MIT18

Embed Size (px)

Citation preview

  • 8/12/2019 MIT18

    1/6

    18.06ProblemSet2SolutionTotal: 100points

    Section2.5. Problem24: UseGauss-Jordaneliminationon [U I]tofindtheuppertriangular1

    U1: 1 0 0a b

    U U1 x1 x2 x3= 00 0 1 0 0 1

    (4points): Rowreduce[U I]toget[I U1]asfollows(hereRi standsfortheithrow):

    0 1 1 0=I c .

    Solution

    1 a b 1 0 0 (R1 =R1aR2)

    =R2cR2)1

    0 0 10 0 1 0 0

    0 1 0bac a 0 1 c 0 1 0 (R2 1 0 c

    0 0 1 0 0 1 1a acb

    1

    R1 1 0 0 1 =R1(bac)R3 0 1 0 0 1

    0 0 1 0 0 c .

    Section 2.5. Problem40: (Recommended)A isa4by4matrixwith 1sonthe diagonal anda,b,conthediagonalabove. FindA1 forthisbidiagonalmatrix.Solution (12points): Rowreduce[A I]toget[I A1]asfollows(hereRi standsfortheithrow):

    1 a 0 0 1 0 0 00 1 b 0 0 1 0 00 0 1 c 0 0 1 00 0 0 1 0 0 0 1

    (R1 =R1+aR2) 1 0 ab 0 1 a 0 0 (R2 =R2+bR2) 00 0 10 1 bc 0 1 b 0(R3 =R3+cR4) 00 0 0 1 0 0 0 10 0 1 c

    (R1 =R1+abR3) 1 0 0 0 1 a ab abc(R2 =R2+bcR4)

    0 1 0 0 0 1 b bc0 0 1 0 0 0 1 c0 0 0 1 0 0 0 1

    .

    Alternatively, write A = IN. Then N has a,b,c above the main diagonal, and all otherentriesequalto0. HenceA1 = (IN)1 =I+N+N2+N3 asN4 =0.Section2.6. Problem13: (Recommended)ComputeLandU forthesymmetricmatrix

    a a a aa b b b

    A= .a b c ca b c d

    Findfourconditionsona,b,c,dtogetA=LU withfourpivots.

  • 8/12/2019 MIT18

    2/6

    pset2-s10-soln: page2

    Solution (4 points): Eliminationsubtractsrow1 fromrows24, then row2 fromrows34, andfinally row 3 from row 4; the result is U. All the multipliers ij are equal to 1; so L is the lowertriangular

    matrix

    with

    1s

    on

    the

    diagonal

    and

    below

    it.

    a a a a a a a a 0 0 ba ba baba ba baA 0 0 00 0 bc da db

    a a a a 1 0 0 0baba cbcbca ca0 ca

    0 ba ba ba 1 1 0 0=U, L= .0 cb cb0 0 0 dc 1 1 1 1

    ThepivotsarethenonzeroentriesonthediagonalofU. Sotherearefourpivotswhenthese fourconditionsaresatisfied: a=0,b=a,c=b,andd=c.Section2.6. Problem18: IfA=LDU andalso A=L1D1U1 withall factors invertible, thenL=L1 andD=D1 andU =U1. Thethreefactorsareunique.

    Derive the equation L1

    1LD = D1U1U1. Are the two sides triangular or diagonal? DeduceL=L1 andU =U1 (theyallhavediagonal1s). ThenD=D1.Solution (4points): NoticethatLDU =L1D1U1. Multiplyonthe leftbyL1 1 andontheright

    byU1,gettingL

    1

    1LDUU1 =L1

    1L1D1U1U1.

    ButU U1 =I andL1

    1L1 =I. ThusL1 1LD=D1U1U1,asdesired.TheleftsideL1 1LD islowertriangular,andtherightsideD1U1U1 isuppertriangular. But

    theyreequal. Sotheyrebothdiagonal. HenceL1

    1LandU1U1 arediagonaltoo. Buttheyhavediagonal 1s. So theyre both equal to I. Thus L =L1 and U = U1. Also L1 1LD =D1U1U1.ThusD=D1.Section2.6. Problem25: Forthe6by6seconddifferenceconstant-diagonalmatrixK,putthepivots and multipliers intoK =LU. (L and U will have onlytwo nonzero diagonals, because Khas three.) Find a formula for the i, j entry of L1, by software like MATLAB using inv(L) or bylookingforanicepattern.

    0 1 1 1 0

    2 11

    =

    Solution (12 points): Here is the transcript of a session with the software Octave, which is theopen-source GNUclone of MATLAB.ThedecompositionK =LU is found using theteachingcodeslu.m,availablefrom

    http://web.mit.edu/18.06/www/Course-Info/Tcodes.html

    1,2,1matrix =toeplitz([2 -1 0 0 0 0]).K

    11 2

    http://web.mit.edu/18.06/www/Course-Info/Tcodes.htmlhttp://web.mit.edu/18.06/www/Course-Info/Tcodes.html
  • 8/12/2019 MIT18

    3/6

    pset2-s10-soln: page3

    octave:1> K=toeplitz([2 -1 0 0 0 0]);octave:2> [L,U]=slu(K);octave:3> inv(L)ans =

    1.00000 0.00000 0.00000 0.00000 0.00000 0.000000.50000 1.00000 0.00000 0.00000 0.00000 0.000000.33333 0.66667 1.00000 0.00000 0.00000 0.000000.25000 0.50000 0.75000 1.00000 0.00000 0.000000.20000 0.40000 0.60000 0.80000 1.00000 0.000000.16667 0.33333 0.50000 0.66667 0.83333 1.00000

    Sothenicepatternis(L1)ij =j/iforjiand(L1)ij =0forj > i.Section2.6. Problem26: IfyouprintK1,itdoesntlookgood. Butifyouprint7K1 (whenK is6by6),thatmatrixlookswonderful. Writedown7K1 byhand,followingthispattern:

    1 Row1andcolumn1are(6,5,4,3,2,1).2 Onandabovethemaindiagonal,rowiisitimesrow1.3 Onandbelowthemaindiagonal,columnj isj timescolumn1.

    1Multiply times that 7 to produce 7 Here is that pattern for = 3:K K I n.2 1 0

    23 2 1 43by3case =(K)(4K1) = 1

    0 12 2 4 2 4Thedeterminant .1 1 2 3 4ofthis K is4Solution (12points): Forn=6,followingthepatternyieldsthismatrix:

    6 5 4 3 2 15 10 8 6 4 24 8 12 9 6 33 6 9 12 8 42 4 6 8 10 51 2 3 4 5 6

    .

    HereisthetranscriptofanOctavesessionthatmultipliesK timesthat7K1.octave:1> K=toeplitz([2 -1 0 0 0 0]);octave:2> M=[6 5 4 3 2 1;5 10 8 6 4 2;4 8 12 9 6 3;3 6 9 12 8 4;2 4 6 8 10 5;1 2 3 4 5 6];octave:3> K*Mans =

    7 0 0 0 0 00 7 0 0 0 00 0 7 0 0 00 0 0 7 0 00 0 0 0 7 00 0 0 0 0 7

    Section2.7. Problem13: (a)Finda3by3permutationmatrixwithP3 =I (butnotP =I).(b)Finda4by4permutationP withP4 =I.

    Solution (4points): (a)LetP movetherowsinacycle: thefirsttothesecond,thesecondtothethird,andthethirdtothefirst. So

    0 0 1 0 1 0

    1 0 0, P2 =0 0 1, and P3 =I.P =0 1 0 1 0 0

  • 8/12/2019 MIT18

    4/6

    pset2-s10-soln: page4

    1 0(b) Let P be the block diagonal matrix with 1 and P on the diagonal: P = 0P

    . Since

    P3 =I,alsoP

    3 =I. SoP

    4 =P

    = I.Section 2.7. Problem 36: Agroup of matrices includes AB and A1 if it includes A and B.Productsandinversesstay inthegroup. Whichofthesesetsaregroups?LowertriangularmatricesLwith1sonthediagonal,symmetricmatricesS,positivematricesM,diagonal invertible matrices D, permutation matrices P, matrices with QT = Q1. Invent twomorematrixgroups.Solution (4points): Yes,the lowertriangularmatricesLwith1sonthediagonal formagroup.

    Clearly, the productof two is athird. Further, the Gauss-Jordan method shows thatthe inverseofone isanother.

    No,thesymmetricmatricesdonotformagroup. Forexample,herearetwosymmetricmatricesAandB whoseproductAB isnotsymmetric.

    0 1 0 1 2 3 2 4 5

    A=1 0 0, B=2 4 5, AB=1 2 3.0 0 1 3 5 6 3 5 6

    1 1No,thepositivematricesdonot formagroup. Forexample,0 1 ispositive,but its inverse

    11 isnot.0 1

    Yes,clearly,thediagonalinvertiblematricesformagroup.Yes,clearly,thepermutationmatricesmatricesformagroup.Yes, the matrices with QT =Q1 form a group. Indeed, if A and B are two such matrices,

    thensoareAB andA1,as(AB)T =BTAT =B1A1 = (AB)1 and (A1)T = (AT)1 =A1.

    Therearemanymorematrixgroups. Forexample,giventwo,theblockmatricesA0 B0

    form

    athirdasArangesoverthefirstgroupandB rangesoverthesecond. AnotherexampleisthesetofallproductscP wherec isanonzeroscalarandP isapermutationmatrixofgivensize.Section2.7. Problem40: SupposeQT equals Q1 (transposeequalinverse,soQTQ=I).

    (a)Showthatthecolumnsq1, . . . , q n areunitvectors: qi2=1.T(b)ShowthateverytwodistinctcolumnsofQareperpendicular: qi qj =0fori= j.

    (c)Finda2by2examplewithfirstentryq11 =cos.Solution (12points): Inanycase,theijentryofQTQisqiTqj. SoQTQ=I leadsto(a)qiTqi = 1

    foralliandto(b)qiTqj =0fori=j. Asfor(c),therotationmatrixcossinworks.sin cosSection3.1. Problem18: Trueorfalse(checkadditionorgiveacounterexample):

    (a)ThesymmetricmatricesinM(withAT =A)formasubspace.(b)Theskew-symmetricmatrices inM(withAT =A)formasubspace.(c)Theunsymmetricmatrices inM(withAT = A)formasubspace.

    Solution (4points): (a)True: AT =AandBT =B leadto(A+B)T =AT+BT =A+B.(b)True: AT =AandBT =B leadto(A+B)T =AT+BT =AB=(A+B).

    1 1 0 0 1 1(c)False: 0 0

    +

    1 1

    =1 1

    .

  • 8/12/2019 MIT18

    5/6

    pset2-s10-soln: page5

    Section3.1. Problem23: (Recommended)IfweaddanextracolumnbtoamatrixA,thenthecolumn space gets larger unless . Give an example where the column space gets larger andan

    example

    where

    it

    doesnt.

    Why

    is

    Ax

    =

    bsolvable

    exactly

    when

    the

    column

    space

    doesnt

    get

    largeritisthesameforAand [A b]?Solution (4 points): The column space gets larger unless itcontainsb; that is, b is a linear

    1 0combinationofthecolumnsof A. Forexample,letA=0 0;thenthecolumnspacegetslargerifb=

    10 anditdoesntifb=

    01 . TheequationAx=bissolvableexactlywhenbisa(nontrivial)

    linear combination of the columns of A (with the components of x as combining coefficients); soAx = b is solvable exactly when b lies in the column space, so exactly when the column spacedoesntgetlarger.Section3.1. Problem30: SupposeSandTaretwosubspacesofavectorspaceV.

    (a) Definition: The sumS+T contains all sumss+t of a vectors inS and a vectort inT. Show thatS+T satisfies the requirements (addition and scalar multiplication) for avectorspace.

    (b) IfSandTarelines inRm,whatisthedifferencebetweenS+TandST? Thatunioncontainsallvectors fromSandTorboth. Explainthisstatement: The spanof STisS+T. (Section3.5returnstothiswordspan.)

    Solution (12points): (a)Lets,s bevectors inS, lett,t bevectors inT,and letcbeascalar.Then

    (s+t) + (s+t) = (s+s) + (t+t) and c(s+t) =cs+ct.ThusS+T isclosedunderadditionandscalarmultiplication; inotherwords, itsatisfiesthetworequirementsforavectorspace.

    (b)IfSandTaredistinctlines,thenS+Tisaplane,whereasSTisnotevenclosedunderaddition. ThespanofSTisthesetofallcombinationsofvectorsinthisunion. Inparticular,itcontainsallsumss+tofavectorsinSandavectortinT,andthesesumsformS+T. Ontheotherhand,S+TcontainsbothSandT;so itcontainsST. Further,S+T isavectorspace.Soitcontainsallcombinationsofvectorsinitself;inparticular,itcontainsthespanofST. ThusthespanofSTisS+T.Section3.1. Problem32: ShowthatthematricesAand[AAB](withextracolumns)havethesamecolumnspace. ButfindasquarematrixwithC(A2)smallerthanC(A). Importantpoint:AnnbynmatrixhasC(A) =Rn exactlywhenA isan matrix.Solution (12 points): Each column of AB is a combination of the columns of A (the combining

    coefficientsaretheentries inthecorrespondingcolumnofB). Soanycombinationofthecolumnsof[AAB]isacombinationofthecolumnsofAalone. ThusAand[AAB]havethesamecolumnspace.

    0 1 1LetA=0 0

    . ThenA2 =0,soC(A2) =Z. ButC(A) isthelinethrough

    0

    .AnnbynmatrixhasC(A) =Rn exactlywhenA isan invertiblematrix,becauseAx=b is

    solvableforanygivenbexactlywhenA isinvertible.

  • 8/12/2019 MIT18

    6/6