Quantum Mechanics 1
Lecture 2: Wave Mechanics
Massimiliano Grazzini
University of Zurich
Outline
The double slit experiment
The structure of the quantum amplitude
Classical limit
Sum over the paths
The uncertainty principle
The free-particle kernel
The wave function and the Schrödinger equation
Free particles
The double slit experiment
screen
detector
source
1 2
A source radiates electrons all of the same energy that come out in all directions The screen has two holes, 1 and 2 through which the electrons can pass
D x
Experimentally one finds that the probability as a function of is the complicated curve on the right with maxima and minima and there are regions of in which the electrons hardly arrive
x x
The detector counts how many electrons arrive at x
A relatively recent experiment has verified this phenomenon with the largest and most classical objects, which are currently the fullerene molecules
(Nairz, Arndt, and Zeilinger (2003)
The fullerene molecule , consisting of 60 carbon atoms arranged in a truncated
icosahedral shape, is the smallest known natural soccer ball.
C
60Given the large dimension one would expect that these molecules behave like particles under ordinary circumstances.
The multi-slit experiment by Nairz et al.
has instead shown their wave nature
The double slit experiment
The double slit experiment
With this experimental setup, it would seem reasonable to assume that:
Each electron must go through the slit 1 or through the slit 2
The chance to arrive to should be the sum of two parts: the chance to arrive through the slit 1 and the chance to arrive through the slit 2 x P
2P
1screen
detector
source
1 2
D
x
The double slit experiment
The sum of the two clearly does not give the previous curve so we conclude that
screen
detector
source
1 2
D x
We may easily check this by closing one of the two slits and doing again the experiment: the results are shown below
P ≠ P
1+ P
2The curve is familiar: it is exactly the
interference pattern expected if waves starting from the source pass through the two holes
P(x)
The simplest way to represent wave amplitudes is through complex numbers
We are led to introduce two complex amplitudes and such that P = | A |
2with A = A
1+ A
2and P
1= | A
1|
2P
2A =
1| A
2|
2A
2Logical difficulty: it is not true that P = P
1+ P
2It is not true that when both holes are open the particle goes from one hole or the other !
If this were the case we could classify all arrivals at in two disjoint classes:
those coming from the hole 1 and those from the hole 2 x
The effect of observation
We have concluded on logical grounds that since we cannot assume that the electron passes either through hole 1 or hole 2 P ≠ P
1+ P
2But we can devise an experiment to check this directly
source
A source of light behind the screen allows us to watch and see through which hole the electron passes
Electron scatter light: if light is scattered behind hole 1, we may conclude that an electron passed through hole 1; if it is scattered close to hole 2, the electron
passed through hole 2
1
2
For each electron arriving at we can now experimentally say whether it came through hole 1 or 2
x
The result is that the probability distributions and are indeed given by the curves on the right, and that
P
1P
2P = P
1+ P
2When we watch the electrons to see from which hole they passed we find
P = P1 + P2When we do not watch we get a different result
P = |A1 + A2|2 ≠ P1 + P2The answer is that we have used light: the light in collision with the electron may alter its motion and its chance to arrive to a given point on the detector
We could use very low intensity, but weak light does not mean weak disturbance When we have low intensity, it means that less photons are used, but this way we may miss the electron
What happened ?
But when we do see an electron, it means that a photon was scattered, and a momentum of order h / λ was transferred to the electron
How about taking very large wavelengths ? This way the photon momentum
would be smaller λ
But there is a limit to this: if too long a wavelength is used, we will not be able to say whether the scattering was behind hole 1 or hole 2
We thus see that any experiment designed to determine through which hole the electron passes must produce enough disturbance to alter the ensuing distribution It was first pointed out by Heisenberg in his uncertainty principle that the
consistency of the new mechanics implies a limitation to the precision to which an experiment could be performed
In our case the principle states that any attempt to design an apparatus to determine
through which hole the electron passed, and delicate enough so as not to deflect the
electron enough to destroy the interference pattern is bound to fail
The uncertainty principle
source
When an electron is deflected the vertical component of its
momentum is changed
Suppose that the screen is not fixed but it is free to move up and down and that we can measure its displacement
p p
l
a
δp
2δp
1For small deflection angles the momentum keeps the same
magnitude and we have
| p
1− p
2| / | p | = | δp | / p ∼ a / l p
1p
2The uncertainty principle
The interference pattern of the original experiment is the sign of a wave-like behavior of the electrons
The two beams of electrons will interfere constructively and give a maximum in the diffraction pattern at the center of the screen
As we move down from the center the next maximum will occur at a distance d such that the beam from the hole 1 has travelled exactly one wavelength more than the beam from 2
The following relation holds
First (central) maximum Second
maximum
λ
a ∼ d
l
The uncertainty principle
If we want to say which hole the electron passed through, the momentum imbalance must be measured with precision better than , that is δp
If we want to see the interference pattern we must control the position of the screen with Δx < d
Combining the two requirements and using λ and we obtain
a ∼ d
l δp / p ∼ a / l
We thus conclude that if we want to say which hole the electron passed
through and at the same time we want to observe the interference pattern we must have
Δp < δp = λ
d p < λ
Δx p = h Δx
Use De Broglie relation
λ = h/pwhich is (as order of magnitude) what the uncertainty principle forbids !
ΔxΔp < h
Δp < δp
screen
detector
source
D
Suppose now that we add more holes….
The structure of the amplitude
screen
detector
source
D
Suppose now that we add more holes….
The structure of the amplitude
screen
detector
source
D
Suppose now that we add more holes….
The structure of the amplitude
screen
detector
source
D
The amplitude will be the sum of the
amplitudes over the interfering alternatives Suppose now that we add more holes….
Eventually, if we remove the screen we obtain
The structure of the amplitude
A = ∑
i
A
iA = ∑ A
pathdetector
source
D
The structure of the amplitude
But what is the form of A
path?
Sum over all the possible paths connecting the source and a point on the screen
It can be obtained by considering the analogy with geometrical optics
A = ∑
paths
A
pathThe structure of the amplitude
In geometrical optics the path of light rays is determined by the Fermat principle: the “optical path” of the ray, that is, the phase
difference in the initial and final points of the path must be minimum In classical mechanics the particle trajectory can be determined
according to the principle of minimal action: the action functional computed along the classical trajectory must be minimum
Starting from this analogy we can assume that the phase of the
amplitude on a given path must be proportional to the classical action computed along that path. The proportionality constant should be
identified with the Planck constant ℏ
Geometrical optics is obtained from wave optics by taking the limit λ → 0
A
path∼ exp{iS[path]/ ℏ}
detector
source
D
Classical limit
Sum over all the possible paths connecting the source and a point on the screen
In the classical limit the action is large in absolute value and oscillates very rapidly In this limit ( ) only the classical trajectory contributes: the
contributions from all the other trajectories cancel out ℏ → 0 A = ∑
paths
exp{iS[path]/ ℏ}
detector
source
D
Classical limit
Sum over all the possible paths connecting the source and a point on the screen
In this limit ( ) only the classical trajectory contributes: the contributions from all the other trajectories cancel out ℏ → 0
In the classical limit the action is large in absolute value and oscillates very rapidly A = ∑
paths
exp{iS[path]/ ℏ}
→ exp{iS
cl/ ℏ}
Classical limit
We can thus summarise this as
Quantum Mechanics Wave optics
Classical mechanics ~ Geometrical optics
ℏ → 0 λ → 0
~
Sum over the paths
We consider a particle moving in one dimension from the point at time to
the point at time x
bt
bx
at
aThe amplitude to go from to is obtained as a b K(b, a) = ∑
x(t)
exp{iS(x(t))/ ℏ}
Although the qualitative idea of the sum over the paths is clear, we need a precise
mathematical definition
The Riemann integral of a function is
defined as f (x)
Δx→0
lim ∑
i
f (x
i)Δx
iWe can follow an analogous procedure to define the sum over the paths
x f (x)
x
ix
i+1Δx
tb
ta
xa xb
x(t)
We divide the time interval in steps of width
t
b− t
aϵ
t
bt
ax
ax
ix
i+1x
bt
it
i+1ϵ
K(b, a) = lim
ϵ→0
1 A ∫
+∞
−∞
dx
1A ∫
+∞
−∞
dx
2A . . . . ∫
+∞
−∞
dx
N−1A exp{iS(x(t))/ ℏ}
For the moment we limit ourselves to consider a free particle
L (x, · x, t) = 1
2 m x ·
2∼ m
2 ( x
i+1− x
iϵ )
2
For each of the values at the time we integrate over t
ix
ix
iS[x(t)] = ∫
tb
ta
L(x, · x, t)dt
We introduce a constant whose precise value will be fixed later
A
≡ ∫ 𝒟x(t)exp{iS(x(t))/ ℏ}
S[x(t)] = m ϵ 2
N−1
∑
i=0( x
i+1− x
iϵ )
2
K(b, a) = lim
ϵ→0
1 A ∫
+∞
−∞
dx
1A ∫
+∞
−∞
dx
2A . . . . ∫
+∞
−∞
dx
N−1A exp{iS(x(t))/ ℏ}
We can then write the amplitude K(b, a)
= lim
ϵ→0
1 A
N∫
+∞
−∞
dx
1. . . . dx
N−1exp { im
2ℏ ϵ
N−1∑
i=0
( x
i+1− x
iϵ )
2
}
= lim
ϵ→0
1 AN ∫
+∞
−∞ dx2. . . . dxN−1exp{ im 2ℏϵ
N−1
∑i=2 (xi+1 − xi)2}∫
+∞
−∞ dx1exp{ im
2ℏϵ [(x2 − x1)2 + (x1 − xa)2]} 2(x1−(x2+xa)/2)2+(x2−xa)2/2
The integral over is now a Gaussian integral and it gives x
1∫
+∞
−∞
dx
1exp { im
2ℏϵ [ (x
2− x
1)
2+ (x
1− x
a)
2] } = π iℏϵ
m exp { im
4ℏϵ (x
2− x
a)
2}
(recall ∫ dx exp{iax
2} = iπ / a )
K(b, a) = lim
ϵ→0
1 A
N∫
+∞
−∞
dx
2. . . . dx
N−1exp { im 2ℏϵ
N−1
∑
i=2( x
i+1− x
i)
2} π iℏϵ
m exp { im
4ℏϵ (x
2− x
a)
2}
We can then proceed in a similar way with the integral over x
2= lim
ϵ→0
1
AN iπℏϵ m ∫
+∞
−∞ dx3. . . . dxN−1exp{ im 2ℏϵ
N−1
∑i=3 (xi+1 − xi)2}∫
+∞
−∞ dx2 exp{ im
2ℏϵ [(x3 − x2)2 + (x2 − xa)2/2]} 3/2(x2−(2x3+xa)/3)2+(x3−xa)2/3
= lim
ϵ→0
1
A
N2iπ ℏϵ 2m ∫
+∞
−∞
dx
3. . . . dx
N−1exp { im 2ℏϵ
N−1
∑
i=3( x
i+1− x
i)
2} 4iπ ℏϵ
3m exp { im
6ℏϵ (x
3− x
a)
2}
Factor
1/ 2Factor
2/3By repeating the procedure recursively we obtain
K(b, a) = lim
ϵ→0
1
A
N( 2π iℏϵ
m )
(N−1)/2
1
2 ⋅ 2
3 ⋅ 3
4 . . . . N − 1
N exp { im
2ℏϵN (x
b− x
a)
2}
= lim
ϵ→0
1
A ( 2π iℏϵ A
2m )
(N−1)/2
ϵ
ϵN exp { im
2ℏ(t − t ) (x
b− x
a)
2}
If we now set A = 2 π iℏϵ we find
m
K(b, a) = m
2π iℏ(t
b− t
a) exp { im
2ℏ(t
b− t
a) (x
b− x
a)
2}
By setting t
a= 0, x
a= 0, t
b= t, x
b= x we can simply write
K(b, a) → K(x, t) = m
2π iℏt exp { im
2ℏt x
2}
It is interesting to observe that the factor 1
2 m (x
b− x
a)
2t
b− t
acorresponds to the classical action along the path from to so we can
write S
cla b
K(b, a) = m
2π iℏ(t
b− t
a) exp { i
ℏ S
cl} K(b, a) = lim
ϵ→0
1
A ( 2π iℏϵ A
2m )
(N−1)/2
ϵ
ϵN exp { im
2ℏ(t
b− t
a) (x
b− x
a)
2}
tb − ta
The free-particle kernel
K(x, t) = m
2π iℏt exp { im
2ℏt x
2}
We now want to discuss some
implications of the free-particle kernel
At fixed time the real part of the amplitude varies with as t x
Re(K)
x
We see that as we get farther from the origin the oscillations become more and more rapid
What is the wavelength at large ? We can find it by solving x
2π = m(x + λ)2
2ℏt − mx2
2ℏt = mxλ
ℏt + mλ2 2ℏt
The free-particle kernel
K(x, t) = m
2π iℏt exp { im
2ℏt x
2}
We now want to discuss some
implications of the free-particle kernel
Re(K)
x
We see that as we get farther from the origin the oscillations become more and more rapid
2π = m(x + λ)2
2ℏt − mx2
2ℏt = mxλ
ℏt + mλ2 2ℏt
If we assume λ ≪ x we can neglect the second term and we find
λ = 2 π ℏ m(x / t)
At fixed time the real part of the amplitude varies with as t x
A classical free particle which moves from the origin in the time interval has
velocity and momentum x / t mx / t t
We can interpret the previous result by saying that when the motion can be described by assigning a classical momentum
to the particle the amplitude varies with a wavelength
p = mx / t
λ = h / p
We can repeat this discussion for the time dependence Suppose we keep the distance fixed and vary t
Re(K)
t
The oscillation period is defined as the time required to increase the phase by T 2π
2π = mx
22ℏt − mx
22ℏ(t + T )
= mx2
2ℏt (1 − 1
1 + T/t) = mx2
2ℏt2 T 1 + T/t
If we assume T ≪ t we find ω = 2 π
T ∼ m
2ℏ ( x t )
2
But is nothing but the classical energy of the particle, so we conclude that
E = 1 2 m ( x t )
2
E = ℏω = hν
The above conclusions can also be obtained by recalling that in classical mechanics the momentum and energy at the time and position are t
bx
bp = ∂S
cl∂x
bE = − ∂S
cl∂t
band that in the classical limit the kernel is
K(b, a) ∼ exp{ i
ℏScl(b,a)}
Events occurring in succession
The amplitude to go from to is obtained as a b K(b, a) = ∑
x(t)
exp{iS(x(t))/ ℏ}
Suppose that is some time between and : The action along any path between and can be written as a t
cb t
at
bS[b, a] = S[b, c] + S[c, a]
This follows from the definition of the action as an integral over time (and on the fact that it depends only on position and velocity)
It is possible to split any path in two
parts: the first part will have end points and and the second part
would have end points and
x
ax
c= x(t
c)
x
cx
bWe thus write K(b, a) = ∫ 𝒟x(t)exp{iS(x(t))/ ℏ} = ∫ 𝒟x(t)exp{i / ℏS[b, c] + i / ℏS[c, a]}
x
t
cc
x a
t b
Events occurring in succession
x
t
cc
x a
t b
We then have to integrate over all paths from to , over all paths from to and finally to integrate the result over all possible values of a c x
cc b In doing the first step the factor S[b, c] is constant, so we can write
K(b, a) = ∫
xcdx
c∫ 𝒟x(t)exp{i / ℏS[b, c]}K(c, a)
In the next step we integrate over the paths at fixed and we find x
cK(b, a) = ∫
xcdx
cK(b, c)K(c, a)
The kernel to go from to is the sum over all the possible values of of amplitudes for the particle to go from to and then to
a b
x
ca c b
Amplitudes for events occurring in
succession in time multiply
Wave function
The kernel K(b, a) describes the propagation of a particle from to a b
We want to generalise this result by considering a particle for which we do not know the initial position
By using the result on amplitudes occurring in succession we can write
Such amplitude has the same probability characteristics as those we already encountered: its square gives the probability of finding the particle at the point and at the time | ψ (x, t) |
2
x t
We will call this amplitude wave function
The wave function ψ (x, t) describes the state of the system at the time t
ψ (x
b, t
b) = ∫
+∞
−∞
dx
aK(x
b, t
b; x
a, t
a)ψ (x
a, t
a)
We call the total amplitude to arrive at (x, t) from the past ψ (x, t)
Towards new principles
We now want to summarise what we have learned so far
Quantum systems have an intrinsically casual behavior. We cannot say what
happens until we observe it. Only the probability distribution of a measurement is a physical property of the system
Probabilities should be computed by squaring transition amplitudes The amplitudes for different alternatives sum up like waves
(superposition principle) and interference phenomena occur The amplitude for a particle to go from to is obtained by summing the amplitude on each possible path (path integral)
a bThe contribution of the amplitude from a single path is obtained by
exp{iS/ℏ}where is the classical action evaluated on that path
SIn the classical limit
ℏ → 0only the classical path contributes
a
b
The state of a system is described by the wave function
ψ(x,t)The Schrödinger Equation
In Quantum Mechanics the wave function determines in a complete way the
state of a physical system ψ
This implies that such function, at a given time, defines also the behaviour of the system in the future
Mathematically this means that the derivative of the wave function must be
determined from at the same time ψ ∂ψ / ∂t
According to the superposition principle, such dependence must be linear We thus write in general
i ℏ ∂ ψ
∂t = H ψ
Where is a linear operator and the factor has been added for convenience H iℏ
Now, given the function its modulus squared gives the probability to find the particle at the time in the position ψ (x, t) | ψ (x, t) |
2
t x
If we integrate over the total probability must be independent on time x
We then have 0 = d
dt ∫ | ψ |
2dx = ∫ ∂ψ *
∂t ψ dx + ∫ ψ * ∂ ψ
∂t dx
Multiplying by and using the definition of we find iℏ H
0 = ∫ ( ψ *(H ψ) − (Hψ)*ψ ) dx
Using the definition of adjoint ∫ (Hψ)*ψ = ∫ ψ *(H
†ψ) (remember MMP1 !) we find
0 = ∫ ψ * ( H
†− H ) ψ dx
Given the arbitrarity of we conclude that must be self-adjoint ψ H
so its eigenvalues are real. Do they correspond to a physical quantity ?
H = H
†To answer this question we can check what happens in the classical limit In this limit we have
ψ = a exp{iS / ℏ}
Where a can be treated as a constant
If we act on this form of the wave function with iℏ ∂ we obtain
∂t iℏ ∂ ψ
∂t = iℏa exp{iS / ℏ} i
ℏ ∂S
∂t = − ∂ S
∂t ψ
Comparing with i ℏ ∂ ψ
∂t = H ψ
we conclude that in the classical limit the eigenvalues of the operator correspond to H −∂S / ∂t
But this quantity is nothing but the energy of the system
The operator is the Hamilton operator for the system H
We now want to see that the path integral formulation that we have outlined naturally leads to the Schrödinger equation
We start from the form of the wave function in terms of the kernel K(b, a)
We focus on a single particle in one dimension subject to a potential with
Lagrangian V(x, t)
ℒ(x, · x, t) = 1
2 m x ·
2− V(x, t)
Choosing t
b= t + ϵ , x
b= x and x
a= y t ,
a= t we can write
ψ (x, t + ϵ) = ∫ dy
A exp { i
ℏ ϵℒ ( x + y
2 , x − y
ϵ , t + ϵ
2 ) } ψ ( y, t)
= ∫ dy
A exp { im
2ℏϵ (x − y)
2} exp { − iϵ
ℏ V ( x + y
2 , t + ϵ
2 ) } ψ ( y, t)
ψ (x
b, t
b) = ∫
+∞
−∞
dx
aK(x
b, t
b; x
a, t
a)ψ (x
a, t
a)
If is small the difference must also be small, or the velocity should be very
large ϵ x − y
If is small the factor oscillates very rapidly and damps the integration to the region
ϵ exp { im
2ℏϵ (x − y)
2} y
x − y ∼ ϵℏ / m
It makes sense to expand the second factor in powers of and ϵ (x − y)
2We start from
ψ (x, t) + ϵ∂
tψ (x, t) ≃ ∫ dy
A exp { im(x − y)
22ℏϵ } ( 1 − iϵ
ℏ V(x, t) )
× [ ψ (x, t) + (x − y)∂
xψ (x, t) + ( x − y)
22 ∂
2xψ (x, t) ]
: comparing the left and right handed side we find
𝒪(ϵ
0)
ψ (x, t) = ∫ dy
A exp { im(x − y)
22ℏϵ } ψ (x, t) = 2 π iℏϵ
m 1
A ψ (x, t) ψ (x, t + ϵ) = ∫ dy
A exp { im
2ℏϵ (x − y)
2} exp { − iϵ
ℏ V ( x + y
2 , t + ϵ
2 ) } ψ (y, t)
and we find
So we can set consistently with what we have found for the free-particle kernel
A = 2 π iℏϵ m
Multiplying by we find iℏ
iℏ∂
tψ (x, t) = − ℏ
22m ∂
2xψ + V(x, t)ψ ψ (x, t) + ϵ∂
tψ (x, t) ≃ ∫ dy
A exp { im(x − y)
22ℏϵ } ( 1 − iϵ
ℏ V(x, t) )
× [ ψ (x, t) + (x − y)∂
xψ (x, t) + ( x − y)
22 ∂
2xψ (x, t) ]
The term of on the right-hand side vanishes because the integral is odd, consistently with the absence of 𝒪(x − y) ∼ 𝒪( ϵ ) 𝒪( ϵ ) on the left hand side
: let’s look again at our equation
𝒪( ϵ )
: at this order we have to keep also the term on the right
𝒪(ϵ) 𝒪 ( (x − y)
2)
ϵ∂
tψ (x, t) = ∫ dy
2 A exp { im(x − y)
22ℏϵ } (x − y)
2∂
2xψ + ϵ
iℏ V(x, t)ψ (x, t)
iℏϵ/(2m)
*
* Use ∫r2dre−ar2 = π
2 a−3/2
This equation can be written in the form i ℏ ∂ ψ
∂t = H ψ If we identify
H = − ℏ 2m
2∂
2x+ V(x, t)
Comments:
Hamilton operator
The classical Hamiltonian of the system is
H = p
22m + V(x, t)
The Hamilton operator is obtained from the classical Hamiltonian with the replacement
p → − iℏ∂
xSchrödinger equation
The Schrödinger equation is a first-order differential equation in the time variable : the form of the wave function is sufficient as initial condition t ψ (x, t = t
0)
The K(x
2, t
2, x
1, t
1) ≡ K(2,1) is the Green function of the operator
iℏ∂
t+ ℏ
22m ∂
2x− V(x, t)
Indeed we have for t
2> t
1iℏ∂
t2K(2,1) = − ℏ
22m ∂
2x2K(2,1) + V(2)K(2,1)
But from we have
as
ψ (x
2, t
2) = ∫
+∞
−∞
dx
1K(x
2, t
2; x
1, t
1)ψ (x
1, t
1) K(2,1) → δ(x
2− x
1) t
2→ t
1+If we define K(2,1) = 0 for t
2< t
1we have
[ iℏ∂
t2+ ℏ
22m ∂
2x2− V(x
2, t) ] K(2,1) = iℏδ(x
2− x
1)δ(t
2− t
1)
The superposition principle
We have already pointed out that the superposition principle requires that the differential laws of the new mechanics must be linear
The Schrödinger equation is a linear first-order differential equation in the time variable t
This means that if and are solution of the Schrödinger equation ψ
1ψ
2ψ = a
1ψ
1+ a
2ψ
2Is also a solution
a
1, a
2∈ ℂ
Moreover, if in the state characterised by the function a certain observable takes the value and in the state characterised by the function it takes the value
ψ
1O O
1ψ
2O
2In the state characterised by the wave function the
observable can take either the value or the value O
1ψ = a O
12ψ
1+ a
2ψ
2In the free case V(x, t) = 0 the Schrödinger equation reads
i ℏ ∂
tψ (x, t) = − ℏ 2m
2∂
2xψ (x, t)
i ℏ ∂
tχ
k(t)
χ
k(t) = − ℏ 2m
2∂
2xϕ
k(x)
ϕ
k(x) ≡ E
kThe free-particle
We can look for a solution in factorised form ψ
k(x, t) = χ
k(t)ϕ
k(x) which gives
with E
k= ℏω
k= ℏ
2k
2and real
2m k
Left and right-handed side are now
function of different variables: they must be equal to a constant: let’s call it
Ekand
χ
k(t) = exp{− i / ℏE
kt} ≡ exp{− iω
kt} ϕ
k(x) = exp{i 2mE
kℏ
2x} ≡ exp{ikx}
By using and we obtain and , that is the right dispersion relation for a classical massive particle k = 2π / λ λ = h / p p = ℏk E = p
2
/2m
Recall MMP1: The functions ϕ
k(x) = exp{ikx} form a basis of L
2[ℝ]
The functions ψ
k(x, t) = χ
k(t)ϕ
k(x) = exp{i(kx − ω
kt)} = exp { i
ℏ ( px − E
pt )}
Describe particles with momentum and energy but the wave function is a pure phase ! p = ℏk E = p
2
/2m
The probability is independent on
This means that all values of are equally likely and that the position is completely undetermined
| ψ
k(x, t) |
2= 1 x
x
This is the uncertainty principle !
Knowing ψ (x, t) at t = 0 we can find a general solution of the Schrödinger equation
ψ (x, t) = ∫ dk
2π a(k) ψ
k(x, t) = ∫ dk
2π a(k) exp{i(kx − ω
kt)}
ψ (x,0) = ∫ dk
2π a(k) exp{ikx} a(k) = ∫ dy ψ ( y,0) exp{− iky} = ∫ dy ψ ( y,0) ψ
k* ( y,0)
ψ (x, t) = ∫ dy ∫ dk
2π ψ
k* (y,0) ψ
k(x, t) ψ ( y,0)
Thus we can write
Comparing with the definition of the wave-function in terms of the kernel
ψ (x, t) = ∫ dy K(x, t, y,0) ψ (y,0)
we find
K(x, t, y,0) = ∫ dk
2π ψ
k* ( y,0) ψ
k(x, t)
= ∫ dk
2π e
−ikye
i(kx−ℏk2t/2m)= ∫ dk
2π exp { − iℏt
2m ( k − (x − y)m ℏt )
2
+ im(x − y)
22ℏt }
= m
2πiℏt exp{im(x − y)2
2ℏt }