Upload
sarankumar-thathuru
View
228
Download
0
Embed Size (px)
Citation preview
8/12/2019 17- Matrix Basics 1
1/31
Advanced
Structural
Analysis
DevdasMenonProfessor
IITMadras
NationalProgrammeonTechnologyEnhancedLearning(NPTEL)
www.nptel.ac.in
Lecture17Module3:
BasicMatrixConcepts
8/12/2019 17- Matrix Basics 1
2/31
AdvancedStructuralAnalysis
Modules
1. Reviewofbasicstructuralanalysis 1(6lectures)
2. Reviewofbasicstructuralanalysis 2(10lectures)
3. Basicmatrix
concepts
4. Matrixanalysisofstructureswithaxialelements
5. Matrixanalysisofbeamsandgrids
6. Matrixanalysisofplaneandspaceframes
7. Analysisofelasticinstabilityandsecondordereffects
8/12/2019 17- Matrix Basics 1
3/31
Module3:BasicMatrixConcepts
Reviewofmatrixalgebra
Introductiontomatrixstructuralanalysis(forceand
displacementtransformations;stiffnessandflexibilitymatrices;basicformulations;equivalent
jointloads).
8/12/2019 17- Matrix Basics 1
4/31
ReviewofBasic
ConceptsinStructuralAnalysis
Matrix
Conceptsand
Methods
Structures
withAxialElements
Beamsand Grids
PlaneandSpace
Frames
ElasticInstability
andSecondorder
AnalysisA
dvanced
Stru
ctural
Analys
is
8/12/2019 17- Matrix Basics 1
5/31
1INTRODUCTION
2MATRIX
3VECTOR
4ELEMENTARYMATRIXOPERATIONS
5MATRIXMUTLITPLICATION
6TRANSPOSEOFAMATRIX
MatrixConceptsandAlgebra
7RANKOFAMATRIX
8LINEARSIMULTANEOUSEQUATIONS
9MATRIXINVERSION
10EIGENVALUESANDEIGENVECTORS
8/12/2019 17- Matrix Basics 1
6/31
MatricesinStructuralAnalysis
LOADS
(input) STRUCTURE
(system)
RESPONSE?
(output)
A AA AR A
R RA RR R
F k k D
F k k D
LoadVectorFASupportDisplacementsDR
InitialDeformationsD*in
DisplacementVectorDASupportReactionsDRInternalForcesF*
MemberDeformationsD*
Analysissoftwarepackages
8/12/2019 17- Matrix Basics 1
7/31
Introduction
Thereisanincreasingtendencyamongmodernstructuralengineerstoleanheavilyonsoftwarepackagesforeverything.
Thisinducesafalsesenseofknowledge,securityandpower.
Thecomputerisindeedapowerfultoolandanassetforanystructuralengineer. Itisdangerous,however,tomakethetoolonesmaster,andtomakeitaconvenientsubstitutefor
humanknowledge,experienceandcreativethinking.
Ref.: Preface to Advanced Structural Analysis
8/12/2019 17- Matrix Basics 1
8/31
By definition,amatrix isrectangulararrayofelementsarrangedinhorizontalrowsandverticalcolumns.
Theentriesofamatrix,calledelements,arescalarquantities commonlynumbers,butmayalsobefunctions,operatorsorevenmatrices(calledsub
matrices)themselves.
A A
A
mn a
ij
aij
mn
a11
a12
a13
L a1 n
a21
a22
a23
L a2n
a31
a32
a33
L a3n
M M M M
am1
am2
am3
L amm
order
m n square matrix
a
ij 0 i,j null matrix, O
identity matrix,I
aij 0: i j
aij 1 : i j
diagonal elements
1 0 0
0 1 00 0 1
Matrix?
8/12/2019 17- Matrix Basics 1
9/31
square matrix
symmetric matrix (m= n= 6)
aij ajibanded matrix
sparse matrix
a 0 0d b 0e f c
a d e0 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
74-1
-2
-2
-1
1
1
00
0
0
000
00
0 0
0 00
0 00
00
0
A a
ij
mn
b
ij
ip
cij
l np
dij
ml pe
ij
ml np
mn
B CD E
Partitioning:
8/12/2019 17- Matrix Basics 1
10/31
Avectorisasimplearrayofscalarquantities,typicallyarrangedinaverticalcolumn.
Hence,thevectorcanbevisualisedasamatrixoforderm
1,wherethenumberm iscalledthedimensionofthevector.Thescalarentriesofavectorarecalledcomponents ofthevector.
V V m
vi V m1 vij m1
v1
v2
v3
M
vm
Whatisavector?
Isitatype
of
matrix?
a1 a2 a3 L an row vector
8/12/2019 17- Matrix Basics 1
11/31
Wecanvisualizeamultidimensionallinearvectorspace,m,whosedimensionm isgivenbytheminimumnumberoflinearlyindependentvectors(withrealcomponents)requiredtospanthespace.Vectorsaresaidtospanavectorspace,ifthespaceconsistsofallpossiblelinearcombinations ofthosevectors.Anysetofvectorsthatarelinearly
independentandalsospanthevectorspaceiscalledabasisofthevectorspace.
0x
y
z
V
V
2
13
canbevisualisedin3 vectorspaceas V 2i j 3k
unitvectors
10
0
'01
0
and00
1
provideanorthogonalbasisin3 vectorspace.
havingamagnitudeorlength,
V v
i
2
i1
m
14
Asetofvectors,{V1,V2,,Vn},havingthesamedimensionm,issaidtobelinearlyindependentifnolinearcombinationofthem(otherthanthezero
combination)resultsinazerovector;i.e.,c1 = c2 = = cn = 0,if .
ciVii1
n
0
8/12/2019 17- Matrix Basics 1
12/31
ELEMENTARYMATRIXOPERATIONS
ScalarMultiplication: A aij mn aij mn A
MatrixAddition: A+B aij mn bij mn aij bij mn
A B a
ij
mn
1 bij mn aij bij mn
A+B B+A
A+ B+C A+B +C
MatrixMultiplication: aij mn bij np cij mpAB =C;i.e.,
L L L L L
L L L L L
ai1 ai2 ai3 L ain
L L L L L
L L L L L
mn
M M b1j
M M
M M b2j
M M
M M M M M
M M bnj
M M
np
M
M
L L cij L
M
M
mp
ith row
jth column jth column
ith row
8/12/2019 17- Matrix Basics 1
13/31
A B+C
AB+AC
A BC AB C
B+C A BA+CAAO= O
AI= A
2 13 4
1 2
32
1 42 0
22
0 811 125 4
32
A mn B np C mp
23
1
1 14
2
2
EverycolumnvectorofC isalinear
combinationofthecolumnvectorsofthepremultiplyingmatrixA!
3 1 4
4
2 0
EveryrowvectorofC isalinearcombinationoftherowvectorsofthepostmultiplyingmatrixB!
Matrix multiplication operation does notpossess the property of commutativity;i.e, in general, AB BA
8/12/2019 17- Matrix Basics 1
14/31
TRANSPOSEOFAMATRIX
Transposition is an operation in which the rectangular array of the matrix is rearranged(transposed) such that the order of the matrix changes from m nto n m, with therows changed into columns, preserving the order. If the original matrix is A =[aij]mn,then thetranspose ofA, which is denoted as AT, is given by AT = [aij]
T = [aji]nm.
Transposition is an operation in which the rectangular array of the matrix is rearranged(transposed) such that the order of the matrix changes from m nto n m, with therows changed into columns, preserving the order. If the original matrix is A =[aij]mn,then thetranspose ofA, which is denoted as AT, is given by AT = [aij]
T = [aji]nm.
(AT)T =A
(A)T =AT
(A+B)T =AT+BT
(AB)T =BTAT
ST =AT (AT)T
A nm
T
A mn S nn
Square matrix
=AT A=S
1 2 3
2 0 1
1 2
2 03 1
14 1
1 5
Symmetric matrix
(sji= sij)
AT=AT (i.e., aji= aji ) Skew - Symmetric
8/12/2019 17- Matrix Basics 1
15/31
Theproduct{F}T{D}= resultsinamatrixoforder1 1 ,whichisnothing
butascalar. Suchaproduct,whichissometimesdenotedas ,iscalledinnerproduct (dotproductinvectoralgebrainvolvingtwoandthreedimensionalvectors). Thisproductessentiallyreflectstheprojectedlengthofonevectoralongthedirectionoftheothervector. ThemagnitudeofanyvectorV,defined as maybeviewedasthe
innerproductofVwithitself,i.e., .
Theproduct{F}T{D}= resultsinamatrixoforder1 1 ,whichisnothing
butascalar. Suchaproduct,whichissometimesdenotedas ,iscalledinnerproduct (dotproductinvectoralgebrainvolvingtwoandthreedimensionalvectors). Thisproductessentiallyreflectstheprojectedlengthofonevectoralongthedirectionoftheothervector. ThemagnitudeofanyvectorV,defined as maybeviewedasthe
innerproductofVwithitself,i.e., .
F,D
V vi
2
i1m
V,V VTVorthonormal
vectors
X
i
T
X
j ij 1 if i= j
0 if i j
D,F DT k D
D,F T DT k D T DT k T D [k] is symmetric !
Commutative propertyof inner product: F
T
D D T
F
F
m1 k mm D m1
8/12/2019 17- Matrix Basics 1
16/31
Therankofthematrix[A]isequaltothenumberoflinearly
independentcolumnvectorsofthematrix,andthisnumberisidenticaltothenumberoflinearlyindependentrowvectors.
Themaximumvalueoftherankrofanymatrixoforderm n isgivenbym orn(whicheverislower),andtheminimumvalueis1.
Linearsimultaneousequations:
aij
Xj
j1n ci i 1,2,.....,m
A mn Xn1 Cm1coefficientmatrix
RANKOF
AMATRIX
Thesubspaceinthevectorspacem containingalllinearcombinationsoftheindependentcolumn (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(homogeneousequations)
nullspaceofA(allpossible
solutionsofX)
8/12/2019 17- Matrix Basics 1
17/31
RowReducedEchelonForm
R mn
Irr F r nr
O mr rO mr nr
ArelativelyeasyandcertainwayofdeterminingtherankofamatrixisbyreducingthematrixtoarowreducedechelonformR throughaprocessofelimination(transformingA ascloselyaspossibletoanidentifymatrixI intheupperleftcorner).
Free variablecoefficient 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 FO O
Rank r= 2
2 4 6
2 1 3
3 1 4
4 2 6
X1
X2
X3
c1
c2
c3
c4
AX=C:
8/12/2019 17- Matrix Basics 1
18/31
LINEARSIMULTANEOUSEQUATIONS
AX=C
I rr F r nrO mrr O mr nr
mn
Xpivot
r1X
free nr1
n1
Dpivot
r1D
zero mr1
m1
pivotvariables pivotrowconstants
freevariables zerorowconstants
Xpivot Dpivot F Xfree X Xp XnD
zero O
Case1 : r m and r n
must be satisfied by C for a feasible solution set.
2 4 6
2 1 3
3 1 4
4 2 6
X1
X2
X3
c1
c2
c3
c4
AX=C
RX=D
1 0 1
0 1 1
0 0 0
0 0 0
X1
X2
X3
c14c2 6c1 c
2 3c
3 c
110c
2 62c2
c4
RX=D
zero
zero
c3
10c2c
1
6
c4 2c2
particular + null space solution
8/12/2019 17- Matrix Basics 1
19/31
2 4 6
2 1 3
3 1 4
4 2 6
X1
X2
X3
a
b
(10ba) / 62b
AX=C
2 4 6
2 1 3
3 1 4
4 2 6
X1
X2
X3
0
3
5
6
1 2 3
0 3 3
0 5 5
0 6 6
X1
X2
X3
0356
1 0 1
0 1 1
0 0 0
0 0 0
X1
X2
X3
210
0
For a feasible solution space
X
pivot Dpivot F XfreeD
zero OParticularsolution:LetX3=3
X1 2 X
3
X2 1 X
3
X1
X2
X3
p
210
Nullspacesolution(C=O);LetX3=
X1
X2
X3
n
Complete
solution:
X=Xp+
Xn
X1
X2
X3
n
21
8/12/2019 17- Matrix Basics 1
20/31
CoefficientmatrixAhasafullcolumnrank(r=n),buttherearelinearlydependentrows (rn).Wehavemoreequationsthanunknowns,andweneedtoensurethattheequationsareconsistent(linearlydependent)forasolutiontobepossible;i.e., hastobesatisfied.
Case2: r n m
D
zero 0
XD
pivot
R IO
R mn Irr F
r nr
O mr rO mr nr
IrrO mr r
X
pivot Dpivot F XfreeD
zero O must be satisfied by C for a feasible solution set.
A | C 2 4 6 c12 1 3 c
2
3 1 4 c3
4 2 8 c4
1 2 3 c1 / 20 3 3 c
1 c
2
0 5 5 3c1 / 2 c
3
0 6 4 2c1 c
4
shouldbezeroshouldbezero
1 0 1 c
1 4c
2 60 1 1 c1 c2 30 0 2 2c
2 c
4 0 0 0 c
3 c
1 10c
2 6
(rows 3 and 4 interchanged)
8/12/2019 17- Matrix Basics 1
21/31
2 4 6
2 1 3
3 1 4
4 2 8
X1
X2
X3
a
b
(10ba) / 6c
AX=C
1 0 0
0 1 0
0 0 1
0 0 0
X1
X2X
3
210
0
X D
X1
X2
X3
p
210
For a feasible solution spacec
3 c
110c
2 6
RXDI
O
X
O
D
O
2 4 6
2 1 3
3 1 4
4 2 8
X1
X2X
3
0
3
5
6
8/12/2019 17- Matrix Basics 1
22/31
Case3: r m n
R I F
R mn Irr F r nr
O mr rO mr nr
Irr F r nr
CoefficientmatrixAhasafullrowrank(r=m),buttherearelinearlydependentcolumns(rm).Wehavemoreunknownsthanequations,whichisasituationweencounterinstaticallyindeterminatestructures.
Owingtotheabsenceofzerorowvectors,thereisnoconstraintontheconstantvectorC,andasolutioniscertainlypossible.
However,astherearefreevariablespresent,thenullspacesolutionhasinfinitepossibilities(asinCase1)inthecompletesolution.
X
pivot Dpivot F Xfree X Xp XnD
zero O
8/12/2019 17- Matrix Basics 1
23/31
1 2 3 4
2 1 1 2
3 3 4 8
X1
X2
X3
X4
c1
c2
c3
1 2 3 40 3 5 60 3 5 4
X1
X2
X3
X4
c12c
1 c
2
3c1 c
3
1 0 1 3 00 1 5 3 20 0 0 2
X1
X2
X3
X4
c12c
2 32c
1 c
2 3c1 c2 c3
AX=C
1 0 0 1 30 1 0 5 30 0 1 0
X1
X2
X4
X3
c12c2 35c12c
2 3 c3c1 c
2 c
3 2
d1d
2
d3
Interchange third and fourth columns to preserve the identity matrix on the left. I F %X D
RX=D
c1
c2
c3
3
62
d1
d2
d3
37
3.5
8/12/2019 17- Matrix Basics 1
24/31
It is evident that infinite solutions are possible.
1 2 3 42 1 1 2
3 3 4 8
X1
X2
X3
X4
c1
c2
c3
AX=C
Xpivot Dpivot F XfreeParticularsolution:LetX3=0
X1
X2
X3
X4
p
3703.5
Nullspacesolution(D=O);LetX3=
X1X
2
X3
1 / 35 / 30
I F %X D
1 0 0 1 30 1 0 5 3
0 0 1 0
X1
X2
X4
X3
377 2
X XpX
n 37
03.5
35 3
0
3 375 3
3.5
Completesolution:
8/12/2019 17- Matrix Basics 1
25/31
Inthiscase,thecoefficientmatrixAhasafullcolumnrank(r=n)aswellasafullrowrank (r=m),whichalsoimpliesthatthematrixisasquarematrix(m=n).Suchamatrixissaidtobeinvertibleor nonsingular.
Case4 : r m n
R I unique solution, XDEliminationTechniqueforsolvingAX=C
InthetraditionalGausseliminationprocedure,itissufficienttoreducethecoefficientmatrixAtoanuppertriangularformUforthispurpose,whilecarryingouttheelementarymatrixoperationsontheaugmentedmatrixIfAis asquarematrixandtheequationsareconsistent,theuniquesolutioncanbeobtainedbybacksubstitution.
However,bygoingafewstepsfurther,theAmatrixcanbereducedtotherowreducedechelonformR,andthecompletesolution,ifany,canbedirectlyobtained.
A | C
R I FO O
8/12/2019 17- Matrix Basics 1
26/31
WhenthematrixAissquareandoffullrank,analternativeapproachtosolvingtheequationsisbytheoperationofinversion.WhenthematrixAissquareandoffullrank,analternativeapproachtosolvingtheequationsisbytheoperationofinversion.
AX = CMATRIXINVERSION
X A 1 C
AA1 A1A I
BasicPropertiesofInverseofaMatrix:
A1 1 AA T 1 A 1 TA1 1 A 1AB1 B1A1
DeterminantofaMatrix
Ingeneral,itcanbeprovedthattheinverseexists,ifascalarproperlycalled
thedeterminantofthesquarematrixA,denotedas ordetA or ,isnotequaltozero.
Foradiagonalmatrix,thedeterminantisgivenbytheproductofallthe
diagonalelements.
A A
8/12/2019 17- Matrix Basics 1
27/31
Perhapsthesimplestwayoffindingthedeterminantofanysquarematrixisby
applyingtheprocessofelimination,reducingthematrixtoanupperdiagonalform U.Thedeterminantisdirectlyobtainableastheproductofthepivotelementsinthediagonal (ve signifoddno.ofrowexchangesinvolved).
For a matrix A of order 2 2, the determinant is given by:
det A a11 a12a
21 a
22
a11
a22 a
21a
12
Clearly,anonzerodeterminantispossibleforagivensquarematrixAonlyiftherearepivotelementsinalltherowsofthematrix;i.e.,thematrixhastohaveafullrankforittobenonsingularandinvertible.
det A
aij
ijj1
n
for i 1,2,3,...,nwhere isthecofactoroftheelement aij,whosevalueisgivenbythedeterminant ijofthe(n1) (n1)matrixobtainedbydeletingthei
th rowandjth
columnofthematrixA,withtheappropriatesign(positiveornegative).
ij 1ij ij
8/12/2019 17- Matrix Basics 1
28/31
det A a
11
11 a
12
12 ...a
1 n
1 n
det A a11
a22
a23
a32
a33
a12
a21
a23
a31
a33
a13
a21
a22
a31
a32
det A m det AmdetA n det A for A nndet A
T detAAdjointMethodofFindingInverse
Ingeneral,foranysquarematrixofordern n,provided,det A 0,itcanbeshownthat
where iscalledtheadjointmatrixofA,whichisthetransposeofamatrixwhoseelementscomprisethecofactorsijofA.Thistechnique,however,becomescumbersomewhentheorderofthematrixexceedsthreeorfour.
A1 1detA
A
8/12/2019 17- Matrix Basics 1
29/31
a b
c d
1 1
adbc d bc a
a sym
b c
d e f
1
1a cf e2 b bf ded be cd
cf e2 symdebf fad2be cd bdae acb2
7 3 2 3 3 82 3 7 3 3 83 8 3 8 3 8
1
11.40625
0.734375 0.109375 0.6250.109375 0.734375 0.6250.625 0.625 5.0
Algorithmbasediterativemethodsoffindinginverse:
GaussJordanEliminationMethod LDLTDecompositionMethod Cholesky DecompositionMethod
8/12/2019 17- Matrix Basics 1
30/31
Cramersruleissuitableforsolvingasmallnumberofsimultaneousequations.
Itrequiresgenerationofn+1determinants,whichiscumbersomebyalgebraicformulation.Eliminationbasedalgorithmicmethodsaremuchbettersuitedforcomputerapplication.
CramersRule
Thesolutiontoasetofconsistentequations,canbeshowntobegivenby:
A X C
Xi
a11
a1 ,j1 c1 a1 n
Ma
n1 a
n,j1 cn anna
11 a
1 ,j1 a1j a1 nM
an1
an,j1 anj ann
i 1,2,.....,n
8/12/2019 17- Matrix Basics 1
31/31
Thestabilityofiterativesolutionprocedures(formatrixinversion)andtheaccuracyoftheendresultdependontheconditionofthematrix,whichisameasureofitsnonsingularity.Amatrixissaidtobeillconditioned(ornearsingular)ifitsdeterminantisverysmallincomparisonwiththevalueofits
averageelement,andsuchamatrixisvulnerabletoerroneousestimationofitsinversebyiterativesolvers.
Stiffnessmatricesarerelativelywellconditionedandhavethepropertyofpositivedefiniteness,withdiagonaldominance,wherebytheirinversescanbestablyandaccuratelygenerated.
ConditionofaMatrix
ThesquarematrixAofordernissaidtobepositivedefinite,ifforanyarbitrarychoiceof
anndimensionalvectorX,theproductXT
AXyieldsascalarquantitythatisinvariablypositive.Alltheeigenvaluesofsuchamatrixarereal,distinctandspacedwellapart.
ThesquarematrixAofordernissaidtobepositivedefinite,ifforanyarbitrarychoiceof
anndimensionalvectorX,theproductXT
AXyieldsascalarquantitythatisinvariablypositive.Alltheeigenvaluesofsuchamatrixarereal,distinctandspacedwellapart.
XTAX 0