2

Linear Transformations

pefinitions. Linear Transformation

(a) Let V (F) and WV (F) be two vector spaces over the same
fleld F. Afunction T: V’WV is called a linear transformation (or
vector space homomorpbism)' from V into Wifit sattsfies the
following properties.
(1) T(u+p)=Ta+TB a, BE V.
Le. T carries the sum of two vectors of V to the sum of the
tmages of the vectors. This is known as additive property of T.
(2) T (aol=aTa, t a E Vand aE F
Le. Tcarries the scalar multiple of a vector u E V to the same
scalar multiple of the T image of that vector.
This is knoWn as homogeneous property of T.
Above shows that T preserves the vector space compositions
of addition and scalar multiplication.
Both the above properties could be combined into one as
under.
T (ca+9)=T (ca) +T (9) =cT («)+T (9)
This is known as linearity of T. We shall use this single
criteria instead of satisfying the first twoproperties given above
(6) Linear operator on V.
If in the above definition (a) the vector space Wis the same
as Vthen the linear transformation T:V+ is called a linear
operator such that
T (ca+B)=c T ()+ T (9)

Thus a linear operator is a function or mapping from Vinto

itself. Whenever it is referred that T is a linear transformation on
Vit would mean that T is a linear operator.
(c) Linear Functionals on V.
Again if Tbea mapping from vector space V (F) to its
field F
82 +T (g)
T(Cx+-B) =cT (a)
V’Fs.t.
ie. T: B E
V,eEF
t a, functional on V.
calleda linear
Then Tis Transformation Ion V
Identity
(d) (Identity operator)
L.e.
a E V
IfI(a)=a transformation on V.
identity
then Iis called
Iis linear.
a,BE VandcE F
Let which is a V. S.
then ca +BE V
3)= ca+8 by def. of I
I(ca + cI(a)+1(p)
Ia=a and /B=ß
Hence l is linear
O on V.
(e) Zero transformation
IfO (a)=0¬ V transformation on V.
then O is called zero
is linear
O transformation
so that
Let o, B E Vand cE F V. S.
ca +3 E Vas V is a
(c1+B)=0 by definition of zero transformation
O of V
=0+0as 0 is the additive identity
=c0+ 0 as in a V. S. Oc=0
c0 (a) +0 (9)
0(«)=0, O(3)=0
(operator) on
Hence O: V’Vis a linear transformation
Zero functional
IfO(a)=0e F aE V.
then O is said to be a zero functional.
Zero transformation from V (F) to W (F).
If O()=0' E W, aE V.
then O is called zero transformation from V to W.
Hence T: V>V such that T(a) -0 ¬ VuE V
then T0 i.e. zero operator
T: V’F such that T ()=0 E FV aE V
then Ta) ie. zero functional
T: V’W such that T(a)-0' E WV aE V
then T=0 L.e, zero transformation.
 LinearTransformationg
83

Note. If T: V’ Wthen we shall denote the additive iden-

titiesof 'and by 0and 0' respectively 0 will stand for zero of
ield. O will stand for zero operator, functional or transforma-
be
the case may
tion as Addition of linear transformations
Let 7; aDd T, be linear transformations on V into Whence
two linear transformations is defined as
sumof

i T7, is a mapping which carries any element a E V

in W.
tnthe sum of its 71 and T, images
To prore that T+T, is linear.
Let o. B E VandcE Fso that ca-+B E V
then (T, + T,) (Ca+)
-T, (cr +B)+ T, (ca+8) by def. of T;+T,
(c T«t T,9) +(¢T,a+ T,p) as
T, and T, are L.T.'s
=c [T,a+Ta]+ [TB+T,9]
-c [(T+T) a]+(7,+T) . by def. of T;+T,.
Hence T,+T, is linear.
(8) Scalar multiplication of a linear transformation.
W and a be a
Let T be a linear transformation on V ipto
scalar E F, then the mapping aT is definedF. as
(aT) a a [T («)], t aE V and a E scalar multiple
.e. aT carries an element ae Vto the same
of Timage of o.
To prove that aT is linear
Let o, BE Vand cE Ftben ca+ß EV.
(aT) (ea+ p)=a [T (cu+B)] by def. of aT
=a (cT (a)+TB] as T is lincar
=¢ aT (c) +aT(B)
=c(aT) a+(aT) B by def. of aT.
Hence aT is linear.
transformation.
() Additive inverse of a linear
«=-[T(«)]
Iie mapping -T defined as (-T)
T:
Btalled the additive inverse of lipear transformation
1o prove that -T is linear.
Let a, ßE Vandce Fso that ca tBE
i. (-T) (ca+ B)=-(T (ca +B)J T' is linear
- c T («) +T (8)) as
 me-T («))+[-T(9))
c-T) ut(-) Bby def. of - .
Hence -T is linear.
Cor. T(-«)-T («) VaE V
T(-)-TI(-1) ]
-(-1)7as-Ta:: T(aa)=aT (a)
T(«-B)T|(a+(- B)I
-T («)+T (-) by linearity of T
-T(a)-T (3).
() If T:V’W then zero vectors of V andW correspondo

Tcarries zero vector 0 of V to zero vector

(Meerut
0 of W.
l.e.
Let « = Vso that 7T (a) -= a' say E W L
Now a+0=« where 0 E V
then T (u) -=T (a +0)- T(¢)+-T (0)
Or a'=a'4T (0) wher c ' E W.
Above shows that T (0) is the zero vector of W.
i.e. T (0)=0'
where 0 and 0' are the additive identities of V and Wrespe:
vely.
(j) Vector space isomorphism.
If T be a mapping from V(F) to W(F) such that
(1) Tis one-one.
(2) T is onto.
(3) T preserves the vector space compositions. Ths
other words means that
T (ca+9) =cT («) +T (9) + a, BE Vand c e F
then the two vector spaces V(F) and W (F) are said to be isom
phic and expressed as V W.
Theorem 11. Vector space of linear transformations.
Prove that the set of all linear transfornmations from V (
W(F), denoted by L(V, W forms a vector space Over the
with respect to addition and scalar multip lication of
linear tra
mations as defined in (f) and (g). (Kanpur 69, Meerut 68,
We shall now verify that the set L (V, W)
axioms of a vector space. satisfies
L(V, W) is an additive abelian
Let Ti, T, T,E L (V, W) group.
then
 §3. Composite Linear operator.
We have seen io last article that if
T,: V’Wand I, : U’Vtben T,T, is a linear transformation
from U’ Wwhereas T,T, is not defined. However in the case of
linear operators i.e. when U=V= W, then both T,T, and T,I,
will be defoed but T,T, is not ia general equalto T;7; i.e. multi
plication of linear operators is not commutative in general. Let
us illustrate this by the following examples.
Ex. 1. Let T, and T, be linear operators (L. 0.) from R (R)
to R² (R) defined as follows.
T, (X1, X)(X, X) ..(1)
 Lineat Alze,
T, (X;, X)=(x, 0).
Now T,T, (, x)=T, [(T, (*;, x,)] by def. of T,I,
=T, [(x, 0)] by (2)
=(0, x}) by (l).
Again T,T, (M, x)=T, (T, (X4, x)] by def. of T,T,
=T, (x, x) by (1)
=(x, 0) by (2)
Above shows that T,T,#T,T, i.e. T; and T; are not commut,
tive.

As another example let us consider the following.

Ex. 2. Let Vbe the vector space of all
polynomial functions
in one indeterminate x with real coefficients. Let D be the differential
operator and let T be the linear operator multiplication by x.
(DT) f (*) =D[Tf (*)]=D [xf (x))
=1.f (x)+xf' (*) ...(1)
(TD] Í (x)=T(Df(1)]
--xf
Hence DT#TD.
..(2)
In fact[(DT-TD)] / (x)=f (*)=l[f (*))
DT-TD=I i.e. the identity operator.
Also prove that
(TD)²=T'D²+TD
(TD f (x)= TD ((TD) f (*)]=TD
-T (D [x f (*)}
[xf'(*)] by (2)
-T|ð' (x) +xf" (x)]
=xf' (x) +x*f" (*).
Again (T'D') f (x)= TDS (0)) .. (3)
-T*|[' (x)]=*f" (x)
Also (TD) f (x) =xf'(x) by 2.
(TD+TD) f (x)=x f (x)+ xf" (x)
-(TD³ f (x), by (3)
(TDPD+TD.
Ex. 3. In the vector
space of Ex. 2. if the
as
operator T be defineu
then DTl but TD+1.
LinearTransformationg
89

Let
f(*)=po+Px+pa.
+...e V.

d
-Po+P t: t.
=p»tPixtPt...=f (*) =1 ()=If(*)]
Since (DT)f (*)=/[S (*)] +f(x) e V.
. DT=I the identity operator.
(TD)f(*) =T(D (f(*))

=T [p+2pyx + .J

-P1x+Pt...fx)AI[f (*)J.
.. TDl.
Hence DT4TD.
$4. Properties of multiplica tion of linear operators.
Let A, B,C be three linear operators on a vector space VE) and
O
and I are the zero and identity operators on V.
Then the following algebraic results hold good.
(1) A0-0A=0
(2) Al=IA=A
(3) A (B+C)=AB+AC. Right distributive law.
(4) (A+ B) C=AC+BC. Left distributive law.
(5) A (BC)=(AB) C. Multiplication is associative.
0) k (AB) =(kA) B=A (kB) where k is a scalar.
multiplication by scalar is associative.
uet a E V then by definition I (a)=, 0 ()=0 ¬ V.
AB («)=A [B (a)].
A0(«)=A [O («)]=4 (0) =0-0 («).
.. A0=0
 Exercise 7
Ex. 1. (a) Let T: Vs (R) ’ V, (R) defined by
T (a1, ag, 4s)=(4, az), a1, ag, ag E K.
Prove that Tis a linear
transformation.
Let u=(4,, ay, aa), p=(b, b2, ba) E
T («)=(a, as) and T V; (R)
.:. (Ø)=(b, b)
catp-(ca,tbi, ca,+b,, ca,+ b,)
Linear Transformatlons 93

T(cu+B) (ca, + bg, ca, tb,) by def. of T

(ca,, ca,) 4+ (b,, b,)
c(ay, a,) -+(by, b,)
cT (a)+T (B) by I.
Hence 7T is a lincar transformation.
(b) T': V, (R) > V, (R) defined by
T (a,, ay, 4y) =(a1-ay, ay-a4).
Prove that T is lInear.
Let a=(a,, ag, a4) and B(b, b,, b¡)
T(*)-(a-a4, a,-a,), T (9)=(b,-b,, b,-- b,) (1)
(ca-+p)(ca, +b,. ca,+b, Ca, + b,)
T(ca+p) =(ca, +b,-ca,-b, ca,+b,-ca,-b)
-[c (a,-a,) +b,-b,, c (a- 4,) +b,-ba]
-[c (a,--a,), c(ai -a,)]+(6,-b, bË-b,)
=c |(a - a,), (a, - a,)]+(b,-by, bË-b,)
cT («) +T (9) by I.
Hence T is linear.
(c) T: V, (R)’, (R) defined by
T(a1, ag, a,)=(3aj -2a, +ag, aj-3u- 2a,).
Prove that T is linear.
If a=(b1, b¡, ba) then T (9)=(3b,-26,+b3, b,-3b,-2ba)
T (ca+B)=T (ca, +bj, ca, + bg, ca,+ b,).
=[(3ca, +36,--2ca,-2b,+ ca,+b,),
ca, +b,-3ca,--3b,-2ca,-26,].
=[c (3a,-2a,+ a,)+(36,-2b, +b,).
+c (a,-3a,-2a,) +(bË-3b,-2b,)]
=[c (3a,-2a, +a,), c (a,-3a,-2a,)]
+[(36,-2b, +ba), (bi-3b,-2b,)]
=cT («) +T (Ø).
Hence T is linear.
(d) T=V, (R) ’ V, (R) defined as
T (a1, a,) =(4,t+az, a). Prove that T is linear.
Let a=(4,, 4), B=(b,, b,).
T(«)=(a,+a,, a,), T (3)=(6,+b, b,)
cx+9=(ca, +b1, ca,+b,).
T (ca+9)=(ca+b,+ca,+b, ca, +b,) by def.
-(c (a,+a,) +b,+b, ca, +b,)
=(c(a,+a,), ca)+(b,+b, bj)
CT (a) +T (B). Hence T is linear.
 64

R defined as
Linear Aly,
(e) T-V, (R)
T (a,, a,)=, ,.
T(ca+9)-T (c a, t b,, c a, t h,)
-(c a, +b,) (c a, +b,)
cT(o)+ T(B)c a, a, +b, b
Thus T (c a+ S)tc T(o)4 T (9) and hence T is not line.
(0 T- i, (R) ’ V, (R) defined as .
T(a,, a,)=(a, +1, 2a,, a, +a,).
We know that if Tbe a lincar transformation from V..
the zero's of the two spaces correspond i.e. T(0)=0'
Hence T(0)= T (0, 0)-(0, 0, 0) (0,0, 0) ie, 0'
Hence T is not lincar.
Ex. 2. Let V(F)be a vector space of all mxn matrices
the field F and P and Q be two fixed matrices of order m%m and
nXn respectivelv over the same field F
T: V’V defined as T (A)=PAQ.
Show that Tis alinear transformation.
Now P is mX n and 4 is mxn PA is m n andois
nXn.

PAQ is mxn matrix and as such belongs to V (F)

Let Aand BE V(F) so that k A+B is again a matrix of
mxn type and as such belongs to V (F).
T(k A4 B) P (k A+ B) O by def.
-(k PA+ PB) Q
-(k PAQ+ PBÌ)
-kT (A) +T (B). Hence Tis iinear.
Inmatrices multiplication is distributive w.r.t. addition.
(b) Let V be the vector space of all n xn
field F and Mis an orbltrarily fixed matrices over he
matrix in V.
T:V+V: T(A)= AM + MA
Prove that T is linear.
T(k A+B)-(k A+ B)
M4 M (k A+ B)
-k AM 4BM k MA+ MB
-k (AM MA) +(BM +
kT (4)+T (B) MB)
Hence Tis linear.
(c) Let V be the
vector space of all
the fieldR which we know
are a+ ib is a vector complex numbers over
space V (R)
such that a, b E R, whose elements
 l:T (a+ ib) =0- ib E
Linearity of T. Let a=aib, 3=c+ idV(R).
chea ka+§=(k atc) +i(kb +d)
Tka+9)=(k a+c) -i(k b+d)=k (a-ib)+(c - id)
=k T(a) +T (3) and hence T is
linear.
Note. If the vectot space be V(C)
i.e. V(C)={a+ib : a, bE C}
cbeathe linearity property will not be satisfied as in this case k
lt be chOsen to be p+iq and it may be verified that
T (k a+3)# kT(a)+T (3).
Ex. 3. Let V (R) be the vector space of integrable functions
onR.

Drfine T: V>R:T()=| f) dr, fE , a, be R.

Prore that Tis a linear functional.
Tis a functional al-right as it maps elements of vector space
to its field R. We bave now to show that T is linear.
Let f, and f, e Vso that cf+f,also E Vwhere cE R
I(ef+f)= (cf ()+f, () du

=cT()+T ()
Hence T is linear.

Ex. 4. Describe explicitly the linear iransformation T from

Pto F such that T (e)=(a, b), T (e2)=(c, d) where e-(1, 0),
=(0, 1) are unit vectors.
(1, x) E F and we are required to find T (X1,
X) under
Aven conditions that
T (e,)=(a, b) and T (e,)=(e, d):
and hence
NOW we know that fe,. e.} is a basis set for F
I.c. of the elements
eu0r (X1, x,)E F' can be expressed as
of the basis set.
Clearly
(, X)=x, (1, 0)+x, (0, 1)=x, eitx ez
, x)=T(3, e,tx, e,)=x T(e)+*, (e,),
by linearity of T.
T (Zj X)=x, (a, b) +x, (C, d)
 96

- (a x,, b x)+(cx, dx,)

Linear Al
(ax,+cx, b x,+d x)
Ex. 5. Describe explicitly the line ar transformation
T: R'> R such that T (2, 3)=(4, 5) and T(1,
Just as in last qucstion we shall first establish that
vectors of the domain of T form a basis for the domain
0)=(0,0).
Tie. R'. For L.I. of (2, 3), (1, 0)
a (2, 3)+ b (1, 0)==0=-(0, 0)
(2a+b, 3a)-(0, 0)» 2a-+b=0, 3a=0 ’ q=0, h
Hence they are linearly independent.
Next we have to show that they span R.
Let (*1, x) E R'=a(2, 3) +b (1, 0)=(2a+b, 3a)
3 b= 3x,--2x,
2a+b=X, 3a==X .. a=,
3
.. (K, x)= (2,3)+ ?x-2x, (1, 0)
3
Hence they span B.
i. T(ky, x)-T(2,3)+M (1, 0) by (1)
T(2,3)-4T(1, 0)
by linearity of

Ex. 6. Find alinear

(4, 5)+
3x1-2x0, 0)=(
3

T(1, 0)=(1, 1)and T(0, 1)transformation T: RR>R² such th

square with vertices (0, 0), -(-1, 2). Prove that T maps
(1, 0), (1, 1) and (0, 1) into
parallelogram.
Proceeding as in Ex. 4
(X, x4)==x, (1,
T (x1,
x)=x 0)+X, (0, 1)
(1, 0) +x, T (0, 1)
T(X,
X,)=x, (1, 1)+x, (--1,
Let the
2)-(x,-- X, t...0)
,
and their T
given verlices of the
images be A', B', C', square
D',
be A, B, C, D respective'
A'=T (A)=T (0,
0)=(0, 0)
on putting x=0,
LmearTransformations
97

B'=T(B)=T(1, 0)=(1. 1) on putting x=1, x,=0 in I

C'-T(C)=T (1, 1)-(0, 3) on
D'=T(D)=T (0. 1)==(--1, 2) putting x=1, x,=1 in I
on putting x=0, x,=1 in
above that A' B'= C'
It is clear from D'/2 and also slope
B'slope of C' D' and A D' = B'
ofA' C=/5. Hence a paralle-
logram.
1 If U(F) and V (F) are two
vector spaces and Ti, T, are
incar transformations from U into V. prove that the mapping T
detinedas
T (a)=c T, («)+T, (a), « E U, c E F
isa Iinear transformution from U into V. (Marthwada 71)
T,:U-’ V, T,: U’ Vare linear transformations
T1 («) E V, T, (a)EVaEU
T(«)=c T, (a)+T, («) also belongs to V a E V, as
Vis a vector space.
Hence T is a mapping from U into V.
T is linear. Let a, BEU and ke Fthen
ka+8EU
:: T(k +g)=c T, (k a+B)+T, (k a-+B) by definition of T
=c (k T; («)+T, (B)'+k T, («)+T, ()
as T, and T, are 1linear transformations
ck T, («)+c T; (g)+k T, (a) +T, (P)
by Distributive Law in V
=k [c T; («)+T, («))+c T1 () +T, (8)
ckake in a field
=k (c TË («)+T, («)]+c T; (B)+T, (B)
a (a+p)=a +aB in a V, S.
kT («)+T (g) by def. of T
Since T (k «+ B)=k T («)+T (B) therefore T is linear trans
formation.
97. Range and Null space of a linear
transformation.
Definition. Let TV (F)+ W(F) (Marthwada 71)
Oniess T is an isomorphism in which case T in one one onto
otherwise T may be many one and into
When T is many one then there maymapplng.
besame many vectors
so element in
domai
In
n V (F) whose image under Tis the in W(F).
particularthere may be aset of vectors in the domain V(F)
whose TImages in W(F) is the zero vector of W(F)-
 98

Similarlv when Tis not onto then all the

Line
elements
ar Alzu
are not the Timages of thec elements of the domain set of W
this case the range of T will not be the whole of
in
W(F).
Hence we define below range of 1 and null space of 7
Range ofTie R(T).
If T: Wthen range of Twill consist of all those
of Wwhich are T imoges of certain vectors x of the domain
other words.
yeclar
R(T)-(ye W:y=T (N),xE V}
It is evident that R (T) is a sub sef of W.
Null space of T(or kernel ofa transformation)
rT: V W then the null space of T will consist of all th0se ,
tors x of the domain V which are ma, ped on to zero of W or who
T image is the zero of W. In other words.
Null space of T-{xEV:T (x)=0' W}
It is evident that null space of T is a sub set of domain V.
Theorem 71. T:V’ Wwhere T is a linear transfor mati
Prove that range and null space of Tare sub spaces of W and
respectively.
Criteria for a sub space
If Ube a sub set of a vector space V then Uwill be a sub spa
of V if whenever a. B E U then ca +8 also belongs to U whe
cE F
Range of T={y:y=T (x) where x E n
Let , and y, E range of T so that
V}=T(x;) and y,=T (x}), where x,, x, E V
Now cy,y,=cT (x) +T (x,)
-T (cx) + T(x)
-T (ex+ x) by linearity of T.
But V is a vector space and
X1, x,EV» cxtx,E V.
Hence (cy);+ y)=T (cxit x,) where
Therefore cv, ty, also belongs to R(T)cx+x; E V.
when y, and y, E N
which therefore is sub space of W.
Null space ofTx:T()-0}
Let x1 and x, E null space of T So that
T (x,)=0, T (x) =0
 LincarTransformatlons 99

also E V, which is a vector space

Now cx tx,
T(cx, X) -cT (x,) +T (x,) by 1inearity ofT
-c0 +0'0

+x, E V'also belongs to null space of T when ever

Hence cx,(T).
lo N
I, , E space of 7 is a sub space of V
Hence null
Theorem 72. Define rank T and nulliry T where T: V’ Wis
dimensio
linear transformation and hence p:ove that ifV be finite
V=n then
nal sa dim
Rank T +Nullity T=dim V=n
(Benaras 1969; Meerut 1969, 72, 73, 74, 75)
Nullity T. 7he dimension of the null space of a linear trans
formation Tis called nullity Tand is written as n (T).
Rank T The dimension of the range space of a linear trans
formation T is called rank T and is written as p (T).
Now we have to prove that
Rank T4 Nullity T=dim V=n.
Since is finite dimensional and null space of T i.e. N (T) is
asu sp ce of V therefore it is also firite dimensional and say
dim N(T)=k where k<n.
Hence the null space of T has a basis set say
A-{,, a......}
consisting of k elements. By definition of null space we have
Ta, =0, Ta,-0'...Ta; -0'
The above basis set A can be extended. (Th. 12. P. 54)
so that the extended set B{a,, a.g.k, k+1 an
Consisting of nelements is a basis set of V which is of dim n.
We shall now establish that the set
Ca{Ta741-.... Tan}
of additional n- k vectors is a basis for
Consi-ting of Timages
Tange of T ie.
R(T) and in that case
dim R (7T) =n -k and dim N (T)=k
dim R (T)+ dim N (T)=n -k-+k==n-dim V.
Now in order to prove that set Cis a busis for R(T) we have
o prove the
tollowing
() The set C is L.
(2) I.
The set C spans the range of T.
100 Linear Algeb,
Linear independence of C.
Consider Ck+1 Tek+t...+C, T¡n0'
dn]=0
’ T(Ck+1 ak41t...+Tcn
by linearity of
Above shows that the vector
d=Ck+1 k41tCn On belongs to null space of T as its Tim.
is zero of W.
But the set A={d,, g..ag} is a basis set of null space and a
such this vector a Enull space can be expressed as a 1.c. of th:
elements of A

Hence b, ayb, agt+...+bu a=Ck+1 Ch+1t.Cn an

or
or A linear combination of L. I. set B=0
Therefore b, b..by=Ck+1=..=C1=0
Hence cn+1 Tan41t... Cr T a, '.
’ Cn+1=0...Cn0,
Therefore the set C={T n41 ..T an
consisting of n-k vectors is L, I.
2. The set C spans the range of T.
Let B E R (T) then by definition
BT(a) where E V

or BT2 a ay as set B is a basis for V

or B=2 a, T («) by linearity of T

But Ta-Ta,...Ta: =0' as A is a basis of null space

B a, T (a)
f-k+1

Hence Bis a l. c. of T (as+)...T («,)

which are elements of C.
dim R (T)-1-k, and dim null space of T=k.
Rank T+nullity T=n-k+kedim v.
Linear Transformatlons 101
Exercise &

1. Les be a vector space aid Ta linear transformation from

V' into V. Prove that the following two statements about T are
equivdlent.
i) The intersection of range of T and the null space of T is
V.
the zero space of (Agra 80)
(6) If T (Ta)0 then T (e)=0
We shall prove that
a ’ band b ’ a
a b given R (T) n N(T)={0}
To prove that
T (Ta)-0 > T («)=0.
Let us put Tu=ß
by definition of R(T), BE R (T) ...(1)
Also we are given that
T(Ta)=0 or T ()=0 .. BE N(T). ...(2)
Hence Bbelongs to both R (T) and N (T).
.. BE R (T) ^ N(T) B0 i.e. T («)-0
as R (T) n N (T)=={0} given
b> a given T (T«)0> T («)=0 ..(3)
To prove. R (T) n N (T)={0}
Let ß E R (T)) ^ N (T)
.. BER (T) and BE N(T)
Now B ER (T) + BT(a) for some #E V
and ßE N (T) > TB) 0> T(Ta)=0 ’ T («) =0 by (3)
’ B=0 as BT(«).
Hence R (T) ^ N(T)={0}.
2. Let Vbe a finite dimensional vector space and T a linear
operator on V such that rank T=rank T. Prove that the range and
null space of T are disjotnt.
Since T is a L. O. on V hence T² is also a L. 0. on V.
Also dim V-Rank T+ null1ty T=Rank T+nullity T
Nullity T=Nullity T² by given condition
or dim null space of T=dim null space of T
Now « E null space of T ’ T («)=0 .(1)
T [T («)]=T0) ’ T' («)-0
¬ null sp ace of T:
Dull space of T C null space of T? ...(2)
 102

Since by () dim of null space of

|inear Aye
dim null Npace ol
sub sel of he olher nd
as such one canot be
have null spacc of Cnull spaco of 7 lience we
so that (2) and (3) logcther give
null space of Tnull space of 7".
Now from (3) it follows that
T (a)(0 > T («) )
and hence as proved above in 2nd part of Bx. I we have
R(7) N N(7){0},
Ex. 3. Let Vbe n dimensional vector space over
and let 1 be alinear Iransfornation from Vino Vsuch thut
the field p
cnd mull space of Tare identical. Prove that n is even.
Rank (T) Nullity (T) dim Va given
Also R(r) and N(7) are given to be identical
and hence dim R() dim N (T)
or Rank T= Nullity T
2 Rank (7)=, and hence is even
As an example ol such a linear
transformation for which
R(I) and N (T) are identical we have below.
Let Vbe the vector space of all
degree less than or cqual to I together wilh polynomial functions of
zero function over R.
Let D:V - Vbc the differential
function to its derivative.
operator which takes every
Now we know that
always zero.
differentiation of constant functions is

N (D)=u: D
(u)0)=set of all cons tant functions.
R (D =B:B-D (a)}=set of all
because a' are functions of const ant functions.
D(a) i.e. B will all be
degreo less than or equal to l and hence
constant functions.
Another example.
Let T: V, (R) -’ V, (R)
delined as
T (4, b)= (b, 0) v a, b E R
...(1)
It is easy to verify that T is linear.
N(T)=(a:T («)=0}
T()=T (u, b)==(b, 0) -)=(0, 0) ’ b0
=(a, 0)
N (T)={(a, 0) : T (a,
0)=(0, 0)} ...2)
uE R
t Linear Transformations
103
R(T)={8:BT(«)}.
BE R(T)-T(e)
-T (a, b)=(b, 0)
Now (a, bh) = a (1. 0) +b (0, 1)

.: 5-T(«)=T (a. b)
=T (as, t ba,)=aTa1 +bTa.
=a T(l, 0) 4+b T (0, 1) by linearity of T
=a (0. 0) +b (|, 0) from (1)
on putting appropriate values of a and b.
Hence R (7) is spaned by vectors (0, 0) and (1, 0).
We can exclude the zero vector
and can say that R (7) is
spanned by (1, 0)
R(T)={a (1, 0} : a e R}={(a, 0) : a E R} ...(3)
From 2 and 3 we get N (T) =R (T).
Clearly V; (R) is vector space which is finite dimensional
Whose dimension is 2
as il, 0), (0, 1)} is a basis for V, (R).
Ex. 4. Veriy that the transformations defined below are linear
ransformations from R into ' . Find the range, rank, null space
and nullity for each

() T (X1 X))=(j-X¡, Xy-X1 -)

Linearity of T. cz+B=c (x, x)+U }:)
=(cx}+)' CXt')
.. T (ca +B)=(cxtJitcxtJ'y CX1+1- CX_- )2s CXt )
=c (x1tx2)+U 'a, c (x-x)
+1-), cx,+y)}
c1 (a)+I (B).
Hence T is lnear.
Null space of I. Let a=(x, *) then
N(T)=(E R:I (2) ==0 E R}
But T a)=(x,- X+X,
X)=0'=(0, 0, 0)
X-4=0, x+x=, x=0
X0, x,=0
.2=(x; Xj)=(0, 0) =0 E R'
 ar A
104
Hence null space consists of only zero vector of the dota

T= dimension of null space =0.

Nullity R: B=T (3) for Some aE
T={BE
rangeofspace of T will consist of all tripiets oi,
Hence space
Range

such that (x,, X) E R?

N, N, Now in order to find Rank T. we should find a bas
R(I), and the number of elements in the basis set will be rank

BEER (T)=T (*)=T (X1, X,)

0)+x, (0, 1)
Now (x1, x) E V, (R)= %4. (1,
(I4tXy)=xTatx,Ta
.:. BraT(X X,)=T by linearitye
=X, T(1, 0)+ x, T (0, 1)
=X (1, 1, 0) +x, (1, -1, 1)
values
by (1)on putting appropriate
i.e. xl, X=0 and then x=0, x,=1.
Since BER(T)=x1 (1, 1, 0) + x, (1, -1, 1)
therefore the vectors (1, 1, 0) and (1, -1, 1) span R (T)
Let us see if they are linearly independent.
Consider a (1, 1, 0)+b (1, -1, l)=0'--(0, 0, 0)
> (a+b, a-b, b)=(0, 0, 0) ’ a=be0
Hence L. I. and as such the set
{(1, 1, 0), (1, -1, 1)}
is a basis set of the R (T) consisting of two vectors and therefor
dim R (T)= Rank T=2.
Alternative. We had found nullity T=0
and dim V=dim R=2,
Also Rank T+ Nullity T=dim V=2.
Rank T+0=2 or Rank T-2.
(b) Proceed as above.
Null space of T={0}
Nullity T=0, Rank T=2.
Also the set {(1, -1, -1), (1, 1, 0)}
is a basis set for R (T).
Ex. 5. Let T, be a
linear
that set of all linear transformationstransformation
S on V, (F) onfor which1
is a sub space uf the vector
of all linear transformations.
We know that the set space
of all linear transformations ona
space forms a vector space say A
 Linear Transformatfons
105

A-T:T is a L. T. on I, (F))
. R he the set of all such L. T's. S s.. T,S-0 where T,

B--S: Sis a L. T. for which T,S-0, I,e A

We have to prove that B is a sub space of 4. Cleariy B is a
suh set of A
Let S, and S, e B so that T,S,-0, T,S,-0.
In order to prOve that B is a sub space we should that
CS+ S, where CE Falso belongs to B whenever
S and S, E B.
Now T, (CS,+S,)-c (T,Si)+ (T,S,) by Linearity of T,
C04 0-0
Above shows that CS, +S, also belongs to Band hence B is a
sub space.
Ex 6. What do you wnderstand by invariant sub space and
hence prove that range and null space of a linear transformatlon T on
Vare invariant under !.
Let T: V’V, and W be a sub spacc of V
Then W is said to be invariant under T if for all
aE Wthe vector T a E W
E W>T («) E W
Itshould be clearly understood that the above statement does
not mean that if we have a vector B E W then B some T («)
with e E Wie. it does not mean that for every vector E Wis the
Timage of some vector belonging to W.
R()B:B=T (a) for some a E V}
We know that R (T) is a sub space of VasT:V-+
.. BE R(T) > BEV
and BEV> T(9) E R (T) by def. of R(T)
Hence BE R (T) T(B) E R(T)
.. R (T) is invariant under T.
N (T)=(a:T («) =0, a E V}
We know that N (T) is a sub space of Vas T: V+Vand as
Juch it contains the zero vector.
E N(T) > T(a)=-0 and 0 E N(T) which is a sub
space.
«E N (T) ’ T (a) E N(T)
N (T)is invariant under T.
106
a iinear
Linear Algebr,
f T: V (R)’ V, (R) is
Ex. 7. (a)
defined by
T(4, b, c, d) -(d-b+c+d, a+2c-d,
a-+b+3c-3d) transjormati
for a. b. c, de R, then verify thau
Rank I +nullity T=din V, (R)
We know that the set A=e, eg, eg, ea is a basis
set for
Va(R)
where e=(0, 0, 1, 0) etc.
We shall by the given deinnition of Tfind the T
bas1s.
ima zes of
Te,-T (1. 0, 0, 0) =(1, 1, I) Putting a=1, b=(=del
Te,=T(0, 1, 0, 0)=(--1,0, 1) b=l, ac=d=)
Tes=T (0, 0, 1, 0)=(1, 2, 3) c=1,a=bd=)
Te=7 (0, 0, 0, 1)=(l, -1, 3)
d=l, a=bmç=
Now any vector « E V, (R)=aejt age, t
as A is a basis set. ageatuae,
BER(T)=T (3)=T (4,e,+ ajest ase, tagea)
=a, T(e)+a,T (e,)+4,T
(es)ta,T (e)
by linearity of T
=a, (1, 1, 1)+a, (1,0, 1)+ a, (1, 2, 3)
Now we shall verify that B ta, (1 -, -3)
vectors E V; (R) can be ER(T) expressed as l.c. of four
vectors or not. expressed as al. c. offewer number of
For
these four vectors, tbis we compute a
matrix, whose roWS are
1
B= 0
2 3
1 -1 -3
we sball reduce this matrix to
echelon form
Apply R,tR, Ry--R,, R--R,
Bo 2

0 -2 14
 Linear Transformations
107

Again apply R, 2R, Rg R,

2
) 0

Then thc non zcro vectors in the above

ic. (1. I, ). (0, I, 2)}
will constitute generating set for R (T).
Verificat ion. « E V, (R)= (4, b, c, d)
8E R(I)= I(a) =(a -b+c+d, a+ 2c--d, a+b+3c-3d)
-(a-b+cd) (1, 1, 1)+(6-+c-2d) (0, 1, 2)
-=k, (1, 1, I)+k, (0, 1,2).
Also the above set is L. I. and hence
(1, 1, 1), (0, 1. 2)} is a basis set for R (T) and since it cont
ains (wo elements therefore
dim R (T) i.c. rank T=2.
...(1)
Basis for null space of T
z¬ N (T) if T(«) =0
T(a, b, c, d)-=(a -b+ctd, a+ 2c -d, a t-b+ic -3d) =(o, 0,0)
a+b4ctd-0 We have to solve these fora, b, c, d.
a+b+3c -- 3d-0
Coefficient matrix
1 -1 i
0 2
I 3 2 2 -4
by R,-R R¡-RË
by R,--2R,
-2
0 0
Above is echelon form of matrix.
The equivalent sys te m of equations are
a-b+c+d-0 b=2d-c
b+c-2d=0 a=-2c+d.
0-0
The number of free variables is 2 ramely c, d and the values
ol a, b depend on these and hence nullity T dim N (7T)=number
of iree variables=2.
108 Linear Alge,
a=-2, b-1
Choosing c=1, d=0,
b=2.
Choosing c=0, d=1, a=1,
2, 0, 1}
.:. ((-2,-1, 1, 0), (0, above system is L.I. also
constitute a basis for N (7) as the
Dim V, (R) =4, Dim (R(T)]=2, Dim [N (T)
dim V (R)
.. Rank T+Nulity T=
ormation definod:
Ex, 7 (b). Let T: R-’R' be a linear transj
T(x,y, z)- (x+y, y tz).
null space of
Find a basis, dimension of each of the range and
and verify the theorem
Rank T+nullity T=dim V.
basis for R then
From given definition if fe, e, e} be a
T(e)-(1, 0), T(e,)=(1, 1), T(e,)=(0, 1) T images of
Now form the matrix whose rows are the above

1 0
i.e. 0 1
01 0 0
by R,- R, and then R-R,
i.
Above is echelon form and hence the non zero vectors e
{(1, 0), (0, 1)}
will constitute a basis for range of T and hence dim [R (T)]=2
i. e. Rank T=2
Basis for null space of T.
aE N (T) if T(3)=0
I (4, y, z)=(x+}, y+z)=(0, 0)
*ty=0
y+z=0.
Clearly there is only one free variable z
i.e. y=-z, X=-}=z.
Choosing z=1, x=1, y=-1 and hence (l,-1, 1} is a basls
set for N (T).
Chearly T (1, - 1, 1)=(0, 0)
.. dim (N (T)]=Nullity T=1.
Also dim R°=3.
Rank T+ Nulity T=2+1-3=dim R
Linear Transformations 109

Ex. 8. (a) Eind alinear transformation T: R ’ R4 whose

vectors
range is spanned by
(1, 2, 0,--4), (2, 0, --1, --3)
The basis ef R is A={e, eg e}
Now range is spanned by ((I, 2, 0,--4), (2, 0,-1,3)}
We may further include a vector (0, 0, 0, 0) in the above and
.his inclusion does not efect the spanning property so that
B-{(1. 2, 0,-4), (2, 0,-1,-3), (0, 0., 0, 0)}
Now there exists a unique L. T. T such that Tx=y; by Th.
22 P. 78.
. Te,-(1, 2, 0,- 4), Te,(2, 0,-1,-3). Te,-(0, 0, 0,0)
Now «E R° ’ ¢=(a, b, c)= ae,+be,+ce,
T (a, b, c)=T (a e,t+b e,tc e,)
aTe, +b Te, +cTe,
or T(a, b. c)=a (1, 2, 0, --4)+5 (2, 0,-1,-3)+c (0, 0, 0, 0)
or T (a, b, c) (a+25, 2a.-b,-4a--36)
Above expresses explicitlyT: R'’ R
Ex. 8. (b) Descrtbe expllcttly a linear transformation T:
R ’ R' whose range space is spanned by
(1, 0,--1), (1, 2, 2)}
Proceeding as above
T (a, b, c)=(a+b, 2b, --a +25)
§8. Singular and Non Singular Transformations.
Definition. A linear transformation T: V’ W is satd to be
non slngular if the null space of T is the zero space {0} i.e. it consists
of only the zero element.
Now N{T}=(« E V: T(«)=0}.
There could be so many a'E Vwhose T image is zero of W
but if only =0 is such that T («)--0' and for no other 0,
T(3)=0' then T is said to be non singular.
In other words ifT («) -0' > «=0 then T is non singular.
We can also say a0, T(«)0' then T is non singular. Twill be
said to be singular if the null space of T' contains at least onc non
Zero Vector i.e. a0 > T («)=0'.
$9. Injective (one-one) surjective (onto), bijective (one-one
onto) Transformations.
I. T:V ’ Wis Said to be suriection if it is onto ie. each
clement yE codomain Wis the T image of some element xE