Linear Mappings
Linear mappings
Eigenvalues and Eigenvectors
Linear Mappings
to Linear mappings are functions defined
on vector spaces that preserve linear
combinations
Definition
Given two vector spaces V and W
say that
we T V W is a linear
V
mapping if it verifies tu ie and taek
a Trutv Tcu Tcu
b T avi a Tail
Example The mapping [Link] 1122
defined by is linear
Tf Xy
µ
Example The mapping D P IP defined
by Dpw p ex is linear
Example Let A be matrix of size
a
Mxn The mapping [Link] Rm defined
T E AI is linear
by
Example The mapping filth R2 defined
f is not linear
by
f g Yy
Properties of linear mappings
Let T V W be a linear mapping
then
1 T Ou Ow
2 T f ie T Il
3 T aint azitzt tan Un a Fluitt tannin
11251122
Example The mapping f defined
by f
f g tf
is not linear
4 414
since
11
Kernel and Image of a linear mapping
Definition
Let T V W be a linear mapping
He define the kernel and image
of T respectively as follows
Ker F LIEV TIE 0
IMT TCHEW I EV
E Ker T is a subspace of V
B Im T is a subspace of W
Theorem
et T V 7W be a linear and
mapping
let ai uh ten be a
system of
generators of V Then SHUTThiel Twins
is a
system of generators of Imt
Example Find the kernel and image of
the linear fi 1123 31123
mapping
defined by
Injective surjective and bijective mapping
A function f A B is injective
if and only if
V [Link] if Xty then flat fly
or equivalently
Vx
YEA if fCxt fly then
y
B A function fA B is
surjective if
and only if
Vb EB there exists a cA such that
b fca
A function is bijective if and only if
it is both injective and surjective
T
bi
n
i
3
55
v w
Example T
W
V
Example T
not injective
not surjective
Example T
Bie
V
w
Theorem
Let T V 7W be a linear mapping Then
L T is injective if and only if kerf 253
2 Tis surjective if and only if 1mF W
Other characterizations
For any given linear mapping f V 2W
we have
1 f is injective if and only if for each
linearly independent set lui Ui In
the set fail f lui flint is
linearly independent
2 f surjective if and only if for each
is
system of generators of v lui UI In3
the set 4 flail futz faint is a
system of generators of W
3 f bijective if and only if for each
is
basis of V lui UI um the set
fate futz fluin is a basis of
W
Operations with linear mappings
Given two linear mappings f g V 2W
and a scalar 2E IR we define the
operations
f tg V W f g CEI fix Hgtx
af V W Hf I Ifk
Theorem
With the operations defined above
the set of mappings between
linear
two vector spaces V and W is itself
a vector space
Moreover the set of linear mappings
between V and W has additional
structure
Theorem
Given two linear
mappings f V W
and W U their composition
g
V U defined by go f GHQ
gof I
is also a linear mapping
Recall that if a function is bijective
then it has an inverse function
In particular if f V w is a bijective
linear then there exists
mapping
a function f i W V such that
V TE V f f F J
the W f f in D
Theorem
For each bijective linear function
w V
f V 7W its inverse
mapping f
is also a linearmapping That is
f lair BJ af ca t f o tu r EW
Fa GEIR
Linear mappings and matrices
In this section we will see that linear
mappings between finite dimensional
vector spaces have a matrix representation
The mechanism to achieve this matrix
representation consists in coordinates
using
with respect to some basis of V and W
More concretely let TiV w be a linear
mapping let B be a basis of V and let
C be a
mapping of W For any rector
I EV the
image of I under T is TIEW
The coordinates of E and TE with respect
to the corresponding bases are ITB
and TI Ic We will see that there
exists a matrix MEB such that
TI c MFB I 3ps
The matrix MFB is the matrix representation
of T when we fix the bases B and C This
matrix converts the coordinates IIB into
the coordinates Tx Ic
D The
notation MFB has been chosen to
remark that the matrix represention
of T depends of the bases B and C
In other words it
changes with the choice
of bases
Matrix associated to a linear mapping
In this section we will construct the matrix
associated to the linear mapping T V 2W
We start by fixing a basis for V and
W Let B be a basis for V and let C
be a basis for W In particular let
B hbi 52 In
Then we represent every IEV
can as a
linear combination of B using the
coordinates IIIB That is
I _x b it xz5ztXzbz t 1 Xn b n
j LI B
Now using the linearity of T we have
TI T X b it x5bztx5bzt tin bn
X T bi [Link] zTlb3 t txnTCbn
Since TI is a vector of W we can
take the coordinates of TE with respect
to the basis C
Therefore
TI c x Tibi ctxz Ttbd c t tXnlTCbnDc
This last equation can be written as a
matrix equation
TI c Ttb c Tibi c Tibor I B
Comparing with the equation Ctx MET IIB
at the beginning of the section we see that
MEB Tibi c tcb c TC b n c
Matrix equation of a linear mapping
Let V and W be two vector spaces of
dimensions n and m respectively and let
B and C be bases of V and W respectively
Given a linear mapping T V W the
matrix equation of T with respect to B
and C is
TI c MFB ITB
which given the coordinates IIB of a
vector I EV with respect to B computes
the coordinates Tac of its image TI
with respect to C
MFB is the matrix associated to T with
respect to B and C that is the matrix
of size Mxn given by
MFB ft lb Dc Tl bd c LTCb n c
B The notation MFB has been chosen so that
in the equation THE MFB IIIB everything
indexed with B is collected on the right
and everything indexed with C is collected
on the left
A special case is when V W Nevertheless
B and C may not be the same bases
So a second special case when V W and
B C
1123
Example Let fi 1122 be a linear mapping
such that
tiny p [Link]
We would like to write the matrix associated
Ez and Ez the
to f with respect to
112 1123
canonical bases of and respectively
The matrix is given by
ft D fl D
Example Let D Pz 1B be the linear mapping
defined by Dpcx p Cx Consider the
canonical basis B I X x x3 Write the
Matrix associated with D with respect to
B the second special case discussed above
This matrix is given by
ME'B DII B Dl xDB DW B D 1 37 B
2
O B 2 B 2x B 3 B
i L
Example Let T B Dy be the linear mapping
defined by Tpcx7 xp Cx Consider the
3
bases B I Itx xtx x 1 and
C l X x x3 x4 of Pz and 1174
respectively
The matrix associated to T with respect
to B and C is
ME'B Tcl c Tata c Text x 4 c Tlx4x3 c
2
1 4 3 3
X e x1 c c 1 1 43 c
O O O O
I 1 O 0
o 1 o
O O I 1
O O O I
Associated matrix Injectivity and Surjectivity
In this section we will relate some properties
of a linear mapping with some properties
of its associated matrix
Consider the linear
mapping TV w between
to finite dimensional vector spaces Since
we can associate a matrix with T
we will use the tools developed for
matrices to study the mapping T
For this we need to fix a basis for V
B Sibi Tsz ibn
We will also need to fix a basis for W C
With these two bases we can construct
MFB the matrix associated with T such
that
TETE MFB ITB KIEV
Recall that the matrix MEB is given by
MFB CTCbD c TCb d c IT Tbh c
We have the following facts studied
previously
Studying a set of vectors is equivalent
to studying its coordinates
Since B is a linearly independent set
T is injective if and only if
T bi TC52 TI bn is a linearly
independent set
Since B is a
generating system of V
T is
surjective if and only if
Ttb T 1521 TTbn is a generating
system of W
i First we will study the relationship between
of T and the pivot columns of MFB
the injectivity
We know that T is injective if and only if
CITI Dc CtTbdIc Ttbnbc is linearly
independent But these vectors are the columns
of MF Therefore we have
T is injective if and only if the columns of
MiB are linearly independent equivalently
all the columns are pivots
B Tis injective if and only if b EMEK IIB
IE IMT has a unique solution
Clearly the number of pivots in MFB is
independent of the chosen bases B and C
B We can relate the injectivity of MEB with
B
the rank of M
rank MFB pivots columns of ME'B dim V
T
byinjectivity of T
all the columns are pivots
1123
Example Let fi 1122 be a linear mapping
such that
tiny [Link]
We have already found the associated matrix
with respect to the canonical bases
I H
L
We can that all the columns of ME
see
E
1122 21
are pivots equivalently rank ME dim
therefore f is injective
Example Let D Pz 1B be the linear mapping
defined by Dpcx p Cx As we have
previously seen the associated matrix with
respect to the standard basis B is
i
This matrix only has 3 pivots and one nonpivot
column Therefore D is not injective
Equivalently D is not injective because
rank MDB'B 3 dim 113 4
Example Let T B Dy be the linear mapping
defined by Tpcx7 xp Cx Consider the
2
bases B I Lt x X1 x2tx3 and
C l X x x3 x4 of Pz and 1174
respectively
The associated matrix is
MFB O O O O I 0 O O
I 1 O O O l O 0
o o n o
o
O O I 1 O O O I
0 O O I 0 O O O
In this case every column of MFB is a pivot
and therefore T is injective Equivalently
T is injective because
rank ME'B 4 dim 1B
Now we will study the relationship between
the surjectivity of T and the pivots of ME
We know that T is surjective if and only if
Tcb Dc Ttb273g TCb n c is a system of
generators of RM where m dim W In other
words the matrix equation
To c MEB I B Tse W
should always have a solution for any DEW
B
For this to be true each row of ME must
have a pivot in order to avoid contradictions
of the form 0 1
T is surjective if and only if the columns
of MFB constitutes a system of generators
112M
of dimW equivalently each
m row
of MFB has a pivot
BT is surjective if and only if the equation
5 c MFB ITB always has a solution for
any DEW
D We can relate the
surjectivity of T with
the rank of MFB
rank ME'B pivots rows of ME'B dim W
9
by the surjectivity
of T each row has
a pivot
1123
Example Let fi 1122 be a linear mapping
such that
tiny p [Link]
We have already found the associated matrix
with respect to the canonical bases
H
L
EZ
We can see that not every row of ME
has a pivot therefore f is not surjective
Equivalently f is not surjective because
EZ
rank ME 2 1 dim 1123 3
1125 21123
Example Consider the linear mapping T
defined by TC E AI where
a
I
L
Clearly the matrix associated to T with respect
to the canonical bases is A
ai
L
Since every row of A has a pivot we have
that T is surjective Equivalently T is
surjective since
rank A 3 1123
dim
To finish this section it must be remarked
that T is bijective if and only if every
now and every column of MFB has a pivot
For this it is necessary that MFB is square
and invertible
Moreover a mapping can be both not injective
and not surjective An example of this
is the mapping D B Ps defined by
Dpc x p Cx
Associated matrix and change of bases
Let V and W be two vector spaces and
let T V W be a fixed but arbitrary
linear mapping
Recall that the matrix associated with
T depends on the bases chosen for V and W
Nevertheless the linear mapping should not
depend on our choice of bases
This implies that there is a relation between
two matrices associated with T but with
respect to distinct bases
Indeed these relations exist and constitute the
center of attention of this section
Recall that V and W are vector spaces and
the linear mapping T V W is fixed but
arbitrary
Let's choose two distinct bases for V B and B
and two distinct bases for W C and I
Then we have the following relations between
coordinates
i B Pn I B TEV
B B
Tv c DIE WEW
Iz
BIBI CEE the corresponding
where and are
matrices of change of bases
Recall that MFB denotes the matrix associated
with T with respect to B and C and
B
ME denotes the matrix associated with
T with respect to B and E
We can conveniently summarize the relation
between these matrices with the following
diagram
wet V
C B
MT
C B
Ee T BIB
c B
B
ME
Our goal is to find a relation between
MFB and MFB To do this we proceed
as follow
Consider the matrix equation associated
with T with respect to the bases B and C
To c MFB T B Vie V
Vote that To c Ie To E an
JIB BIB I Substituting these relations
in the above equation we obtain
Ie To e ME'BBIB HB
Lastly multiplying both sides of this equation
P
by P cc E
we
E
obtain
c
To e MEB BIZ TIE
Comparing this last equation with
tote MEB I B we deduce
MFB P MEB p
E c B B
change of bases formula
for the matrix associated
with a linear mapping T
This formula is the relation between MFB and
MFB that we were looking for
Observe how the notation can help us
remember the formula above as follows
On the left hand side of the formula
5
we have M with super indeces E and Bi
On the right hand side these same indices
are on the exterior while B and C
appear on the interior Moreover notice
that the subindex B y below the superindex
C and similarly the subindex C y below
the super index C
We can also use the diagram
WE v
C B
MT
C B
Tec T BIB
c B
B
ME
to remember how to write the change of
bases formula Notice that the arrow connecting
directly B and E is labeled ME Also
notice that B and I can be connected
alternatively via B B c c
Writing from right to left the labels
of this alternative path we get MEB PBe
Since the direct path and the alternative
path connect the same two vertices
B and I we will say that they are
5
equal
i
M MIBpapers obtaining
the desired change of bases formula
1123
Example Let fi 1122 be a linear mapping
such that
tiny p [Link]
We have already found the associated matrix
with respect to the canonical bases
Now consider the following bases for
R2 and 1123
Hill D
a
Hill fill o
We will compute the matrix associated with
with respect to B and C The change
f
of basis formula is
ME'B cIe MI B
You can verify for yourself that
to L
Eeif
iii f
1125 31123
Example Consider the linear mapping T
defined by TC E AI where
a
I
L
Clearly the matrix associated to T with respect
to the canonical bases Es and Ez is ME EA
Consider the bases of 1125 and 1123
B I 0 O L O
Holl
The matrix associated with T with
respect to B and C is
given by
es
ME'B P Mfa p B
C Ez Es
You can verify on
your own that
II
L
and M
1 o