THE THEORY OF MATRICES BY F. R. GANTMACHER VOLUME ONE CHELSEA PUBLISHING COMPANY NEW YORK, N.Y.Conraoite © 1050, ay Curzars Ponta ComPusy 1950, Crauara Puausunee Coxravr CCorvisauy 1969, ur Cususta Punusnise Comrase se Cone Corrie @ 1060, vr Cusaaza Poa Linasay of Coxonees Cxrata Cano None 58-11779 ‘rae exnsent wows, PEALE 15 10 YouRRE, fa BxoUst rmasshimios Be KA Hisoon, or THE Rose LANacAce 008 TRORIYA MATRITS wy FR, Gaxruacnee (Paxtnaxep) DEDALUS - Acorvo -IAG me mer ot mates COTTA 30200006220 SBN a0s4.0131-4 Proven gs rn Usinem Seas om Aw PREFACE, ‘tur warms carco.es is widele applied nowadays i yatlous branches of ‘nalhemaion,mechanirs, thorteal pysi theoretical retical euginre fag, ef. However, neither inthe Soviet north foreign Tieratare thee book that gives a sufficiently complete acount of the problems uf tr theory and of is diverse applications. The poesat base kan actempt te thie fap in the mathomatin! erat "The bok is based on letore courses on the theory of mattis and its ‘pplieations thst the sth ha given sreerl troy He couse af the et ‘seventeen years atthe Universities of Mon’ ae Tis anita the Nose Institut of Physi! Technolgy ‘The back is meant ot onlt for mhathsmatisians (uncengeadnates end ‘tear students) But ae for specilste in alid feds (physien eng neering) who are interest in maheratis an its apiliatins, "Therefore theanthor bas endeavoverd to make his acount ef the mater an ceca posible, assez only thatthe reader Is aequtinted withthe theory of determinants and with the asl court of higher uathomatien within the Brozramme of higher telnieal sduraion, aly a fw led seston i {he last chapters of the baok seguir ational automate Rnewieze on the part of the rues, Morava the author has tried to keep the indie vidal chapters af a pombe inopondont of eachother, Pat example Chapter Vs Panctions of Matra, dee ot dea athe material er tained in Chapters [and 11. At these places a Chapter ¥ where fonda ‘ental sneepts introduee? in Chapter IV are being wn! foe the fe! tes the corresponding refrences are given, Thnx 4 rrr who i depute swith the rudiments ofthe theory of ater van inmuediatly bein with ‘aig the chapters thst iteest hin, ‘The book ease of two pats containing fifteen ehaptes 1m Chapters Vand TI nformtion abun matties and Linear operators {8 developed initio and the connection hatnaun aperesee and maiees Ieintrodaeed, hapten IT rspounds the theoretical tass of Cans’ eimination rete 04 certs ausivinied effeetive methods nf moving arse of Tene ‘ustions fhe large mT this chapter the vue ao Breomes tequinte tb the teenie of opeeaing ith ratice hat ae dived int rest ‘sue "tees© p These five chapters vonstitute a first conese on matriers and their applic nection with the redurtion of matrices to a normal form. ‘This reduction smateiees "V4 and ind. Passacn : eran tah teeig cnt of oe mate with el mates Sinerame canes with waiter marca te Chptr it we expen ger hwy yee ares ot the form tah ve td Bae nin reomlr at he fhe dno Hans te ry of rls pol of ates 38 hnadn Webrstran thor of seeeatry dvr co hes fig Ierpenin inp Ronse thy of mal inte wh es iveeyafrser delopent ef Welt’ ary By mrt Brn, for thry-ih utr bos tht he a sued ying the ‘Spoon of ti tha-we cabo Chapey Neto Src ‘SEpenl toe} AB nthe a font cs, "he aban tite apd ite ny af nso ew differen tone income eet In Chater XIU we eaphin the somata pce prope fmt sit nih hon. omens and estes te impr pps auton of tela) nen Marto ea a he ear oe Drobiy and 3) tary proprio fete tien in mc ie The mate meth fing amogehe Marky shai eee intaclok [5] V1 Nonanordivend ted nthe ac the at OF triton probable 8imogentar Mra ein wh te nomen utes are wih mse man hes eecatt mat) Thee proper ne vrains ar cnn with anther import def ssn osteo tes oe tse a th pation neve sled MeO Ke jan ith ‘Stator oti ban haps NU nly era rss Seoain ar peste‘ rener on id ed me a te ee set emo Chapter XIV emphasis of mats tegitemsottiteesat tine ae tt Peel place H4.0) in thn chapter wong to he theory ofthe melpheasve Sgr (Pritts anit cohen we Tater et salen "Thne pons areas mil abana Se Saline ft eh fi ve ty ashe men these of inapis) i coneten hide pale Ay of motions we ah gine estan sane of SY. Bg Retons rete mate hry tna Witte eon Hoe ne vty anima Bifano, wh eal spac velo he nnn se ton cee pe {erm of healton ne ee oars ;“ Punrace 518 of Chapter XIV we ive a brlet survey of sume results of the fundamental inventigatins of TA, Lapno-Danilevkif on anastiefunetions ff several matrices anther apletions to diferent systems “Th ast ehapire, Chapter XV, dela vith te aplication of the theory st quadratic foron (in patienor, 9 Hankel forms) tothe Routh-Hurwitz Groblem of determining the mr of rome of polynomial jn the right Puitplae (Re > 0}. "The frst sstionsof te chapter contain the elasieat ‘Mature ofthe problem. bn § Sere give the tore of &. ML, Lyaruaos in (Chiat sablty eriterion la set up which isequvalent tothe Routh Thterioe, Together withthe ability eriteron of RouteTTurwita we iv agi ef thin chapter, the eomparnicly litle known eiterion of Lfuard tod Chipart in whic ihe user of detersinen inequities i oy sbout Tr of thet nthe Roath-Hurwiteerteion, atthe sad of Chepter XV we exhibit th ele connection between stabil. ity problems and two remeshable theorems of A. A. Markov end P. Th Gheyunow whieh were obtained by thew celebrated athorscn the basiof the Cpnusan of certain contnted fractions of special fps in sres of derea ing powers ofthe argument. Here we give matrix proot of these theorems Tin ten, i bet wommmary of the contents of this book FR, Gamtmachor PUBLISHERS’ PREFACE ‘Tis Ponswans wt 50 thank Profesor Gantmacher for bis kisdhes Communieating tothe teatsltor new versions of sveral paragraphs of the ‘rind Russnrlangusge book ‘The Dutiishers lc take pleasore i thanking the VB Deutscher Verne ‘ter Woaensehaten, whose wy pblehedtraslations of Resins seni Tonks into the eva langage elude a covterpart of the present work, for abet hind spirit of cooperation in aren tothe sof thei Formals inthe preparation of the preset work. ‘No material chungesteve ben tad in ve extn translating the present ‘wor fom the Rowson except for th replace of several parusrephs Dy The new versions supplind by Professor Gantmacher. Some changes inthe Feferences ulin tbe bibliography have been made for the bene of the Bolohdanguage render. CONTENTS Presser. Pomusteas’ Paarace 1. Marmecas Axo OPERATIONS ON MAHICES BL. Matries, ave notation oe § 2 AAdition and multiplication of ectangulr trea —— 5 8. Square meio $4 Compound matrices Minors of the Severe mates TL Tus Avconrruse oF Gauss ap Soa OF fs APPLICATIONS 51. Gauss aiinatin method § 2, Mechanea interpretation of Gaus slenithm 58. Splverter’ determinant int meen 1 A. The deeumpmsition oft aavare matric int tiangu fas § 5. The parition of a matrix into Docks. The technique of pernting with parttoaed matsigen, The generals leo of Get TL Lantan Ovenatons 1 aN DIMENSIONAL Vicon Space 81. Veet apnens § 2 ATinear operator mapoing an ninenions ape into wn ‘mimensiona spy ene 8. Aaattion and sultiplcation of ‘eannformation of cardia 5, Mquivalen mateees, he rank an operat " Basin ot an operator. Sven's $8 Linear operstons ann fn aeinensonal wpace inte hell47 Charasteritc values and characteristic wotors of a Tnear operator ——— 8 8, Linear operator of simple strctare | Tax Chuanscrinistte PoLYNOMTAL AND THE MINIMAL POLY: — 7 MOMESE OF A METRES. 4 1 Addition and multiplication of mate polynamfalen.— 78 § 2 Right and let division of matrix polpnomial——— TT $3 The generaliced Récnt hore $4 The characteristic polynomial of « mate. The aint fae 5. The method of Fader forthe simultaneoas computation ofthe nations ofthe characteristic polyromial and of the adjoine matrix —— - 8 § 6 The misinal polynomial ofa matrix. @ FUNCTIONS OF BIATRICES. cnn 4§1. Definition of «fonston of tein 8 42 The Lagrange-Sylvctr iaterplation polyoma 101 § Other forms of the definition of ft). The components tthe mateo Representation of functions of matriow by means of eres 110 Application of» faction ofa matric to the integration of ft pate of Hinear atiesentiah sqaienr with eorsant efficients 8 Stability of wotion inthe ease of liner epsom. Equivstenr Transronartons oF POLYNOMIAL MATHICES. [Awabrase Trrony oF ELSMENTany DIvisons.. $1. Blewentary (ranformations of « plynomial mats § 2. Canonical form of & 3 Invariant polynomials and elementary divisors of 4 poly ‘waa mais ‘Avriterion fo similarity nf mates ‘The normal forme of & mate "The clementary iviors ofthe matty ft} YL Mernsx Equations § 8, A general method of onstructing the teansforming matrix 189 $9 Avotber method of eostrocting @ transforming mation 164 VIL Tus Sracerone oF 4 Linear OPExaron 1 aN mDiMEn: MORAL SPACE wenn 15 $1. The minimal polynomial fm vestor and a apace (ith respect tp given Liner aperstor)—————————— {2 Decomposition into invariant subepasee with co-plae sina) polim onee ns 2. Congruenee. Paetr space |. Devomposition ofa space into epee invariant subspaces 18 5. The normal form of mtg ee 1 ‘ 1 5 Tnvarisat polynomial. Blementary divisors ————— 198, ‘The Jordan normal form of » wate son ‘Krylov’ metho of transforming the seular equation —— 202 as $1, The equation A208 ‘The savin omae A ‘The equation AY —¥1 2 sa $4. Tho sealer equation (0X) =O § 8. Matrix polynomial equations $6 Tho esteection of mls rots ofa nosingune marin 231 § 1. Theextrection of wth rts of a singular matrix. 204 8. The logaritnn of ttt 28 eral eoosierations Metriaston of m spuce——— res Sooo 43. Grams erterian Sor linear dependence of veto 4 85 ‘The geometrce! meaning of the Gramian and some equalities ~ 8 6 Orthogonalisation ofa sequsnee of stor $7 orthonormal hese § & Te adjoint operator§ 9 Normal operator ine unitary spt mann 258 10. The sprtre of normal, eemitin, aed ualasy operons. 270 E11, Postvesemidefinite sud postivedetinite hermitian ops erator — aan a 512, Polar desomponition af «linear aperatrin w unitary space Capley's termes 2 418. Linear operators ina euliden space 280 G16, Polar devompostion of an operator and the Cayley fr: mola inn eucidern spate $28, Commuting normal operators X. Qeapmarte Ano Himars1aw FoRns. 2 $1, Transformation af the aries ina quadeatie form 294 § 2 eduction of @ quadratic form ton sum of squares. The Tae of inet. aan 26 The methods of Lagrange and Jacobi of edacing quat- ithe are 9 80m 9 Om $4 Positive quareti forms — — 508 § 5. Redaction af quadratic fom to prinipal Banna 308 8 Penels of quadeatie £61 =n nna 10 7, Rstromal properties of the sbarorteristio wallow of « ren Tae peel of fet omen ———— a7 8 Smllonilatons of «apse with depres of Frotoma= $28 8, Hermitian forme —. 1 $20. Hankel forms ————— Bncioonseny 51 woe 89 CHAPTER I MATRICES AND OPERATIONS ON MATRICES $1. Mateos, Basie Notation 3, Lat re w given number fe? Deroscioe L: A retangsler aoray of musiers ofthe fod r fem dat ean ae a ‘called a matrix. When m=n, the matric clld square ad the number ‘agua! ton, 0 called its order. dn the general voos the mare esl rectangular (of dimension wn). ‘The arbors tat cont! he matrix (recalled ite eloment. Noramos: In the Goublesubscript notation for th elements, the fist ‘ubecrpt says denats she row art the sco eubeit the eum eo {ining the given lene. fea ‘As an alternative tothe notation (1) for » atric shal abo yee the sbresaton lime @=1,3, ” a ‘Often the matrix (1) wil also br denoted bya singe letter, for example A HEA iva square matrix of andor m then ve The Aterminant of «square mattix lay em: B12, ‘Sample ot ramets are het of ol ner the tal ol mae tee aneite soa sre mien, A he anaes at Po ot sel ary ee Seng 0th mr Se ge nts :2 1 Mises 4x0 Mmatx Onweastons We itzmfuse w eaeine notation for deenminants formed frm clments [ese inset fate) an, Oana) ‘ “ce fe oe apn ho "The determinant (2) is called 4 wir af A of orer p, prod SHES HT MANET SH Che oS hy Sm A rectangular sa sins kon) tn (MQ) nae a(hecg) (ERSESISES rem) ‘Dh inary (2°) im which In the notation (2) the ceterminant ofa square matrix be written ns flloms: A weal go) ha ooo fet ae allot principal aI3 an ‘The largest among the ondry of the nos-zero anaes generate Dy a invtin is ella the monk of The atin, Tr the ra of ootuneular halt 4 of diesen n % 4, them ebviewly rman) Uh retenglae mattis vomiting of shite enim is callad mcolunn mate end il be denoted by (ry 26... 20 ih rectangular mates consisting of single re Vetoes! ‘salle a rove marie wil be denote By [3 9-5 "A oquare astin in which all the slenenteauteide the iain diagonal $2. Auosanas asm Mesritseano om Mvtawes a 4 0...0F Ody 0 Jo ota cated diogonad mati and ie denoted Wy? adg'¢ or by (ay dy ooo Ba Suppose that’ m qusatitis 9.5 Ym hive Tinea wn homegeneos cxprenon interme of other qnantites Fyre Aan thy that = th Hat Hs be, 7 He Mat Eat Ee Og oF more cael 12s) « ‘The transformation af Ue anaitcn F444 Ste she gamatition SY Been by meen te forma 4) Kae as ft ian ‘The coefficient af thi tnnsformation fry reenter mats (2) sf dimension m % 1 ‘The linet trunnormstion (4) determines the mats (1) mqtel nd Tn the nest setian we shal eae he ase epovation on retangalar ties using the peerties of the eae 3msformat ‘ng poi $2 Addiion and Multiplication of Rectangular Matrices ‘We shall define the tase operations om mateaes: addition of matrices, sukiplication of x matrix by a numer, and multiplication of matron 2, Suppone that the quantities 9... fe are expressed terms of the quantity 2u 2s By ments of the lieu transfor ation Son 6 ) o eB et rns amt af EB4 1 Mntoes ax Maer Oreacoss and the quantities itor of tho sane quantities 86 ote tna of the teanatormation ee CH Bem o Thea 2m. o ni Sethls 6 In aworlanoe with this formate the fllowing etinition. Dermsinox 2: The sa of tro v9rtongalar maricee A=! ay band Be | hal both of dimension ia % mie te mate C=! ey [of the some dimension schon elwanth ave Ye sane of fhe eoresponding element of the « %. “e tla ( he operation of forming the sno af given wotices scaled addition, Beomple Weel la late ata ata] Definition 2 ony setae matric eal mens int matrie of the ranafori tion (2) fete sim of the corficiow! Metron of te tanEormations (5) and (6) "Fn tho dofition of watri ston it Sows immo the shi ‘operation hae the properties of cmmutatisty acl associativity Lateaaea 2 (sm) FOS A+ REO) Here, B, aul ( av abitety restaneularatiees al of equal dimension. "The operation of alton of matricenexteuds vx nator wa to the co ofan arbitrary Snite mua of sands 2. tet ue mltpy the qoanttios ys. yneo--ote Mk the eausformation 5) hy some number a of &"Thes Masa) te GHB mh ce with this, we oenulae te folowing definition $2 Ansenox axp Mruzmnicsnis of Marstons 5 Denmos 2. The product of a matrix at A=I2.-ss8) bya umber auf Pisthe matris C= 1 ry ms aL Soya) whore elements are obtoned from the vorreqpending He faonts of A by malipiation by c=nd, where 4 ee The operation of forming the product of a matric by « wander ix ett multiplication ofthe matrix bythe sumer Reample 1 Mimonsion ng wand ace ‘The difference A ~ of two rectongular matiers of ena dimension a defined by AaBa4s ve, He A ina guace mates of oder wala member of thew ad 3% Suppose thet the quant pe quantities ety cose are exprened In emmy of ‘We quantities ysis us by the transformation Beate CSU) o ‘and thatthe quantities yy, yp... ye ae exprres in terms ofthe goat Fa dees oay by the forme ne Baa 2am ® Then onebititing hae expres or he 8) xe ‘transformation : 7 ren® 1, Marnces ann Marie Onisunoxe Gam 0) ZenSiraer Zi Sioa) In seondance wth tis we formulate th falling defini Dvversan & ‘The prac! of tuo rctngsler malice lay tase | jie bans | le an Ban yg Pago ba ek Beton Oy Banoo AG 4] Bec eel nwhich the eeoment cy th tert of the £08 owt te clan Gt the predict" of the throw of the fort matrir Anta the 30h ola ofthe ond matris B Bint Gems ooo a The operation of forming the grata of given matrices i aed wetrir ulipietion. Beampte, ie jase wh ll om RJpetseren eh cadetars an ten tan stale a | Tie te Bes ba Be Day Da a ve By Detiniton the wticent matrix of the tuneformatin (10) the rod f the cut iint aries of (5) a (8) Note tht thr operation of selipication of oa retanyuler matrives om nig De cari ont when the mune of claus ofthe fist fertor set tothe member of rons of the xeon In paren, sedition anes pone wh bol factors rr ahaer ears aoe mH $2 Aupmns ws Mceeinseacto om Mataiess 1 ‘aye rder should verve tha even in this apesial case the mitiplettion Tinatrces dost have the property of coumutatiity. Bor example 8 SLE th (a alls dl HE AB= BA, then the satrices A nnd Mar aed penmtable or commuting, Beample, The matrcoe ‘re permutsbl, brotne 1-6) tis ery cont verify te nna and ls te derbi properts of wih aon wih reap ta mi Limo = 4180), 24 Mem 40+ Bo, ipbeation exten natal wa ae ‘ae of several fase ‘When we make a ofthe mulhintivation of retagtlaewatees, we eam wile the ree tratortstion nats tnty boos Hye a nage Hag i Sane, bata ats to bt asinge mates equation st a Jou eel es Qn Om zy lon hana on a8 1, Marston aa Maru Orenenioss Here £5 (ay ta. 44) aM. Y= (Yeoh) A clusen ein ‘and A= | a | in a vetongolar matrix of dimension mm Tet us feat the special cate when in Phe praduet C= AB the second fectar is a uate dinar mateie B= joy .-)- Then i flows from (11) thit tered CEL Bm FEMA we fe Sante Bt) fab ann tome) ye bes oa saad Hao ad Ae a a Sina, 2. 00 | ag all a An Be | Decco | em eae, | Bede oad baa Paes dee Teac: When roar mar ie maint he it (et ty pal mi at), thew tba (eof te at Thali tates ronets 4 Suppose hat «square matrs = jg | Ie the produet of two retan= (Golar matrices Ao and B= Dy | of dimension e Xe al Xm, fer Pon <= Bam | Jean +> fim || [itn Mae es tae | ia jectp oe a fy I it pik a wo = $2 Aono xn MoverescarIO a» Marmicos * ‘We shall eotalish she important Binet Courhy formula, which expremen the detersinant (| ln tems of te aesnone of ant Jon On ea, 2 na [Poa = Bn | : ~Lenen® once) ce Maa <*> So hi rl cote ttn apa 2 aa see (Set ae a SL According to this forme the determinant of © ihe avn ofthe products of pousibe minors ofthe mast (oth) oer” of int the curvspond- ng minors of the eae order of B Derivation of the Binet Cauchy formula. By (18) the determinant of ‘Cean be represented i the form ay |aobow lea bet o- Manbee | Be ET phat ba tm > then among the numbers #4 2 eam two that are equal, eo that ore a (15) Wer. Hosein isc [o Now etm sm. Then nthe sm on he ight hat side of (2) a he ena inh ae te he op yon {fresh All te rominine sans of (15) eon be itn grupo ‘mers ech hy combing none pp tens naman ter from enh other cay inthe onder of the wulseipts eu voy (90 hat dy there are always at ‘stemond on the Hight hao se of he mate an Go mt ave ns fee Se eairtand win ot (8) a0 6) are to We sent etersw 1 Blastces aso Blase Openers within each nek gop the bso ay, #5. BAY ON alte sate ft of value). Now seit ene such group the som of the correspon Zretayty coe anda, BoB Yas Bat os Br AG OE) Bette: ad ada o> tae Te pi ke aG go mpalt $e) ence from (15) we btn (1, Example ae ortanet ten added toe ede) ja sect | A Bectieet oot tee tart bon gH Bae | ie al ‘hertoe formas (14) yids the nll Cauchy entity eet aeet de nda ea) [sgh on Bec he Bag RA Be on aed 1,2...) inthis, we ob or Setting = 6 W= de abode eorapet ce ME ST | Haj and 6 (F125) are eral mumtens, we deduce the welhnown ‘equity dvb ade tod ete the equity sign bole i and ony if all the snbers a ane propor tional to she conrespig mambers (11,2)... m) Beample 2 (heed onan OLE bea.) ja sehen are tha lancal sea nets lee Here CEC... de ia sama ane of the byes my Eee SN hee nee Pel = $2. Annis Axo Mruewnacano oF Mermus n smerefore for # > 2 eu Oh os et By eel bak Let sons the spell ete where ane Bare sure mates of ane and the rine ender n. When we set w= in (1H), er ive at te ‘wal-koown multiplieetion theorem Zor determine: ate AB\=|4\-1B) as) tet, in nother notation, ‘Thum the determinant ofthe pratt of two sqwere warios ds opel ‘he predct of the determinants of the faciors ‘% The Binet Cauchy formu eas ws, the Roueral cise keto expres ‘he minors of the product of ts rectangular mation Sa termi of the ‘minor of the factors Let A= Ileal), B= lls 2 C= Hoy o= a2, Weooasider an sebitrary minor af C 9 ee mats Lome fr th Memes this minor i the ra fro rectangular masrives: eae sii Sh 8 = $3. Sqvaae Marices 13 We onus @ polynom eld (integral ratione!fanetion) with orfiients i He af et Thea by f(A) we shall ean the mate HA) tgt A ot gd ‘We define in this maya pipe in @ mari Suppo tht ff) th prt of wo plans (1) ond). Hey = 1k, on ‘The polynomial [() is obtained from y(#) and hj) by mltpiestion term iby term and coetion of sinilar terms. Uy thi We anak uae ofthe malt heaton rule for powers: #- =P =*, Since ll thew operations remain ‘aid when the sealar ironed by the matris Soll fom (20) that Kaya atayncay ease, in parton! DLA)ACA) =A} AY 22) ‘ey two polynomials in one and the same matrir ere always pormndable Beamples et the aquence of elemen's a for which k— [eal be called the ph neperiisgonal(enbtioge ltegte by "the sgusge mutes ones win which aitth elements of he first superdagonl stv nnteand al the other elements fe sero, The matsix HF will ln he denote simply by Hl Then cto tee os lone ho we 10), Win vor eatrngu 1, Barmcat ann Blarere OvaxaTios By chose equations, Similarly, i's tbe square matrix of ender» in whic al he ements of ‘he few sbiggnel are nite aed al others ar ze, et | no Joao nF PS j Neca ‘We leave i tothe rear to verify the follows properties of the matrices Wad F 1 When an acliteany rvetanpelar watrir tof dimension 0 is ultiptet nthe Left bythe mate TE (eB) of order my hen the rs By dare shifted upwast (or dnemeart) by one plas, he first (lal) raw CE ecppnars andthe Tel (fiat) row of te pret ¥6 fied by cero or example yo 2 of fos meee sel te A oo ys aS as al poolaaaal oooh yoo op ls oo hee Bm ALi ma Dione wal (mmm al 2. When an arbitrary rectangular mati A of dimension we Xa 46 Imultiplod on the right by the mates IF (or B®) af order ny then all te aumas of ave sifed ta the ight (eff) by plac, the fast (erst) ‘St of. icappeas nd Phe frst loc) alum ofthe producti ied Tyee For ample, $9. Syeane atoms jane mm [OPO E! yo ees] oss elle mags Ci jessy acai (ereel taco 1B Asgusre maisin Ai called singuler if |4|=0. Ocheewio ix ellad ssnsingler Tet A= [aif be » nonsingular matrix (| 0). ‘he ine tenformation with eoetfcint matrix A Let on consider Bam e we diun When we rpard (28) as equations forza 4, determinant of the system of uations (23) i and observe thatthe Secumption, different ‘rom nero, thes we ea express, +34 im terme OF Yoo Me BF smeans of the well-nown formulas: oon toast Bove Me ema te tou os A eottebe leans Mat Be Musnn Ml We nave thus obtaied the “inverse” tion (25) susformation of the teunstorma "The coefsivnt tmatrix of thie transformation Mt ibe called the invvrse matrix of A. Prom (24) is easy toate that es “ Beem 5) 2 YB ai the siebeie complement (ie ofr) of the lement ay the determinant | | Gt ay6 1. Marmcas aso Marsex Orenumioss Por example. if ees seam tee edad - ral ile: eet aha Venter amas tema [By forming the compasite transformation ofthe given trenaformation (23) apa the nverne (24), either order, wecbtan in bath case the ett transformation (withthe wait mitts ax coffcint mati); thenefre at z. e8) The validity of equation (25) ean aio be established by dineet mil ‘ation cf the matroer 4 and A“ In face, by (25) we have? Say Bout sy, ay as Smiley Laaaye= Sein chy SAwu=by iad is ea tee thatthe matrix oguations ax and X4=B (|4} 20) on Inver anton other than X°— A! or by sltiplyng oth ses of he frst (weund) exmation ow the Tet (right) byt" a os the as lative property of mate neitpieaton weber (26) in both ‘eon ef cuatro fos C9) tan oe ‘psa wher Ta et $2 Soran Baress ” Inthe same way it eam be sh that eae of the mtr equations AX=B, EA (Alo), where X aod Hare rectangular matics of equal dimensions yuwr mites of appropriate order, have onc and only one soltin.| x “Band X a, (2) epottvely The meteons (20) are the "eft" and the ‘right™ quotients fn "viding" B by 4. Prom (23) ed (28) we deduee (se p. 12) that ey Ste 0 Sty. 80 that r= Ty. On comparing this with (28), we ees When a rstongeler matric is multiplied on the lft ot om the right By enomsingulr matric, the rank of the original matrix remains wnchanged Note that (26) implios| 4 |-[ At ae Matar Por any two non-singular matrses we bare (Bye (2) |& All the matrices of order» form a ing” with nnit element 5. Since inthis sing the operation of miplintion by a namber of x ie tne, and since there exists a bass of w inal fcepertent ates kn terms of which all the matics of order w ean be expressed linens the ag of matrins of order an alee” Seer et sr mtn ot af fe Seen oh sommaaie {sad danine pecs wet ews mss ew rene By for aon nmr Wai, Madr era "eran avery matin A= [ith lent 7a eee fn ‘Me tom = 3 uaa wine Bae hematin rer x whch tee 2 the6 1. Marwons awe Mares Ootsamows All he nquate matrices af otro form a oomumatative group with respect tothe operation of aition Al th om-snle matriges f order form (aan eomamttive) gra ith repel the apetaten of maison ‘A equate mute A= lay 3 called per frienralor lower tea gular) Hell the slate Bele (are) the main diagonal re zm Scep ccs] es | eo ban moe A diagonal matric i spovial ease beth of an per tring mts sand a lower ernngidar mata ‘Sinee the ietermingn afm (rangular mais i equal to the praduet of ite diagousl cements # tehanzae (and, Ie ettienler,w digo) toute ‘swon-singularif aed onl If lls Uiggonal lementsateditferent from ze). Te ie ny to weit that the wm ad the prvdet oF te ding (upper triangular, lower range) mater cago (appr ruta oer trimester) mites wn tt the inerse of gq non-sivgular diagonal (upper triangle, ower tranawlar) mutrivew matri of the same typeof 1. All th dingonal motion of ander» form aeommtatve grap water the operation of wdtion, me all the upper teinngnlar maviow or all he lower tranglar wari 2. All the nung Aiagvnal nates fore « commutative group ‘ander mlipiaton 4, All the susingelar upper (lower) trangelae ma commutative) group under aipleation joe oom 4 (non 44. Wessoneludethisstction with a forthe importa operation ox matrees ~sramepoition 1 tie operation hs he ssc preety (a2) ema (4 rope Cocming the group ce me foreange 8), a 8 POR os $4 Courocsn Maracas, ats » 2 occymy RET 2,000 the the fromapone AF fj, whore of 0a EHD, ym ee) eA is of divnsion mm then A feo dimen 1 oan to eorife the felling proportion" L(A Byam, 2. * a. wa, a aysiu Ma square mate: $= || 4g eoinctue with ts ranapine (8! = 8), then ‘vis cll symmetric. In spmmetei mtris elements th sre aymmet ally placed with zespet to the main degonl are equal. Note that the Drotuct of two symmetiie matrices ie not, i genera, npmmetre. By 2, nis holds if and only if the two given symmetric mations are permutabe fa sqvare mst Kf difers fem its traaspee by #Saetor = (T=), thesieiscalledsheoaymuctric. Tu askew pinetrie mate any to lements that are symmetrce to the main digponal differ from ach other by a factor —1 and he dingo eaments arenes. Prom 3 it fotlows that te produet of two yerataie skew-symmetric rateees i 4 symmetrio ates.” $4. Compound Matrices. Minors of the Inverse Matrix A, Let A=| af bem given matrix, We consider al posse minors of of onder p Spm) 4G) (SES Ss) a Theme fh in 8% ene X'=(2) the aber of bin tion of» bjecte talon pat a tiie. In order to arrnege the minors (31) in A aquare array. we enumersto ip suze definite gedee—esicographis ore, for exampie—all the ¥ combinetions ofp indies selesed from sng the indices 12,0. somaya prtivns wo feec. Ie tdeas wens ma nmeéaviae a smite msi C1 fh) fe ssp mins 1 (7.$4 Conroesn Mareces, Misine a 2” 1 Matmeps Axo Matnix Orumarioxs ¥ 1 the combinations of noe is <4 < ooo < foam hs mae vanced cael) aoe bp=o. 09 Ayo, hers cia - soot ste Indice 1,2, Sova gewencky Com prion oF (AB with C247 ane ($A) sows that the equations (38) ae tai eh ther th ant (89) at eer ‘Singe the element Buy of the averse matric uf ty aster amined br (38), equation (38) mort bold CHAPTER IT {THE ALGORITHM OF GAUSS AND SOME OF ITS APPLICATIONS 51. Gaua’s Elimination Method 1 bt aaah Hash bar eat ber a eset hate t+ OI bbe anyvcm of» linear eqetions in » unknowns 24.1 od sides 4, “mst form this ses maybe written a with ight: ae a Bre 2 (25, 405.00420) amd 9 leah isthe sqvare oeficient matric edison singular, ther we ca rewrite thin a ym) are columns and orn explicit fom: ‘Thus, the fost of computing the elements of the inverse mattis Ana gghlg i ogaialent to he tock of solving the wate of eqntions (1) tor rbtrary sight nl sides jy Yo The elements of the inverse ‘atex are detcruiged by the formulae (25) of Chapter I Iowevee, the acto cemputation af the elements of A hy Use formulas i very teas for inge x. Therefore effeive methods of empating the terents of a matrix—and hence of solving w system of linear equatons-—are of seat prneticl a” Ue] aca rap oe rnc ante Pope a Sao a 3 NS24 TE Tex Ataonrruat op Gate ax Sows AvvutcaTIos Inthe present ehapter we expound the Portia bass of soe of these outhodssshey are voriante of Geusts elimination method, whose aequsint Ue the render fiat onde hie algabra sone at seal 2 Suppose that in the system of equations (2) we have ay 0. We ‘minute 7 From all the equations begin withthe son by ading ‘othe secon eqastion the first mltpled by — 23, to the thin the fist snatigtied by —, and The system (2) ‘ts now been replaced hy the equiveleat system aunt at ve ‘econ of the annoy ae tense tome of the at 1 sto ee eve ty te formas, Suppose that af) 70. Then we elim 1 ae ani of these (8) and ot the aes han Fae He corey ars aan pelea? ‘eae ote an theme sgh sis coneted it he prectig eet by te formal a o_o eau ay, aw 6 i @ Fae? Spat al pe oF Contnwing the eleorithe, we goin — 1 steps fom the original stem (0) to the tlnngularreourrnt syst peer deadline tally, tosime the principe of lineer sxperpention of ores: 1. Under the combined action of te ystems of forts the coresponting ioplncomente are added together 2 When the magnitudes of al he forces arc mati by ene andthe ane reat nwober, then al he placements are msltiplad by fhe same suber ¥ t 2. Bleonastean, Ierenmnararion o Gaur Ataourame 2 “We dencte by ou the eotficent of ingluence vf the point () on the (Os bey the displacement of (2) under the seon of & unit forse Biplist at (2) i k=1,2,..-0 0) (Big 2), Then under the eambined ution Pine foees Fi, Ps... Pe the displacements ya, re deteranined fy te formals ». (3) ‘Comparing (28) with the origina! spstam (1), we ean interpret the task ot nlvng the wate of equations (1) us follows The daplacoments va ye... Hs bring item. ae are required to Find the cearespning forces Py, Py Pe ‘We denote by Sythe stativl system that is braid from 8 by introdue ing p fixed hinged sapport atthe pists (1), (2),---, |p) (B= 8). We denote the confieien’s of Influence for the reweiing wovable points GED. fm) of the apts Ny by Bo kept) (we Big. for p= 1), ‘Tha worsFciot of cen by regarded a the diaplasnat at the point (i) under the aetion of aie face at (4) anda the retina By, Ry | the fz points (1). 2h... (ble Therefore pquO® the other nana, under the same forees the daplacaments of the | feiten 8 atthe pointe (1), (2),..-5 (ph aresero a Roun Biny += 0 | -s