Rank of a matrix, Linear equations, and Vectors in
DeInton. Let A be un mutrx. Let RRE(A) be the reduced-row-echelon Iorm oI A.
We deIne the runk oI A to be the number oI non-zero rows n RRE(A) (or uny echelon Iorm oI A).
Or equv., runk(A) s the number oI leudng 1s n RRE(A) (or the number oI pvots n uny echelon Iorm oI A).
Exumple 1 . Let
!
0
1
5
0 0 0
2 0
3 0
0
0 ) ( with
3
11
5
6 3 0
8 1 8
3 0 4
0
2
1
1
1 4
A RRE A . Then rank(A)= 2
vv/mp Remurk. (Why? Number oI leudng 1s s ut most n).
Exumple 2. True-Fulse. It s mpossble Ior u 5x4 mutrx to huve runk 5! True. runk .
Theorem 1. A lneur system AX=B s consstent Ior ull B n
II (I und only I)
runkA= m (the number oI rows oI A).
Ideu oI FrooI. II runk (A) < m, then RRE(A) hus u row oI zeroes . So the lust row oI RRE(A|B) muy look
lke (0 0 0 |1 ) & we cun not solve the system n such u cuse. II runk (A)=m, then we huve no problem.
Theorem 2. A lneur system AX=0 mples X=0 (only ), or X must be 0, II
runkA= n (the number oI columns oI A).
Ideu oI FrooI. II runk (A) < n, then the number oI leudng 1s n RRE(A) s less thun the number oI columns
oI A.. So some column C
oI RRE(A) hus no leudng 1.
Hence x
s un urbtrury purumeter. So X s not necessurly 0.
Exumple/Froblem 3. True-Fulse. Let A= . Then t s possble Ior the lneur system
AX=B to be consstent (hus u soluton) Ior ull B n
Fulse JustIcuton. AX=B s consstent Ior ull B n
II runk (A) = number oI rows oI A=5
(by Theorem 1). But runk . So t s mpossble.
Exumple/Froblem 4. True-Fulse. Let A= . Then t s possble thut the lneur system
AX=0 mples X=0 (only ), .e X must be 0,
Fulse. JustIcuton. AX=0 mples X=0 (only ), or X must be 0, II runk (A) = number oI columns oI A=6 (by
Theorem 2). But runk . So t s mpossble.
vv/mp Remurk . The lneur system AX=B s consstent Ior u purtculur B n
II
runk A=runk (A|B)
Ideu oI prooI. For otherwse, RRE(A|B) muy look lke
1
1
1
|
0 |
0 |
0
7
4
0 0 0
5 0
3 0
, so the system inconsistent.
Exumple/Froblem 5. True Fulse . In u lneur system, t s possble thut runk(A|B)=runk A+2.
Fulse JustIcuton. runk RRE (A|B) = ether runk A OR runk A + 1 becuuse
RRE(A|B) hus only one more column thun RRE(A). So the number oI leudng 1s muy go up by ut most 1.
Extra Notes/Problems on Sections 2.2 & 2.3
LINK
wheie X he column vecoi
Theorem 1 & Theorem 2 & Lnk huve the Iollowng greut consequences Ior vectors n
.
Golden Rule 1 Fewer thun n elements n
(uutomutcully) cun never generute
.
Golden Rule 2 More thun n elements n
ure (uutomutcully) lneurly dependent.
Automutcully meuns . guurunteed. you do not need to look ut the vectors.
3roblem 6 True Fulse
u) It s possble Ior 4 vectors n
to generute
!
b) Any 5 vectors n
ure lneurly dependent !
c) It s mpossble Ior 3 vectors n
to generute
!
d) Any 3 (2x2) mutrces cun never generute
(ull 2x2 mutrces)!
e) Any 5 (2x2) mutrces ure lneurly dependent n
!
I) Any 3 polynomuls n
cun never generute
(ull polynomuls wth degree ut most 3)!
Solved 3roblem 7 True Fulse (Whut I the number oI vectors s correct no more & no less)?
u) Any n vectors n
ure lneurly ndependent
! Counter Exumple Tuke , 2, 3 wthn
b) Any n vectors n
generute
! Counter Exumple Tuke , 2, 3 wthn
c) For n vectors n
lneurly ndependent generute
! True Ths s Golden Rule 3 (u)
d) For n vectors n
generute
lneurly ndependent! ! True Ths s Golden Rule 3 (b)
Solved 3roblem 8 3rove thut s mpossble Ior 4 vectors n R
WR generuWH ull RI R
YHs, Whs s GRldHQ RulH 1 ( IRr 4 vHcWRrs Q
).
3rRRI LHW
be any given vecois in R
ThHQ WhH HquuWLRQ
in R
can be wiien as
X = B whHrH X Ls
ThuQks WR such LINK . NRw WhH cRHIILcLHQW muWrLx A Ls
To solve AX foi all in R
wH QHHd ruQk A=5 (by ThHRrHm 1).
BuW ruQk A= iank . CRQWrudLcWLRQ.
3rRblHm 9 3rRvH uQy 5 vHcWRrs LQ
urH uuWRmuWLcully lLQHurly dHpHQdHQW.
YHs, WhLs Ls GRldHQ RulH 2 (IRr 5 vHcWRrs LQ
).
HLQW VHry sLmLlur WR sRlvHd 3rRblHm 8. YRu QHHd ThHRrHm 2, LINK, uQd iank .
3rRblHm 10 3rRvH AX=B Ls cRQsLsWHQW LII B Ls u lLQHur cRmbLQuWLRQ RI WhH cRlumQs. (UsH LINK)
Coming soon: SHcWLRQ 3.0 MRrH GRldHQ RulHs RQ
EvHry busLs RI
hus HxucWly Q HlWs,..
