EgYhIfEn that if A B In
for a nxn matrix A then BA In as well
On the other hand if A is an mxn matrix
B S t A B Im then it is
wtf
FfÉ
we shall
thtInvof
here
AB 12
B
butBAFII
In
today's
class we
first address this question
of rectangular matrices in
general
the we learn
along way
column sank row rank
rank nullibs them
Let A be a mxn matrix
R A linear span of the rows ofA
Row space of A
R A IR dim R A In
This is because we know that
B is LI
if B is a basis of R A then
1 I subset of IR has cardinalitis
any
n thus I B I n
dim R A now rank of A
dim C A column rank of A
FÉ A can be considered same as C AT
CA can be considered same as R AT
We shall show
row rank of A column rank of A
Let R be a sow reduced echelon matrix and
let r be the no of non zero rows of R
then row rank R r
Now
if R R2 Rr be the sows
consider the combination
α R α R 0
th
then the K component of α R 2 Rr
is α Kith component of α R α Rr
is α
α 42 or 0
row rank R r
what about column rank of R
K C.kz Kr
ᵗʰ column
If K 1 then
any
column prior to k
is a zero column
For
any
other column I i s.tk j kit
ChanFk ᵗʰ columns op K C
j
i K Cj Kies
Case Suppose s
then all
bij th entries with esi are 0
Tet be the entries I i
Gj i
Cj as Cke
so
CFisbans C 12
and have 1 s at different position
they
with all other entries are 0
subset of std IRM
i e an basis of
hence linearly independent
dim C R r
so column rank of R r
So in a nutshell
For a RREM R
column rank of R row rank of R
5 the no of nonzero rows
Now if R be the row reduced echelon form
of a matrix A
Then rows of R E R A
row space of R E R A
Also know A R
we
EEE
finitely many elementary matrices
A
igniffling many elementary
matrices
Thus rows of A E R R
R A R R
So RCA R r
now rank of A r the no of non zero
rows of the RREF of A
what can we about columns
say
Im Let A be a mxn matrix and D be
an invertible matrix mxm matrix
Let B D A Then
i N A N B
2 column sk of A column rank of B
Note as a result
A EI ET R
Column rank of A
column rank of R
row rank of R
row rank of A
matrix A
Corollary for any men
column rank of A
row rank of A
rank of a matrix A
This is defined as
Also
no of columns of A n
r n r
Is non zero rows of R
rank of A Nullity of A
FNA
Known as the rank TullitisTm
For matrix A
Imf any
myn
n rank of A unllitis of A
1 e
x EN A An 0 DA x 0 Br 0
KEN B
NCA N B
Now A D B N B E N A
hence N CA N B
2 B DA
If Cj's are columns of A
Cj's are columns of B
then DC
Cj
let column rank of A K
and re No be a basis of C A
i
Clearly
0 dig Cj
Define Wi is Cj Do
Then coi we is a basis of C B
hence column sk of B K we are done with
the proof
claim we LI
First we w are
Subfuse α wi 0
Ex a 0
D
divi o as D is invertible
α o i
So w we is LI
Next we show w we spans B
Let w E C B
Now w Epic β C ECCA
β DC 855 94
D I B g
EECA u
Vivi E 5 Du
Et r w
C B
w wa spans
Hence w wu is a basis of CCB
There fore
column rank
of B K column tank of A Go
The system of egu Ax and
Lmark b is solvable if
only if rank A rank Alb
Coming
back to the question when
an mxn matrix A is left invulible or
invertible
right
we have seen for square matrix
it is same as invulibility
Here we
give condition in terms of rank
Im Let A be an mxn matrix
A is invertible rank A r m En
right
matrix c t AC Im
s
II suppose a
C nxm matrix
columns of C then
If C Cm
have A 5ᵗʰ staba
we
Cj gang ARM
ym
what does this mean
AC
if C
z α A t In An
where A Az An are columns of A
So it
ej
E C A I
says
IRM CCA Rm
C A z IRM
dim CCA m
rank A m
We already know rank A is the number
in RREF of A and hence
of nonzero rows
rank A n
So rank A m n
get
we
E Conversely
If rank A m En
given
then C A IRM
ns.t AC e j 1 m
Cj
Elas
C Im
Ime A mon matrix
is invertible rank A n Em
A left
If A left invulible
AT night invertible
nxm
rank AT n m
E rank A en Em
