Thanks to visit codestin.com
Credit goes to www.scribd.com

0% found this document useful (0 votes)
62 views35 pages

Unit 5 Notes

Uploaded by

kavipriya28400
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
62 views35 pages

Unit 5 Notes

Uploaded by

kavipriya28400
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 35

UNIT V - Trace and transpose -

Hermitian, unitary, normal


transformations, real quadratic
form
Vector space
• Definition : A nonempty set V is said to be a vector
space over a field F if
• 1) V is an abelian group with respect to addition
• 2) For 𝛼 ∈ F and v ∈ V , there is an element 𝛼v ∈ V F
satisfies the following conditions. 𝛼
• (i) 𝛼(u + v) = 𝛼(u) + 𝛼(v) V
• (ii) (𝛼+ 𝛽)v = 𝛼v + 𝛽v v 𝛽
• (iii) 𝛼(𝛽 v) = (𝛼𝛽)v, 𝛼v
u

• (iv) 1.v = v
• where 𝛼 , 𝛽 ∈ F and u, v ∈ V .
Inner Product space
• Definition : Let V be a vector space over a complex field F.
• It is said to be an inner product space if for any two vectors u,
v in V, an element (u,v), called inner product is defined in F
such that,
• (i) (u , v) = (v, u )
• (ii) (u, u ) ≥ 0, and (u,u) = 0 if and only if u = 0.
• iii) (𝛼u +𝛽v, w) = 𝛼(u,w) +𝛽(v,w) F

• For any 𝛼 , 𝛽 ∈ F and u, v, w ∈ V. V


u
(u,v)
v
Matrices : Trace
• Let A = (aij) be an n × n matrix.

• a11 a12 a13 . . . . . . . . a1n


• a21 a22 a23 . . . . . . . . a2n
• a31 a32 a33 . . . . . . . . a3n
•A= .. . . . . . . . . . . .. . .
• .. .. .. ..........
• .. . . . . . . . . . . . .. . .
• an1 an2 an3 . . . . . . . . ann
• The trace of A is the sum of elements on the main
diagonal of A.
We denote the trace of A as tr(A).
• Thus, tr(A) = a11 + a22 + a33 + ........ + ann.
n
That is, tr(A) = a
i =1
ii

• Lemma : For A,B ∈ Fn and λ ∈ F


• 1) tr(λ A) = λ tr(A)
• 2) tr(A + B) = tr(A) + tr(B)
• 3) tr(AB) = tr(BA)
Proof
• 1) Let A = (aij) . Then λ A = λ(aij)

n

⇒ tr(λ A) =  a = λ  a
n

i =1
ii ii = λ tr(A).
i =1

• That is, tr(λ A) = λ tr(A)


• 2) Let A = (aij) and B = (bij).
Then A + B = (cij), where cij = aij + bij .
n n n
• Now tr(A + B) =  (aii + bii) =  aii +  bii
i =1 i =1 i =1

= tr(A) + tr(B)
• Thus, tr(A + B) = tr(A) + tr(B) .
• 3) Let A = (aij) and B = (bij). n

Then AB = (cij), where cij =  a k =1


ik bkj and BA = (dij),
• where dij =  b a
n

ik kj
k =1

• tr(AB) =  cii =  (  a b )
n n

ik ki
i =1 i =1 k =1

• If we interchange the order of summation,


n n

• cii = (   a b ) = (   b a )
n n

ik ki ki ik
k =1 i =1 k =1 i =1
n

• =  dkk = tr(BA).
k =1
• Fact: If A is invertible, then tr(ACA−1) = tr(C).
• Proof: Let B = CA−1. Then, tr(ACA−1) = tr(AB) = tr(BA)
= tr(CA−1A) = tr(C).
• Definition : If T ∈ A(V ), then trace of T is the trace of
m1(T), where m1(T) is the matrix of T in some basis of
V.
• Result : Let A ∈ Fn and suppose that K is the splitting
field of the minimal polynomial for A over F, then an
invertible matrix C ∈ Kn can be found so that CAC−1 is
in Jordan form.
Lemma : If T ∈ A(V ) then tr(T) is the sum of the
characteristic roots of T.
• Proof: Assume that T is a matrix in Fn.
• If K is the splitting field for the minimal polynomial for
T over F, then in Kn, T can be brought to its Jordan
form J.
• That is, J is a matrix whose diagonal elements are the
characteristic roots of T, and each root appears as
often as its multiplicity.
• Thus, tr(J)= sum of the characteristic roots of T.
• Since J is of the form ATA−1, we have, tr(J) = tr(T).
Hence the lemma.
• Definition : If A = (𝛼ij) ∈ Fn then the transpose of A,
written as A′ is the matrix A′ = (𝛾ij) , where 𝛾ij =𝛼ji, for
each i, j.
• Lemma : For all A,B ∈ Fn
• (i). (A′)′ = A
• (ii). (A + B)′ = A′ + B′
• (iii). (AB)′ = B′A′ .
• Proof: (i). Let A = (𝛼ij) and A′ = (𝛾ij) where
𝛾ij =𝛼ji, for each i, j.
• Therefore, (A′)′ = (𝛿ij) where, (𝛿ij) = (𝛾ji) = (𝛼ij ) = A
• (ii). Let A = (𝛼ij) and B = (𝛽ij).
• Now A + B = (λij) where λij = 𝛼 ij + 𝛽ij .
• Now ( A + B)′ = (𝛿ij) where 𝛿ij = λji = 𝛼ji + 𝛽ji.
• But A′ + B′ = (𝛼ji)+ ( 𝛽ji)= (𝛿ij) = (A + B)′.
• Thus, (A + B)′ = A′ + B′
Let A = (𝛼ij) and B = (𝛽ij).
• Now AB = (λij) where λij = ∑k 𝛼ik 𝛽kj .
• By definition, (AB)′ = (μij) where
μij = λji = ∑k 𝛼jk 𝛽ki .
• On the other hand, A′ = (𝛾ij ) where 𝛾ij = 𝛼ji and B′
= (𝜉ij ), where 𝜉ij = 𝛽ji.
• Hence the (i,j)th element of B′A′ is
∑k 𝜉jk 𝛾ki = ∑k 𝛽kj 𝛼jk = ∑k 𝛼jk 𝛽kj = μij .
• Hence (AB)′ = B′A′ .
• Definition : The matrix A is said to be symmetric
matrix if A′ = A.
• Definition: The matrix A is said to be skew symmetric
matrix if A′ = −A.
• Definition: The matrix A is said to be Hermitian
matrix if A∗ = A.
• Definition: The matrix A is said to be skew Hermitian
matrix if A∗ = −A.
Unitary, Hermitian and Normal transformations

• Definition: A linear transformation T ∈ A(V ) is said to


be unitary iff (uT, vT) = (u, v), ∀ u, v ∈ V .
• Lemma : If T ∈ A(V ) is such that (vT, v) = 0
for every v ∈ V , then T = 0.
• Proof: It is given that (vT, v) = 0 for every v ∈ V . Let
u,w ∈ V . Then
• ((u + w)T, u + w) = 0 ⇒ (uT + wT, u + w) = 0
• ⇒ (uT + wT, u) + (uT + wT,w) = 0
• ⇒ (uT, u) + (wT, u) + (uT,w) + (wT,w) = 0
Given: T ∈ A(V ) is such that (vT, v) = 0
• But (uT, u) = 0, (wT, w) = 0. Therefore,
• (wT, u) + (uT,w) = 0 .........(1).
• Equation (1) holds for arbitrary w ∈ V and therefore it
must hold if we replace w by iw. Hence
• (iwT, u) + (uT, iw) = 0 ⇒ i(wT, u) − i(uT,w) = 0
• ⇒ (wT, u) − (uT,w) = 0 ⇒ (wT, u) = (uT,w) .........(2).
• Substituting (2) in (1),
• we have, 2(wT, u) = 0, ∀u ∈ V .
This implies, wT = 0 ⇒ T = 0.
Definition : An element T ∈ A(V ) is said to be
nilpotent if Tm = 0 for some positive integer m.
• Results :
• i) A non zero matrix is nilpotent iff all its
characteristic roots are equal to zero.
• ii) If V is finite dimensional over F, then T ∈ A(V ) is
invertible iff the constant term of the minimal
polynomial for T is not zero.
• iii) If T ∈ A(V ) has all its characteristic roots in F, then
a basis of V can be found in which the matrix of T is
triangular.
Lemma : If F is a field of characteristic zero and if
T ∈ AF(V ) is such that tr(Ti) = 0 for all i ≥ 1, then T
is nilpotent.
• Proof: It is given that T ∈ AF(V ).
• Let the minimal polynomial for T be
p(x) = xm + 𝛼 1xm−1 + 𝛼 2xm−2 + 𝛼 3xm−3 + ....... + 𝛼m.
• ⇒ Tm + 𝛼1Tm−1 + 𝛼2Tm−2 + 𝛼3Tm−3 + ....... + 𝛼mI = 0.
• Taking trace on both sides,
tr(Tm) + 𝛼1tr(Tm−1) + 𝛼2tr(Tm−2) + 𝛼3tr(Tm−3) + ....... +
𝛼mtr(I) = 0.
• i.e, 𝛼mtr(I) = 0.
• Since tr(I) ≠ 0, we have 𝛼m = 0.
• That is , the constant term of the minimal polynomial
for T is zero.
• Therefore by result (ii), T is singular.
• Let K be an extension of F. Let us consider T as a
matrix in Fn. This implies, T ∈ Kn. More over all the
characteristic roots are in Kn.
• Hence by result (iii), we can bring T in to a triangular
form.
Hence by result (1), we have T is Nilpotent.
Lemma : If F is of characteristic 0 and if S and T in
AF(V ) are such that ST − TS commutes with S, then
ST − TS is nilpotent.
• Proof: To prove this lemma, let compute the
following result for any k ≥ 1.
• (ST − TS)k = (ST − TS)k−1 (ST − TS)
• = (ST − TS)k−1 ST − (ST − TS)k−1 TS
• ie, (ST − TS)k = (ST − TS)k−1ST − (ST − TS)k−1 TS
..........(1)
• But it is given that ST − TS commutes with S. That is,
(ST − TS)S = S(ST − TS) .
• (ST − TS)2 S = (ST − TS) (ST − TS)S
• = (ST − TS) S (ST − TS)
• = S(ST − TS)(ST − TS)
• = S(ST − TS)2
• i.e, (ST − TS)2S = S(ST − TS)2
• In general, (ST − TS)k−1S = S(ST − TS)k−1
...........(2)
• From (1), (ST − TS)k = (ST − TS)k−1ST − (ST − TS)k−1TS
• = [(ST − TS)k−1S]T − [(ST − TS)k−1T]S
• = S[(ST − TS)k−1]T − [(ST − TS)k−1T]S
• i.e, (ST − TS)k = S[(ST − TS)k−1]T − [(ST − TS)k−1T]S
......(3)
• Put (ST − TS)k−1T = B.
• Substituting this in (3), we have, (ST − TS)k = SB − BS.
• tr(ST − TS)k = tr(SB − BS) = tr(SB) − tr(BS) = 0.
• Therefore, tr(ST − TS)k = 0 ⇒ (ST − TS) is nilpotent.
Definition
• Let V be an inner product space over a field F. A basis
S = {v1, v2, ......, vn} of V is said to be an ortho normal
basis if
• i) (vi, vj) = 0 if i ≠ j
• ii) (vi, vj) = 0 if i = j
Theorem : A linear transformation T on V is unitary iff it takes an
orthonormal basis of V in to another orthonormal basis.

• Proof: ( Necessary Part )


• Let T ∈ A(V ) is unitary. Let {v1, v2, ......, vn} be an
orthonormal basis of V .
• Claim : {v1T, v2T, ......, vnT} is an orthonormal basis of
V
• Now (viT, vjT) = (vi, vj) = 0 ( since T is unitary and i ≠ j )
and (viT, vjT) = (vi, vj) = 1 ( since T is unitary and i = j )
• That is, (viT, vjT) = 0 for i ≠ j
and (viT, vjT) = 1 for i = j
• Hence {v1T, v2T, ......, vnT} is an orthonormal basis of V
Sufficient part : Let {v1, v2, ......, vn} and {v1T, v2T, ......,
vnT} are orthonormal bases of V .

• Claim : T is unitary.
• Let u, w ∈ V . Since {v1, v2, ......, vn} is a basis for V,
every element in V can be expressed as a linear
combination of these basis elements.
• u = 𝛼1v1 + 𝛼2v2 + 𝛼3v3 + ....... + 𝛼nvn
• w = 𝛽1v1 + 𝛽2v2 + 𝛽3v3 + ....... + 𝛽nvn
• That is, u = ∑i 𝛼ivi and w = ∑i 𝛽ivi
• Then , (u, w) = (∑i 𝛼ivi , ∑i 𝛽ivi )
• (u, w) = ∑i 𝛼i (vi i,vi ) = ∑i 𝛼i ......
i
(1)
(u, w) = ∑i 𝛼i  ( vi,vi ) = ∑i 𝛼i
i i ...... (1)

• Also,
• uT = ∑i 𝛼iviT and wT = ∑i 𝛽i viT
• Now consider,
• (uT, wT) = (∑i 𝛼iviT , ∑i 𝛽i vi T )
• = ∑i 𝛼i  i (viT , viT)
• = ∑i 𝛼i  i ………(2)
• From equations (1) and (2), we have,
(uT,wT) = (u,w) for all u,w ∈ V .
• Hence T is unitary.
Lemma : T ∈ A(V ) is unitary iff TT∗ = I

• Proof: Suppose T is unitary.


• Then (uT, wT) = (u,w) for all u,w ∈ V .
• Now consider, (u, wTT∗) = (uT,wT) = (u,w) = (u,wI).
• That is, (u, wTT∗) = (u, wI). This implies, TT∗ = I.
• Conversely suppose, let TT∗ = I.
• Then, (u,w) = (u,wI) = (u,wTT∗) = (uT,wT).
• That is, (u,w) = (uT,wT) for all u,w ∈ V .
• Hence T is unitary.
Theorem: If { v1, v2, ....., vn} is an orthonormal basis of V and if the
matrix of T is (𝛼ij) then the matrix of T∗ is (𝛽ij) where
(𝛼ij) is the conjugate transpose of (𝛽ij). That is, 𝛽ij =  ji
• Proof: Let viT = 𝛼i1v1 + 𝛼i2v2 + ....+ 𝛼ijvj +......+ 𝛼invn and let
viT∗ = 𝛽i1v1 + 𝛽i2v2 + ....+ 𝛽ijvj + ..... + 𝛽invn for i = 1,2,...., n.
• Consider, (viT∗, vj)= (𝛽i1v1 + 𝛽i2v2 + 𝛽ijvj + .... + ......+ 𝛽invn , vj)
• = (𝛽i1v1 , vj) + (𝛽i2v2 , vj) + ......+ (𝛽ijvj , vj) + ......+ (𝛽invn , vj)
• = 𝛽i1(v1 , vj) + 𝛽i2 (v2 , vj) + ......+ 𝛽ij(vj ,vj) + ......+ 𝛽in (vn , vj)
• = 0 + 0 + ....... + 𝛽ij(vj ,vj) + 0 + .... + 0 = 𝛽ij .1 = 𝛽ij.
• That is, (viT∗, vj) = 𝛽ij .
• There fore, 𝛽ij = (viT∗, vj) = (vi, vjT).
• That is, (vi, vjT) = 𝛽ij ...........(1)
• Now, (vi,vjT) = (vi, 𝛼j1v1 + 𝛼j2v2 +....+ 𝛼jivi +....+ 𝛼jn
vn)
• (vi, 𝛼j1v1) + (vi, 𝛼j2v2) +....+ (vi, 𝛼jivi) +....+ (vi, 𝛼jn vn)
• =  j1 (vi, v1) +  j 2 (vi,v2) +....+  ji (vi, vi) +....+  jn (vi, vn)
= 0 + 0 + ...... +  ji 1 + ..... + 0 =  ji
• That is, (vi, vjT) =  ji .........(2).
• From equations (1) and (2) we have 𝛽ij =  ji
• Hence the Theorem.

You might also like