TERM PAPER ENGINEERING MATHEMATICS MTH101

Topic:
Illustrate the definition of rank and its characteristics by column vectors with a 3X4 matrix. Show that for a square matrix the linear dependence of the row vectors implies that of the column vectors and conversely. Show that for a non square matrix either the row vectors or the column vectors must always be linearly dependent.

DOA: 20

th

Oct, 2009.

DOS: 30thNov, 2009.

Submitted to:
Mr. Naresh Kumar Deptt. Of Mathematics.

Submitted by:
Abhishek Sarkar Roll. No.R !90"#$0% Re&.No "0902"2"%%% lass #.'ech(M)*%

ACKNOWLEDGEMENT
+ take this opport,-it. to prese-t m. votes of tha-ks to all those &,i/epost 0ho reall. acte/ as li&hte-i-& pillars to e-li&hte- o,r 0a. thro,&ho,t this pro1ect that has le/ to s,ccessf,l a-/ satisfactor. completio- of this st,/.. 2e are reall. &ratef,l to o,r 3OD for provi/i-& ,s 0ith a- opport,-it. to ,-/ertake this pro1ect i- this ,-iversit. a-/ provi/i-& ,s 0ith all the facilities. 2e are hi&hl. tha-kf,l to o,r s,b1ect teacher Mr. Naresh 4,mar for his active s,pport, val,able time a-/ a/vice, 0hole5hearte/ &,i/a-ce, si-cere cooperatioa-/ pai-s5taki-& i-volveme-t /,ri-& the st,/. a-/ i- completi-& the assi&-me-t of prepari-& the sai/ pro1ect 0ithi- the time stip,late/. 6astl., 2e are tha-kf,l to all those, partic,larl. the vario,s frie-/s , 0ho have bee- i-str,me-tal i- creati-& proper, health. a-/ co-/,ctive e-viro-me-t a-/ i-cl,/i-& -e0 a-/ fresh i--ovative i/eas for ,s /,ri-& the pro1ect, their help, it 0o,l/ have bee- e7tremel. /iffic,lt for ,s to prepare the pro1ect i- a time bo,-/ frame0ork.

INTRODUCTION
What is a matrix? 'he matri7 is a recta-&,lar arra. of -,mbers. A- item i- a matri7 is calle/ a- e-tr. or a- eleme-t. S.stem of -m -,mbers, arra-&e/ i- a recta-&,lar formatio- alo-& m ro0s a-/ - col,m- a-/ bo,-/e/ b. the brackets 89 is calle/ a matri7. Matrices are a ke. tool i- li-ear al&ebra. O-e ,se of matrices is to represe-t li-ear tra-sformatio-s 0hich are hi&her5/ime-sio-al a-alo&s of li-ear f,-ctio-s of the form f(7* : c7, 0here c is a co-sta-t. 'he matri7 m,ltiplicatio- correspo-/s to compositio- of li-ear tra-sformatio-s. Matrices ca- also keep track of the coefficie-ts i- a s.stem of li-ear e;,atio-s. <or a s;,are matri7, the /etermi-a-t a-/ the i-verse matri7 (0he- it e7ists* &over- the behavio,r of sol,tio-s to the correspo-/i-& s.stem of li-ear e;,atio-s a-/ )i&e- val,es a-/ )i&e- vectors provi/e i-si&ht i-to the &eometr. of the associate/ li-ear tra-sformatioWhat is the rank of a matrix? 'he matri7 is of ra-k r 0he- it has at least o-e -o-5=ero eleme-t of or/er r a-/ ever. mi-or of or/er hi&her tha- r va-ishes. #riefl. the ra-k of a matri7 is the lar&est or/er of a-. -o- va-ishi-& mi-or of the matri7. 6i-ear al&ebra st,/ies li-ear tra-sformatio-s 0hich are represe-te/ b. matrices acti-& o- vectors. )i&e- val,es, ei&e- vectors are properties of a matri7. 'he. are comp,te/ b. a metho/ that &ive importa-t i-formatio- abo,t the matri7 a-/ ca- be ,se/ imatri7 factori=atio-. 'he. have applicatio-s i- areas of applie/ mathematics as /iverse as fi-a-ce a-/ ;,a-t,m mecha-ics. A-. ;,a-tit. havi-& - compo-e-ts is calle/ a vector of or/er -.

Computation - Findin the rank of a matrix !!!!!!! 'he easiest 0a. to comp,te the ra-k of a matri7 A is &ive- b. the >a,ss elimi-atiometho/. 'he ro05echelo- form of A pro/,ce/ b. the >a,ss al&orithm has the same ra-k as A, a-/ its ra-k ca- be rea/ off as the -,mber of -o-5=ero ro0s. o-si/er for e7ample the !5b.5! matri7

2e see that the seco-/ col,m- is t0ice the first col,m-, a-/ that the fo,rth col,m- e;,als the s,m of the first a-/ the thir/. 'he first a-/ the thir/ col,m-s are li-earl. i-/epe-/e-t, so the ra-k of A is t0o. 'his ca- be co-firme/ 0ith the >a,ss al&orithm. +t pro/,ces the follo0i-& r o0 echelo- form of A:

0hich has t0o -o-5=ero ro0s. 2he- applie/ to floati-& poi-t comp,tatio-s o- comp,ters, basic >a,ssiaelimi-atio- (6? /ecompositio-* ca- be ,-reliable, a-/ a ra-k reveali-& /ecompositio- sho,l/ be ,se/ i-stea/. A- effective alter-ative is the si-&,lar val,e /ecompositio- (S@D*, b,t there are other less e7pe-sive choices, s,ch as AR /ecompositio- 0ith pivoti-&, 0hich are still more -,mericall. rob,st tha- >a,ssia- elimi-atio-. N,merical /etermi-atio- of ra-k re;,ires a criterio- for /eci/i-& 0he- a val,e, s,ch as a si-&,lar val,e from the S@D, sho,l/ be treate/ as =ero, a practical choice 0hich /epe-/s o- both the matri7 a-/ the applicatio-. 3e-ce the ra-k of the &ive- matri7 is 2.

"#O"$#%&$S: O-l. the =ero matri7 has ra-k B0C. 'he ra-k of A is at most mi-(m -*. f is i-1ective if a-/ o-l. if A has ra-k -(i- these case, 0e sa. that A has f,ll col,m- ra-k. f is s,r1ective if a-/ o-l. if A has ra-k BmC. +- these of a s;,are matri7 A (i.e., m:-*, the- A is i-vertible if a-/ o-l. if A has ra-k -. +f # is a-. - b. k matri7, the- the ra-k of A# is at most the mi-im,m of the ra-k of A a-/ the ra-k of #. As a- e7ample of the DEF case, co-si/er the pro/,ct. #oth factors have ra-k ", b,t the pro/,ct has ra-k B0C.

+f # is a- - b. k matri7 0ith ra-k -, the- the A# has ra-k same as A. +f is a- l b. m matri7 0ith ra-k m, the- A has same ra-k as A. 'he ra-k A has ra-k -,k of A is e;,al to r if a-/ o-l. if there e7ist ai-vertible - b. - matri7 G s,ch that

2here + /e-otes the r b. r i/e-tit. matri7 i- li-ear al&ebra, the i/e-tit. matri7 of si=e - is the - b. - s;,are matri7 0ith o-es othe mai- /ia&-ol a-/ =eros else ever.0here. +t is /e-ote/ b. l if the si=e is immaterial or ca- be triviall. /etermi-e/ b. the context. C'A#AC%$#$S%&CS OF #OW AND CO()*N +$C%O# #o, -e.tor Co/umn -e.tor

Standard matrix notation

A/ternati-e notation 0 A/ternati-e notation 1

$($*$N%A#2 %#ANSFO#*A%&ON OF A *A%#&3: 'he follo0i-& operatio-s, three of 0hich refer to ro0s a-/ three to colo,m-s are k-o0- as eleme-tar. tra-sformatio-s: +. ++. +++. +$C%O# A-. ;,a-tit. havi-& - compo-e-ts is calle/ a vector of or/er -.'herefore, the coefficie-ts i- a li-ear e;,atio- or the eleme-ts i- a ro0 or col,m- matri7 0ill form a vector. 'h,s a-. - -,mbers 7", 72, ....., 7- 0ritte- i- partic,lar or/er, co-stit,te a vector H. (&N$A# D$"$ND$NC$:'he i-tercha-&e of a-. t0o col,m-s (ro0s* 'he m,ltiplicatio- of a-. col,m- (ro0* b. a -o- =ero -,mber. 'he a//itio- of a co-sta-t m,ltiple of the eleme-ts of a-. col,m(ro0* to the correspo-/i-& eleme-ts of a-. other col,m- (ro0*.

'he vectors 7", 72,........,7- are sai/ to be /inear/y dependent4 if there e7ist r -,mbers I", I2,..., I-. Not all =ero, s,ch that I"7" J I272 J....J Ir7r : 0

if -o s,ch -,mbers, other tha- =ero, e7ist, the vectors are sai/ to be li-earl. i-/epe-/e-t. +f I" : 0, tra-sposi-& I"7" to the other si/e a-/ /ivi/i-& b. 5 I", 0e 0rite i- the form 7": K272 J K3 73 J....J Kr7r 'he- the vector H" is sai/ to be li-ear combi-atio- of 72,73,........,7-.

$3A*"($:- Are the vector x": (",3,!,2*, x2:(3,5$,2,2*,x3:(2,5",3,2* li-earl. /epe-/e-tL +f so e7press o-e of these as a li-ear combi-atio- of others. SO6: 'he relatioi.e., I"7" J I272 JI373 :0 :0

I"(",3,!,2* J I2(3,5$,2,2* JI3(2,5",3,2* ,

is e;,ivale-t to I" J 3I2 J2I3:0

3I" 5 $I2 5 I3:0,

!I" J 2I2 J3I3:0, 2I" J 2I2 J2I3:0 As these are satisfie/ b. the val,e I":", I2 :",I3:52 0hich are -ot =ero, the &ivevector are li-earl. /epe-/e-t. Also, 0e have the relatio-, H" J H2 J 2H3 : 0 #. mea-s of 0hich a-. of the &ive- vector ca- be e7presse/ as a li-ear combi-atioof the others. Appl.i-& eleme-tar. ro0 operatio-s to the vectors 7", 72,73, 0e see that the matrices 7" A: 72 73 #: 7" 72 7" J 72 M 273 a-/

have the same ra-k. 'he ra-k of # bei-& 2, the ra-k of A is also 2. Moreover 7", 72 are li-earl. i-/epe-/e-t a-/ 73 ca- be e7presse/ as a li-ear combi-atio- of 7" a-/ 72. Si-ce 73 : (7"J72*N2. Similar res,lts 0ill hol/ for col,m- operatio-s a-/ for a-. matri7. +- &e-eral, 0e have the follo0i-& res,lts:

if a &ive- matri7 has a li-earl. i-/epe-/e-t vectors (ro0s or col,m-s* a- / the remai-i-& vectors are li-ear combi-atio- of these r vectors, the- ra-k of matri7 is r. co-versel., if a matri7 is of ra-k r , it co-tai-s r li-earl. i-/epe-/e-t vectors a-/ remai-i-& vectors if a-. ca- be e7presse/ as a li-ear combi-atio- of these vectors.

QUESTION2: Show that for a non square matrix either the row vectors or the column vectors must always be linearly dependent. ANS2)R: 'aki-& vectors 7"(",3*, 72(3,5$* a-/ 73(!,2* 2riti-& i- the form of vectors i.e. I" J 3I2 J ! I3 : 0 ...........(i* 3 I"5$ I2J2 I3 : 0 ..........(ii* #. solvi-& these t0o e;,atio-s i.e., m,ltipl.i-& (i* 0ith 3, a-/ (ii* 0ith " 0e &et 3 I" J 9 I2 J"2 I3 : 0 .........(iii* 3 I" 5 $ I2 J 2 I3 : 0 ..........(iv* S,btracti-& (iv* from (iii* "! I2J"0 I3:0 "! I2: 5 "0 I3 . I2N I3 : 5$NO . I2 : 5$ a-/ I3 : O .s,bstit,te I2 a-/ I3 i- e;,atio- (i* 0e &et . I " : 5 "3 'herefore the relatio- comes ,p to 5"37"5$72JO73 Appl.i-& eleme-tar. ro0 operatio-s 0e &et A: 7" a-/ #: 7" H2 72 H3 5"37"5$72JO73 'his is the col,m- vector. A-/ the ro0 vector comes ,p to A: 7" H2 H3 #: H" 72 5"37"5$72JO73

a-/ # comes ,p to

3ere i- this e7ample the col,m- vectors a-/ the ro0 vectors both comes ,p to 'he li-earl. /epe-/e-t vectors he-ce the &ive- is prove/. i.e. For a non square matrix either the row vectors or the column vectors must always be linearly dependent. 3e-ce the &ive- propert. is verifie/.