Lecture 5 Vector Spaces and Linear Algebra: Vector...
Transcript of Lecture 5 Vector Spaces and Linear Algebra: Vector...
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Lecture 5
Vector Spaces and Linear Algebra:
Vector Representation of Wave States in Hilbert Spaces
Reading:
Notes and Brennan Chapter 1.6-1.7
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Basis vectors (a minimal set of orthogonal vectors that uniquely and completely describes all of vector space) can be added together to construct any point in the vector space.
Example: the unit vectors in Cartesian coordinates.
Similarly, we can think of orthogonal sets of wave functions as vectors and treat them similarly to simple vectors – i.e. add linear combinations of basis wave function sets to result in any arbitrary wave function within the wave space.
Lee Algebra is often used here: Lee was a pure mathematician who described his Linear Algebra as “finally a mathematical formulation that the physicists cannot corrupt with a useful purpose [paraphrase – not direct quote]”.
Fundamental Expansion Postulate
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Y AxisX Axis
Consider the case of two orthogonal wave functions (two states):
If Ψn and Ψm are orthogonal, then
mnnm dv δ=ΨΨ∫∞
∞−
*
However, in general, if Ψn and Ψm are orthogonal, then nothing can be said about
mnnopm Kdv δξ?
* =ΨΨ∫∞
∞−
because ξop may rotate Ψn into a projection of Ψm . However, if Ψnis an Eigenfunction of ξop then
mnnnMnnopM dvdv δλλξ ∫∫∞
∞−
∞
∞−
=ΨΨ=ΨΨ **
Fundamental Expansion Postulate
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
A basis set is a minimum set of functions that completely spans the space (i.e. all regions in space can be derived from linear combinations of the basis set).
Example: Fourier Series or unit vectors x, y, z.
Since Eigenfunctions of a Hermitian operator having unequal Eigenvalues are mutually orthogonal if they are complete, they must form a basis set.
The Fundamental Expansion Postulate states that every physical observable can be represented by a Hermitian operator with a complete basis set of Eigenfunctions, Ψ1 ,Ψ2 , Ψ3, ... , Ψn and every physical state Ψ can be expanded as a linear combination of Eigenstates as
Where each coefficient is given as:
Fundamental Expansion Postulate
∑ Ψ=Ψi
iic
See discussion Brennan p. 42-46 for details
∫ ΨΨ= dvc ii*
This is sort of like a dot-product operation with a unit vector in normal vector Linear algebra: For example: if you want to know the yth coefficient of the vector r=2x+3y+7z, where x, y, and z are unit vectors (basis vectors) simply take r•y
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Proof:
Let Aop be a Hermitian operator with Eigenfunctions, Ψ1 ,Ψ2 , Ψ3, ... , Ψn and Eigenvalues A1, A2, A3, ... An. The arbitrary physical state Ψ can be expanded as a linear combination of Eigenstates as,
And the expectation value, <A> is then given by,
Therefore...
Fundamental Expansion Postulate
∑ Ψ=Ψi
iic
See discussion Brennan p. 42-46 for details
∫ ΨΨ= dvAA op*
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Proof (cont’d):Fundamental Expansion Postulate
See discussion Brennan p. 42-46 for details
( ) ( )
( )( )
( )
s' c theare weights the whereA s,Eigenvalue
theof average a weighted is A Thus,
cAA
,normalized isΨ since and
dvΨΨcAA
toleading ,cancel" termscross" all
δdvΨΨ
since and numbersjust are sc but the
dvΨAc...ΨAcΨAcΨc...ΨcΨcA
dvΨc...ΨcΨcAΨc...ΨcΨcA
dvΨcAΨcA
ΨdvAΨA
2ii
i
2ii
ii
*i
2ii
ijj*
i
i
nnn222111*
n*
n*
2*
2*
1*
1
nn2211op*
n*
n*
2*
2*
1*
1
iiiop
i
*i
*i
op*
∑
∑ ∫
∫
∫∫
∫ ∑∑
∫
=
=
=
++++=
++++=
=
=
Now using the Eigenvalue relationship
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Proof (cont’d):
But how do we find the |cj|2 ‘s? Given:
Before a measurement, the particle can be in an unidentifiable state Ψ (but is made up of a linear combination of Eigenstates). However, once measured, the particle is in a known Eigenstate, Ψi ,with known measurable variable Ai. The term “collapsing into known state Ψi “ is often used. The |ci|2 ‘s are the probability of collapse (probability that a measurement corresponding to Hermitian operator Aop of the particle in arbitrary state Ψ will result in an observable Ai).
Fundamental Expansion Postulate
See discussion Brennan p. 42-46 for details
i*
j
iiji
*j
ii
*ji
*j
ii
*ji
*j
iii
cΨdvΨ
δcΨdvΨ
dvΨΨcΨdvΨ
ΨΨcΨΨ
ΨcΨ
=
=
=
=
=
∫
∑∫
∫ ∑∫
∑
∑Multiply by Ψj
*
Integrate over all space
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Since we can treat orthogonal wave functions as vectors, we can define many linear algebra functions that apply to wave functions as well.
Hilbert Space: Linear Vector Spaces using “Vector Functions”
( ) ( )( ) ( ) ( )
i allfor 0Ψ then0 all If :States all of ceIndepencenLinear ΨΨΨ :State Nulla of Existance
ΨΨΨΨ and ΨµλλΨµ :ntiplicatioScalor MulΨΨΨΨΨΨ and ΨΨΨΨ :Addition veDistribiti
iiii
mom
nmnmmm
lmnlnmmnnm
≠≠
=++=+=
++=+++=+
∑λλ
λλλ
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
State projections onto each other:
While basis vectors are orthogonal, any two generalized states (each made up of linear combinations of basis states) may not be. Thus, we can define an “overlap operation” similar to a dot product in traditional linear algebra. Similarly to the dot product, this operation returns a scalar representing how much overlap is between the two states:
Linear Vector Spaces
( )( )
∑
∑∑
∫∑∑
∫ ∑∑
∫
∫
=
=
=
=
=
=
∞
∞−
∞
∞−
∞
∞−
∞
∞−
ii
*i
jj
*i
i
ji
j*i
jj
*i
i
jjj
i
*i
*i
n*m
n*m
bad
bad
,Ψ and Ψ functions, basis orthogonal used weif againbut
dvΨΨbad
dvΨbΨad
dvΨΨd
dvΨΨd
ijδ
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
What is an Operator in the Wave space formulism?
An operator merely maps one wave state to another with both initial and final states being in the Hilbert space and thus, both initial and final states can be represented by linear combinations of the basis vector states. Formally this is simply:
Thus an operator merely changes the state from one state to another. Consider the simple “5 times” operator: The new state is mathematically different for the old state by a factor of 5.
Linear Vector Spaces
oldopnew ΨΨ ξ=
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
The Fundamental Expansion Postulate can be rewritten in matrix form. Every physical observable can be represented by a Hermitian operator with a complete basis set of Eigenfunctions, Ψ1 , Ψ2 , Ψ3, ... , Ψn and every physical state Ψ can be expanded as a linear combination of Eigenstates as
Matrix Methods in Quantum Mechanics
[ ]
Ψ
ΨΨΨ
=Ψ
Ψ=Ψ ∑
n
n
iii
cccc
c
M
L 3
2
1
321
See discussion Brennan section 1.7 for details
∫ ΨΨ= dvc ii*
Integrals like these that we have been using all semester are often called matrix elements – now you can see why.
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Thus, every generalized state can be represented by a vector, c, which are the coefficients of the basis wave states (coefficients of the basis vectors).
Since an operator merely changes from one state to another, the operator then can be represented as a matrix as well. Consider:
Matrix Methods in Quantum Mechanics
[ ] [ ]
=
=
Ψ
ΨΨΨ
=
Ψ
ΨΨΨ
=
== ∑∑
n
3
2
1
nnn1
31
2n2221
1n131211
n
3
2
1
copa
n
3
2
1
n321a
n
3
2
1
n321c
iiia
iiic
c
ccc
ξξ
ξξξξξξξξ
a
aaa
ΨξΨ sincebut
aaaaΨ and ccccΨ
ΨaΨ and ΨcΨ
MOM
OM
L
L
M
M
L
M
L
See discussion Brennan section 1.7 for details
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
But how do we find the individual matrix elements ξij? Consider our two states Ψc and Ψa:
Matrix Methods in Quantum Mechanics
( )
( )
( )
( )
( )
∫∑
∑ ∫
∑ ∫
∑ ∫∑
∑ ∫∫∑
∑ ∫∫ ∑
∑ ∫∫
∑
∑∑
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
Ψ==
Ψ=
Ψ=
Ψ=
Ψ=Ψ
Ψ=Ψ
Ψ=Ψ
Ψ
=
===
-i
*ji
ij
ii -
i*jj
i -ii
*jj
i -ii
*j
ii
i -ii
*j
-i
*ji
i
i -ii
*j
- iii
*j
c
i -ii
*j
-a
*j
*j
iiia
iiia
iiiccopa
Ψ where,ca
or cΨa
as, rewritten be can side handright But the
Ψca
Ψca
ΨcΨa
ΨcΨa
side, handleft theon ngonly worki and ,Ψfor in gsubtitutinNow
ΨcΨ
space, allover gintegratin and by sides both gMultiplyin
ΨcΨ operator, theApplying
ΨaΨ and ΨcΨ whereΨξΨ
dv
dv
dv
dv
dvdv
dvdv
dvdv
opjiji
op
op
opij
op
op
op
op
ξξξ
ξ
ξ
ξδ
ξ
ξ
ξ
ξ
See discussion Brennan section 1.7 for details
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
The two state vectors can thus be related by a transformation (operator) matrix made up of matrix elements ξij.
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
[ ] [ ]
∫∑
∑∑
∞
∞
Ψ==
=
=
Ψ
ΨΨΨ
=
Ψ
ΨΨΨ
=
==
-i
*ji
ij
n
3
2
1
nnn1
31
2n2221
1n131211
n
3
2
1
copa
n
3
2
1
n321a
n
3
2
1
n321c
iiia
iiic
Ψ where,ca
c
ccc
ξξ
ξξξξξξξξ
a
aaa
ΨξΨ sincebut
aaaaΨ and ccccΨ
ΨaΨ and ΨcΨ
dvopjiji ξξξ
MOM
OM
L
L
M
M
L
M
L
Thus, many Quantum Mechanics problems reduce down to calculating integrals for each matrix element and then solving NUMERICAL matrix equations.
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Consider what happens to the operator matrix if the basis wave set are Eigenfunctions of the operator.
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
( )( )
( )
( )
=
==
=
=
=Ψ=Ψ=
Ψ=
=
∫∫
∫∞
∞
∞
∞
∞
∞
n
3
2
1
nn
333
222
111
n
3
2
1
-i
*j
-i
*j
-i
*j
n
3
2
1
nnn1
31
2n2221
1n131211
n
3
2
1
c
ccc
ξ0
0ξ00ξ0000ξ
a
aaa
ΨΨ
Ψ where
c
ccc
ξξ
ξξξξξξξξ
a
aaa
MMOM
M
L
L
M
MOM
OM
L
L
M
n
ijiiiji
opji
dvdv
dv
λ
λλ
λ
δλλλξ
ξξ
Thus, using a Eigenfunction basis set results in a diagonal matrix whose elements are the Eigenvalues. Working in reverse, if we can diagonalize the matrix, we can find all the Eigenvalues along the diagonal.
[ ] [ ]
Ψ
ΨΨΨ
=
Ψ
ΨΨΨ
=
n
3
2
1
n321a
n
3
2
1
n321c aaaaΨ and ccccΨM
L
M
L
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Consider the case of the Schrödinger Equation (described in detail in the next lecture). This equation can be written as an Eigenvalue equation of the form:
HΨ=EΨ
where H is the Hamiltonian operator (total energy) and E is the energy Eigenvalue.
Considering Ψ to be a vector wave state, we can rewrite the Schrödinger Equation as,
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
( ) 0
0
becomes, thisleft, thefrom sideright thegsubtractin and Ψ defining
space, allover gintegratin and by gMultiplyin
-i
*j
-
*j
-
*j
-
*j
-
*j
*j
=−
=−
Ψ=
ΨΨ=ΨΨ
ΨΨ=ΨΨ
Ψ
Ψ=Ψ
∑
∑∑
∫
∫∑∫∑
∫ ∑∫ ∑
∑∑
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
iijiji
iji
iiji
i
ji
ii
iii
i
iii
iii
iii
iii
EHc
EcHc
dvHH
dvEcdvHc
dvcEdvcH
cEcH
δ
δ
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Continued...Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
( )
( )( )
( )
( )
0
c
ccc
HH
HHHHHHHHHH
form,matrix inor
0
0
becomes, thisleft, thefrom sideright thegsubtractin and Ψ defining
space, allover gintegratin and by gMultiplyin
n
3
2
1
nnn1
3n3331
2n2221
1n131211
-i
*j
-
*j
-
*j
-
*j
-
*j
*j
=
−
−−
−
=−
=−
Ψ=
ΨΨ=ΨΨ
ΨΨ=ΨΨ
Ψ
Ψ=Ψ
∑
∑∑
∫
∫∑∫∑
∫ ∑∫ ∑
∑∑
∞
∞
∞
∞
∞
∞
∞
∞
∞
∞
MMOM
M
L
L
E
EE
E
EHc
EcHc
dvHH
dvEcdvHc
dvcEdvcH
cEcH
iijiji
iji
iiji
i
ji
ii
iii
i
iii
iii
iii
iii
δ
δ
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Example: Consider the system whose Hamiltonian matrix elements,
evaluate to:
What are the possible results when the energy is measured?
Writing the (H-E)Ψ=0 => (H-E)(c1Ψ1+c2Ψ2+c3Ψ3)=0 matrix and solving, we get:
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
Ψ -
i*j∫
∞
∞
Ψ= dvHH ji
−−−
−=
310121
013H
( )( )
( )
( )( )
( )0
310121
013DET
:for solution nontriviala has which
0310
121013
3
2
1
=
−−−−−
−−
=
−−−−−
−−
EE
E
ccc
EE
E
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
What are the possible results when the energy is measured?Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
( )( )
( )
( )( )
( )
( )
4or 1, 3,E
0)4)(1)(3(0)45)(3(
01)2(
010
)3(101
)1()3(1
1)2(E-3
minors,by Expanding
0310
121013
DET
:for solution nontriviala has which
0310
121013
2
3
2
1
=
=−−−=+−−
=−−
−+
−−−
−−−−−−
=
−−−−−
−−
=
−−−−−
−−
EEEEEE
EEEE
EE
E
ccc
EE
E
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
What are the corresponding Eigenvectors ΨA, ΨB, and ΨC, for each of these energy Eigenvalues? For E=1,
Similarly for E=3 and E=4,
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
( )( )
( )
[ ]
ΨΨΨ
=Ψ
=+−=−+−=+−
=
−−−−−
−−
3
2
1
321
321
321
3
2
1
121
yields,unknowns 3 and equations 3 theseSolving
0200002
:of solutiona has which
01310
11210113
A
ccccccccc
ccc
[ ]
ΨΨΨ
−=Ψ
3
2
1
101B [ ]
ΨΨΨ
−=Ψ
3
2
1
111C
Clearly these Eigenvectors are not normalized.
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
If the ACTUAL state of the particle is described by the vector,
What is the probability of each Eigen-energy (the only possible measurable values) being measured?
After we normalize the states we have previously found and the state above, we must evaluate the overlap of these states. Mathematically this is either,
as described previously or:
For i=A, B and C or equivalently in matrix form,
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
[ ]
ΨΨΨ
−=Ψ
3
2
1
232actual
∫ ΨΨ= dvc Actualii*
ActualCCActualBBActualAA ΨΨc ΨΨc ΨΨc •=•=•=
ECE 6451 - Dr. Alan DoolittleGeorgia Tech
Normalizing each of the relevant vectors we get,
Thus, the probability of measuring ...
...Energy=1 is 6/17
...Energy=3 is 8/17
...Energy=4 is 3/17
Matrix Methods in Quantum Mechanics
See discussion Brennan section 1.7 for details
[ ]
ΨΨΨ
−=Ψ
3
2
1
232171
actual
[ ] [ ] [ ] [ ] [ ] [ ]
173c
178c
176c
232171111
31c232
171101
21c232
171121
61c
ΨΨcΨΨcΨΨc
CBA
CBA
ActualCCActualBBActualAA
===
−•−=−•−=−•=
•=•=•=
[ ]
ΨΨΨ
=Ψ
3
2
1
1216
1A [ ]
ΨΨΨ
−=Ψ
3
2
1
1012
1B [ ]
ΨΨΨ
−=Ψ
3
2
1
1113
1C