Goal :

Revise on what you have learned about basic objects in QM : 4 &

> abstract out some

concepts

Establish the basic picture about the math structure of QM for physists.

Introduce density matrix ; If time allows

,
also quantum entropy & entanglement.

Requirements :

can deal with abstract states /

w .o .
reference to
wavefunctions
operators matrices .

Historical remarks : (ref : S .

Weinberg Lectures on RM)

QM (3410 & 3420 and many others)

,
usually describes quantum states by means of
wave functions ,
which are
funcs of positions of objects of the system ,
e.
g. ,
solving harmonic oscillator and hydrogen L
.

>
wave mechanic by Schrodinger
very important ,
show
up constantly
fanot also by means of matrix mechanics by Heisenberg ,
e .
..
8

solving spin system ] !

But fundamental physical ?

why shoul we
single out
position as the observable

Limitations : is sometimes , position is not fundamental. But rather , we want to

describe the system by momenta or

energy ;
Even
further, observables be described
ii) some can not
by position
or momenta at all, e .S . ,
spin !

# P . M . Dirac
"transformation theory" : describe the
principles of QM
in a unified way
.

In the
following ,
I will first summarize the "axiom"
(posulate) of QM. Then
we will dive into details of these
aspects and
reflect on what
you should have
known.
 Summary Saxiom" (posulate) of
of QM : Cref : Cohen Vol1 ,
Ch . 111)

1) State postulate : Physical states

represented by rays in Hilbert
space
are .

space ()
Hilbert A normed complex vector
space.
:

recall from linear algebra addition is associative & commutative

I
,
;
identity of addition (zero vector) ;

multil lication
> .

I
if , II, I"Ext, <
, Ed ,
then <4 + x 4 Elt
# + 41 = 4+ 2 (4+4 + 4" =
4 + (4'+ 4") ,
xk41 = 224
, .

0 + 4 =
4 =
4

S
o

0
0x4
=

↓8 = 0 .

>
-
sinner product)
It with norm ( ) : Hx() > K
.
equips
a ,
,

*
(4) <E , =

1 ( Xi4, + X 242)

(E, )
,
X, ( 41)

Io ,
(E , El
=

= 0
if
,
+

=
xc(,42)

0 .
.

Exercise 1 : 0 (4 , 4) =
/Exiti Exj4; ) = xXj (4 , :, 4i

prove the Gram determinant det (41 , 4; ) : To

S
->
Gij
(i
(4 4.) ,
(4+, 42)--- 14· Hul
G =

14n , 4r) - -
Stuctul
n= 2 case is the
Cauchy-Schwarz inequality.
umk :
superposition principle is
implied b C/. the vector
space!

Rays : A ray is a set of normalized vestors ((4 , 4) =

1) with

4) ~It' If 41 = 84 181 1 = .

From now on
,
we will adopt Dirac's brad ket formalism :
1. StateIt () "Ket" elements in 1

.
2 "bras" <41 : finear functionals defined on i : ( > C, > (4
,
).
*
elements in the dual
space (-
[Xit+ Xnital = X* <4. 1 + X*421 !

Short-hand : < 41 := (4,)

3 . Wave func .
S
4(x,+ ) = <XMX
"Component" in the expansion of 14) : X

in the basis
of position eigenfunctions .

momentum-space wave
funs S
.

Esp + ) < 4). .

=
 Riesz- Frechet theorem :
Any continuous linear
functional
f : <) <@ ,
41 ( f(2) ,
corresponds to a

wave
func .

Ug ,
S . t ., f = <
Hyl , f(4) =<44)

2) Observable postulate : observables are

represented by Hermitian operators,
acting finearly in 2. ruk: note8Ex**
(B)= B At
+

& 8: H < ( , 8(x14 + Xu 142)) = X, 8141) + x281427 .

C8t = 11
-
· 7884) =
(, 84) =
(8 , ) = (E) = C18TN
(Hermitian Conjugate
is called the
adjoint operator of 8
if
.

Hermitian
+
8 Hermitian
If 8 =
,
then it is called I reality)
for
self-adjont) V
Crank: anti-Hermitian operator
8t = 5 -
as =
(4,06)
*

A fundamental theorem in linear algebra :

i) eigenvalues of a Hermitian operator 8 are all heal ;

84) = X /4)
,
then XERR
ii) eigenvectors
(rays) with
different eigenvalues are
orthogonal.
& 14: ) = Xi14i) , if Xix;, then <4: 14j ) = 0.

Some remarks on 1) & 2)

① dimb) = m .
(4) = 2
, 197 + &(e) + ...
xulen) , le =
(i) (i) .
e=

(
,

Tl Tir

~
...
Tin
F
(E) = (8 P , ,
-..

(i) =
② In wave funs .

Language useal in QM .
(4(X) =
(14))
a
physical State 41x)> (4) 1= Sdx 141x)/"
3) Measurement Postulate (Wavefunc . Collacse) : A system is in state 27. If an

observable (Hermitian operator) with eigenvalues /Xu) & eigenvectors [4n))·

2 the probability of obtaining a
particular eigenvalue Xu is :

P(4 > 4n)) =

<44n)? .

ii) If Xu is obtained , 47 collapses to 4n)·

Vie I
rmR : 1 .
What
happens if i) Xu is degenerate i. e ., , ,
S t.,.

4n i) = Xn4n i)
, ,

ii) the
spectrum (xn] is continuous ?

.
2 In Lef . Cohen ,
this
postulate is
separated into three !

4) Time Evolution Postulate : The time evolution of a state rector is governed

by Schrodinger
the equation :

# : Hamiltonian it 4) # 4) =
,
W. A being the observable
associated the total energy of the system.

4(t)) = [ (t , to) 4Hol) if # ,

indep of time, then U = exp *H-toF

↓ Schrodinger
picture

Link : i) Here the state evolves time , while .

assume w. /
operators not

ii) This postulate fact, not as essential as the P 1 P 2 P 3

is , in
, . , . ,
.

whenspace-time) is taken into account

.
Symmetry
But I'm not saying it is not important I mean it will be
,
rephrased. .
 Some Details of
Postulate.
I Hilbert
Space & Wavefuns . on Coordinate
Space V
>
- Saviti */ *

- linear
space of normalizable wave
funcs ·
on V : ((V)

naturally equipped wel an inner product :

(2, 2) : = (dV*(*)[ *)
e .

.:
g V= S" =
((0 , 2) C, 08) dv = sinocade
spheric
,

harmonics
normalizable : Sau 1410 es1 <

S Y10 2)
,

: a basis
B C Vo => ,

& 4)
. .
:
410, 01 = 410, 12) ,

0
, 2) = 410 01 . , 4, 4) = 415, 03 Ve

Q: What if V is not
compact ? i e
.

., Sav = *
,
e .
g .
plane wave in
open
space?

I Composition of
* · Systems :
important in
reality and relevant for entanglement.
( no identical particles for now)

Begin With the following Q :

in
a one particle two potential wells b) two inequir particles in two potential wells

#
or
# ·

V, Vz V, Vz ,
V V

What the Hilbert ?

are
spaces
To answer it , know direct & tensor
we need to Sum direct
,
produce product of
two vector
spaces V& W
.

An concrete example to a comprehensive understanding :

Let V = IR" , W =3 · v = (X , (2) EV ,
w = (Y, , Y2 , 4) E W
.

I direct
-
sum : V + W =' += 15 ,
dim VOW = 2+ ) =5

same for
dimV < i
[dimwa (, . xz, y . ,Yz, ]) = (v, w)
-
E VQW
-
Cartesian product
not bilinear

product : VXW = IR** RT VXW 3

2
·
& direct X dim = 2+ = 5
,

(X. . X , Y , ,n
w)
Y) (V , Cartesiam
,
=
-
>
product
bilinear VQW = * = Ro
'
③ tensor Product : dim VQW = 2x3 = 6
V &W
,
in
-

(x
1Y ,
X ,Ye
, XiY, XcY,, XYr XiY) ,
= vowEVO WO !
 For finite-dimension sub-spaces VOWIVXW . ,

But it is not true for infinite-dimension ones or infinite families .

Basically Vo VX.
,

In physics this distinction is not essential And it is the tensor

,
.
product
that is more
often usea) , b .
C. / we require multi-linear
property.

Situation 1) A system (say a particle) not

only has external d . o .
f's
like its position , but also d f's , like its
internal
spin
.
o.

or color ;

2) Two systems : S ,& S2 to a bigger system .

tak : O is also
useful when constructing the Fock space
!

Example : v= Hilbert
space of
the electron in hydrogen w. o/
spin.
V =
41x) EV being wavefunc
the .

W= Hilbert
space---for spin , which is of dimension 2
.

spin or
spin ↓

5) = VDW Hilbert the election.-

space of
~> is the

Now, some math : Let H , & It- be two Hilbert Denote the tensor
produce
space .

H H2, and the tensor

space · .
produce of two vectors as

147818) ,
W ./ 14) Esti, 187 Este .

It is
easy to see : 11) (X, 147+ X 142)) @1) 2
= X, 14,7@18 + X 2142) @ 187

14) * (X,1+ X182)] = X, 147* 1) + x 2(4)@1E27 .

(2) H, H2 is
spanned by [14i) @ 10% 3) ,
where

((ui)] is a basis of 7, and Sluj)] is a basis

of He.
So dim (7,* Saim (2)
H2) =
<dim() x ·

Let 14) =
[di(i) , 18) =
[ bj/Vj),
j

then 14) * 1)
Zaib; (i) * (; ·
= )

But there exist in H= /t. @H2 rectors that are NOT tensor products of
a vestor 4, and a vector IE7

7) . = RC , Hr**" , both spanned by Sle

Example :
,
le3]
. Then

(e) = 191) @ (f) + 19) 19 .7 = 4 ·

proof : If it does , let it= a,197 + acles & E= b, 191) + balfe]
 bu /

Slazby
a, =

then
4 zb. 1
!
=
> no solution
ab , =0
I
=o

many elements in tensor SHMs of

This
implies that a
product space are

simple tensors
,
not SINGLE simple tensors themselves !

my entanglement !

Answer of the Question in the beginning :

· direct sum ( Hi(Vi)
: = + Hell

(x1))
(14
HER) = ,
,
XIEV => denote as 1417@142) EHIGSth
42(X2)) ,
XcE V

inner product : (4 ,
04
,
4. 042) = <4, 4 / v .
,
+ <42 , 42) v

· tensor product
: ( = H (vi)
, x Hr(z)

7) =47 = 4, 1x1) 42 (x2) ·

=> denote as 1413 * 142]

inner
produce : (1 , 4. 8th) =
St , 4. /v· Sitelve

discussed tensor What

We have
products of Hilbert
space and states .
about

operators ?
Assume two Hilbert
space It . & Str ,
which have bleis & /Ifi7] as

their basis respectively.

Any in the tensor-product ( = ( .@ (

state
spaceH) E :

(4) =
i, j
(ijlei) &Ifj) ·

Let o be any operator acting o n 21. , 98 ...

72.

Then tensor-product operator 8 0., defined as

· 14) =
00 =
Z. (j(0 lei)(Elfil) ,
.

clearly , there is one-to-one

correspondence between

8 & 8, x1 i & & 1 , 08

Besides, Since 8, acts on H. & Go acts on Str ,

= 8,, I = 0 ,

or in this form : IG1c ,

1, 1 = 0 .
 T
.
Complete Orthonormal basis .
anology of = (·Y)
i) orthonormal basis : <eilej) =
Sij i ↑
ii) complete orthonormal basis :
for any 14) ECt , 14) =
[li) <ei14)
>
-
I = 2 lei7(i) (complete relation or
closure)
itAll

iii) resolution of identity : 1 =

EH ceil , if ei
complete & Orthonormal
Link : One can insert this identity operator in any place.
what
if for outinuous spectrum jeigenvalues) ,
e g
.
.
(x).

iv) Pi = (i)(til is an
operator acting from 11 to 1 . Assume [19in]
is
complete & orthonormal .

It is easy to
verify it =
Pi , Dr = Bi &: 1 .

sit all
exercise :
eigenvalues of
: () = <til4]lei)
↑?
·

Usually, Due uses

eigenvalues to index the basis : (lex)·

↑x = (2x)ex)
Mx(4) =<ex14)1ex) :
project on to the
X-eigenspace .

Say an
operator 8 has eigenvalues (X) & eigenvectors [ltx)) ,

then 8 can be written as

Ex &x
< ext
:= =
2x1ex)

v p = 14.) <4) is also a linear operator : ( >

-
H , , if 14:7 @Si ·

& normally we take str = ( , ) .

Now + &P .

vi) .
matrix representation of operator : Assume a
complete & Orthonormal

basis [lei)]
<ille;
Oj =

It is also have but orthonormal

common that we a
complete not

basis (Ifj) .
Then we can have a matrix reples , as well :

<:181 fi >
Example :
Spin system of an electron

two

dimH = 2 . It is
spanned by Veigenvectors of Sz ,
which we denote as

It & 1)
, They satisfy :

SzlE = ) & <(f) = 1 , <+ 1 -) = 0

.

Then z = 1 +> +1 + ( -
E)1-)
1 = 1+)(+ 1 + 1 --

In this basic , 2 has the matrix repres.

5 =
) : i)
What about 5x &
Sy ? ref :
Sakurai 3rd
Chap
.
14
.

(Sx1+) =
= )(+) + gif 1-)
,

+ i-
(5x , = =
= ((+) -
e 1 ))
-

x
= [(x +)(+, (xy -
18x +
, )c -
, 1]

=(f
is
eib 1)
-

H+)c -
-

1
-

= + >+
-