
(
(
AX=C
2 4 6
2 1 3
3 1 4
4 2 6
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
0
3
5
6
(
(
(
(
(
1 2 3
0 3 3
0 5 5
0 6 6
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
0
3
5
6
(
(
(
(
(
1 0 1
0 1 1
0 0 0
0 0 0
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
2
1
0
0
(
(
(
(
(
For a feasible solution space
X
pivot
{ }
= D
pivot
{ }
F
X
free
{ }
D
zero
{ }
= O
{ }
Particularsolution:LetX
3
=3
X
1
= 2 X
3
X
2
= 1 X
3
X
1
X
2
X
3
(
(
(
p
=
2
1
0
(
(
(
Nullspacesolution(C =O);LetX
3
=
X
1
X
2
X
3
(
(
(
n
=
(
(
(
Complete
solution:
X =X
p
+X
n
X
1
X
2
X
3
(
(
(
n
=
2
1
+
(
(
(
CoefficientmatrixAhasafullcolumnrank(r=n),buttherearelinearly
dependentrows (r<m),whichalsoimpliesthatthenumberofrowsexceedsthe
numberofcolumns(m>n).Wehavemoreequationsthanunknowns,andwe
needtoensurethattheequationsareconsistent(linearlydependent)fora
solutiontobepossible;i.e.,hastobesatisfied.
Case2: r = n < m
D
zero
= 0
X =D
pivot
R =
I
O
mn
=
I
r r
F
r nr
( )
O
mr
( )
r
O
mr
( )
nr
( )
(
(
(
r r
O
mr
( )
r
(
(
X
pivot
{ }
= D
pivot
{ }
F
X
free
{ }
D
zero
{ }
= O
{ }
must be satisfied by C for a feasible solution set.
A|C
=
2 4 6 c
1
2 1 3 c
2
3 1 4 c
3
4 2 8 c
4
(
(
(
(
(
1 2 3 c
1
/ 2
0 3 3 c
1
c
2
0 5 5 3c
1
/ 2 c
3
0 6 4 2c
1
c
4
(
(
(
(
(
shouldbezero shouldbezero
1 0 1 c
1
+ 4c
2
( )
6
0 1 1 c
1
c
2
( )
3
0 0 2 2c
2
+ c
4
( )
0 0 0 c
3
+ c
1
10c
2
( )
6
(
(
(
(
(
(
(rows 3 and 4 interchanged)
2 4 6
2 1 3
3 1 4
4 2 8
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
a
b
(10ba) / 6
c
(
(
(
(
(
AX=C
1 0 0
0 1 0
0 0 1
0 0 0
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
2
1
0
0
(
(
(
(
(
X
{ }
= D
{ }
X
1
X
2
X
3
(
(
(
p
=
2
1
0
(
(
(
For a feasible solution space
c
3
= c
1
10c
2
( )
6
RX =D
I
O
(
X
O
(
=
D
O
2 4 6
2 1 3
3 1 4
4 2 8
(
(
(
(
(
X
1
X
2
X
3
(
(
(
=
0
3
5
6
(
(
(
(
(
Case3: r = m< n
R= I F
mn
=
I
r r
F
r nr
( )
O
mr
( )
r
O
mr
( )
nr
( )
(
(
(
I
r r
F
r nr
( )
(
CoefficientmatrixAhasafullrowrank(r=m),buttherearelinearly
dependentcolumns(r<n),whichalsoimpliesthatthenumberofcolumns
exceedsthenumberofrows(n>m).Wehavemoreunknownsthanequations,
whichisasituationweencounterinstaticallyindeterminatestructures.
Owingtotheabsenceofzerorowvectors,thereisnoconstraintonthe
constantvectorC,andasolutioniscertainlypossible.
However,astherearefreevariablespresent,thenullspacesolutionhas
infinitepossibilities (asinCase1)inthecompletesolution.
X
pivot
{ }
= D
pivot
{ }
F
X
free
{ }
X = X
p
+ X
n
D
zero
{ }
= O
{ }
1 2 3 4
2 1 1 2
3 3 4 8
(
(
(
X
1
X
2
X
3
X
4
(
(
(
(
(
=
c
1
c
2
c
3
(
(
(
1 2 3 4
0 3 5 6
0 3 5 4
(
(
(
X
1
X
2
X
3
X
4
(
(
(
(
(
=
c
1
2c
1
c
2
3c
1
c
3
(
(
(
1 0 1 3 0
0 1 5 3 2
0 0 0 2
(
(
(
X
1
X
2
X
3
X
4
(
(
(
(
(
=
c
1
+2c
2
( )
3
2c
1
c
2
( )
3
c
1
c
2
+c
3
( )
(
(
(
(
AX=C
1 0 0 1 3
0 1 0 5 3
0 0 1 0
(
(
(
X
1
X
2
X
4
X
3
=
c
1
+2c
2
( )
3
5c
1
+2c
2
( )
3c
3
c
1
c
2
+c
3
( )
2
(
(
(
(
=
d
1
d
2
d
3
Interchange third and fourth columns to preserve the identity matrix on the left.
I F
%
X
{ }
= D
{ }
RX=D
c
1
c
2
c
3
(
(
(
=
3
6
2
(
(
(
d
1
d
2
d
3
(
(
(
=
3
7
3.5
(
(
(
It is evident that infinite solutions are possible.
1 2 3 4
2 1 1 2
3 3 4 8
(
(
(
X
1
X
2
X
3
X
4
(
(
(
(
(
=
c
1
c
2
c
3
(
(
(
AX=C
X
pivot
{ }
= D
pivot
{ }
F
X
free
{ }
Particularsolution:LetX
3
=0
X
1
X
2
X
3
X
4
p
=
3
7
0
3.5
Nullspacesolution(D =O);LetX
3
=
X
1
X
2
X
3
(
(
(
=
1/ 3
5/ 3
0
(
(
(
I F
%
X
{ }
= D
{ }
1 0 0 1 3
0 1 0 5 3
0 0 1 0
(
(
(
X
1
X
2
X
4
X
3
=
3
7
7 2
X = X
p
+ X
n
=
3
7
0
3.5
(
(
(
(
+
3
5 3
(
(
(
(
=
3+ 3
75 3
3.5
(
(
(
(
Completesolution:
Inthiscase,thecoefficientmatrixAhasafullcolumnrank(r=n)aswellasafull
rowrank (r=m),whichalsoimpliesthatthematrixisasquarematrix(m=n).
Suchamatrixissaidtobeinvertibleor nonsingular.
Case4: r = m= n
R =I
unique solution, X =D
EliminationTechniqueforsolvingAX=C
InthetraditionalGausseliminationprocedure,itissufficienttoreducethe
coefficientmatrixAtoanuppertriangularformUforthispurpose,while
carryingouttheelementarymatrixoperationsontheaugmentedmatrix
IfAis asquarematrixandtheequationsareconsistent,theuniquesolutioncan
beobtainedbybacksubstitution.
However,bygoingafewstepsfurther,theAmatrixcanbereducedtotherow
reducedechelonformR,andthecompletesolution,ifany,canbedirectly
obtained.
A | C
R =
I F
O O
(
WhenthematrixAissquareandoffullrank,analternativeapproachto
solvingtheequationsisbytheoperationofinversion.
WhenthematrixAissquareandoffullrank,analternativeapproachto
solvingtheequationsisbytheoperationofinversion.
AX = C
MATRIXINVERSION
X
{ }
= A
1
C
{ }
AA
1
= A
1
A =I
BasicPropertiesofInverseofaMatrix:
A
1
( )
1
= A
A
T
( )
1
= A
1
( )
T
A
( )
1
=
1
A
1
AB
( )
1
=B
1
A
1
DeterminantofaMatrix
Ingeneral,itcanbeprovedthattheinverseexists,ifascalarproperlycalled
thedeterminantofthesquarematrixA,denotedasordetA or,is
notequaltozero.
Foradiagonalmatrix,thedeterminantisgivenbytheproductofallthe
diagonalelements.
A A
( )
Perhapsthesimplestwayoffindingthedeterminantofanysquarematrixisby
applyingtheprocessofelimination,reducingthematrixtoanupperdiagonal
formU.Thedeterminantisdirectlyobtainableastheproductofthepivot
elementsinthediagonal (ve signifoddno.ofrowexchangesinvolved).
For a matrix A of order 2 2, the determinant is given by:
det A =
a
11
a
12
a
21
a
22
= a
11
a
22
a
21
a
12
( )
Clearly,anonzerodeterminantispossibleforagivensquarematrixAonlyif
therearepivotelementsinalltherowsofthematrix;i.e.,thematrixhasto
haveafullrankforittobenonsingularandinvertible.
det A = a
ij
o
ij
j =1
n
fori = 1,2,3,...,n
where isthecofactoroftheelementa
ij
,whosevalueisgivenbythe
determinant
ij
ofthe(n1)(n1)matrixobtainedbydeletingthei
th
rowandj
th
columnofthematrixA,withtheappropriatesign(positiveornegative).
o
ij
= 1
( )
i + j
|
ij
det A = a
11
o
11
+a
12
o
12
+...a
1n
o
1n
det A = a
11
a
22
a
23
a
32
a
33
a
12
a
21
a
23
a
31
a
33
+a
13
a
21
a
22
a
31
a
32
det A
m
= det A
( )
m
det A =
n
det A
( )
for A
nn
det A
T
= detA
AdjointMethodofFindingInverse
Ingeneral,foranysquarematrixofordern n,provided,detA 0,itcanbeshownthat
whereiscalledtheadjointmatrixofA,whichisthetransposeofamatrixwhose
elementscomprisethecofactors
ij
ofA.Thistechnique,however,becomescumbersome
whentheorderofthematrixexceedsthreeorfour.
A
1
=
1
detA
%
A
a b
c d
(
1
=
1
ad bc
d b
c a
a sym
b c
d e f
(
(
(
1
=
1
a cf e
2
( )
b bf de
( )
+d be cd
( )
cf e
2
( )
sym
de bf
( )
fa d
2
( )
be cd
( )
bd ae
( )
ac b
2
( )
(
(
(
(
7 3 2 3 3 8
2 3 7 3 3 8
3 8 3 8 3 8
(
(
(
1
=
1
1.40625
0.734375 0.109375 0.625
0.109375 0.734375 0.625
0.625 0.625 5.0
(
(
(
Algorithmbasediterativemethodsoffindinginverse:
GaussJordanEliminationMethod
LDLTDecompositionMethod
Cholesky DecompositionMethod
Cramersruleissuitableforsolvingasmallnumberofsimultaneousequations.
Itrequiresgenerationofn+1determinants,whichiscumbersomebyalgebraic
formulation.Eliminationbasedalgorithmicmethodsaremuchbettersuitedfor
computerapplication.
CramersRule
Thesolutiontoasetofconsistentequations,
canbeshowntobegivenby:
A
X
{ }
= C
{ }
X
i
=
a
11
a
1, j 1
c
1
a
1n
M
a
n1
a
n, j 1
c
n
a
nn
a
11
a
1, j 1
a
1 j
a
1n
M
a
n1
a
n, j 1
a
nj
a
nn
i = 1,2,.....,n
( )
Thestabilityofiterativesolutionprocedures(formatrixinversion)andthe
accuracyoftheendresultdependontheconditionofthematrix,whichisa
measureofitsnonsingularity.Amatrixissaidtobeillconditioned(ornear
singular)ifitsdeterminantisverysmallincomparisonwiththevalueofits
averageelement,andsuchamatrixisvulnerabletoerroneousestimationofits
inversebyiterativesolvers.
Stiffnessmatricesarerelativelywellconditionedandhavethepropertyofpositive
definiteness,withdiagonaldominance,wherebytheirinversescanbestablyand
accuratelygenerated.
ConditionofaMatrix
ThesquarematrixAofordern issaidtobe positivedefinite,ifforanyarbitrarychoiceof
anndimensionalvectorX,theproductX
T
AXyieldsascalarquantitythatisinvariably
positive.Alltheeigenvalues ofsuchamatrixarereal,distinctandspacedwellapart.
ThesquarematrixAofordern issaidtobe positivedefinite,ifforanyarbitrarychoiceof
anndimensionalvectorX,theproductX
T
AXyieldsascalarquantitythatisinvariably
positive.Alltheeigenvalues ofsuchamatrixarereal,distinctandspacedwellapart.
X
T
AX>0