8.

04 Quantum Physics

Lecture XII

After the energy measurement

After the energy measurement with outcome Ei , the particle will be in the energy eigenstate ui , and all subsequent energy measurements will yield the energy Ei .

What is the energy before measurement is made?

If (x) is not an eigenstate, the energy is uncertain. A measurement can yield dierent energy values, only probabilities can be predicted. However, an average value of the energy can be calculated: E =
n=1

|cn |2 En

generally valid

(12-1)

Using the denition of the expansion coecients cn , we can also write this as E = = = =
n=1 n=1

|cn |2 En cn cn cn dxu n (x) (x) En

dx (x)En un (x)

n (x) dx (x)Hu = dx (x)H cn un (x)

n

(x) dx (x)H

Expectation value of energy in state (x): (x) E = dx (x)H valid for any potential, not only box potential (12-2)

Hamiltonian operator and energy

If we postulate that a particle of momentum p is associated with a deBroglie wavelength dB = h , then it is represented by a plane wave eikx with a wavevector p Massachusetts Institute of Technology XII-1

8.04 Quantum Physics

Lecture XII

p k = 2dB = 2p = h , or p = hk . Then, since the Fourier transform (k ) of (x) h gives the probability amplitude for the plane wave with wavevector k 1 ikx (x) = dk(k )e , (12-3) 2

the expectation value of momentum is given by p = dkhk |(k )|2 2 p = dk ( hk )2 |(k )|2

(12-4) (12-5)

Note. In PS5 you show using the properties of Fourier transforms that these expectation values can also be expressed as h p = dx (x) (x) (12-6) i x 2 2 h p = dx (x) (x) (12-7) i x It follows that the expectation value of the KE is T = p2 2m = h 2 2 dx (x) 2m x2

(x) (12-8)

How large is the expectation value of the potential energy?

Potential V (x) should be weighted by probability to nd particle between x and x + dx, hence 2 V = dxV (x)| (x)| = dx (x)V (x) (x) (12-9) Since E = V + T it follows that E = = h 2 2 dx (x) + V (x) (x) 2m x2

(12-10) (12-11)

(x) dx (x)H

This is the so-called sandwich form for calculating the mean value (expectation value) of the energy. If (x) is an energy eigenfunction with eigenvalue E0 , i.e. if Massachusetts Institute of Technology XII-2

8.04 Quantum Physics E0 (x) = E0 E0 (x), then H E = =

E0 (x) dxE (x)H 0

Lecture XII

dxE (x)E0 E0 (x) 0 = E0 dx|E0 (x)|2

= E0 ,

(12-12)

where we have used the fact that the wavefunction is normalized. This shows that the constant E0 appearing when we make the product ansatz (x, t) = T (t) (x ) to solve the SE is really the energy of the system. We dene the expectation value O acting on a wavefunction (x) via the sandwich form of an operator O = dx (x)O (x) O Then we have (12-13)

= dx (x)H (x) E = H

(12-14)

The mean energy of the state described by the wave-function (x) is the expectation . We say that the Hamiltonian H is the operator value of the Hamiltonian operator H associated with the measurable quantity energy. The operator T associated with the kinetic energy is 2 2 =h T (12-15) 2m x2 = dx (x)T (x), while the operator V for the potential energy is with T = T simply a multiplicative factor = V (x) V (12-16) = dx (x)V (x) with V = V

Why is potential energy associated with a simple multiplicative factor while kinetic energy is associated with a second derivative?
Because we are working with wavefunctions in real space (x). We say that we are working with wavefunctions in position space or in the position representation. Massachusetts Institute of Technology XII-3

8.04 Quantum Physics

Lecture XII

Another possibility is to work in momentum space (the momentum representation). Then the wavefunction should be the probability amplitude in momentum space, which is just the Fourier transform (p) of (x). Then to calculate the KE we have p2 to weigh 2 for each p with the probability to nd the particle momentum between m p and p + dp: 2 p p2 = dp |(p)|2 (12-17) 2m 2m p2 = dp (p) (p) (12-18) 2m We see that in momentum space the KE operator is simply a multiplicative factor 2 p2 = p T = 2m 2m in momentum space (12-19)

How to calculate the potential energy V (x) in terms of the wavefunctions in momentum space (p)?
Note. In PS5, you have shown that x = dp (p) ih (p) p n n x = dp (p) ih (p) p Consequently, for any potential function V (x) =
n=0

(12-20) (12-21)

a n xn

(12-22)

we can calculate the expectation value V as V = dp (p)V ih (p) p (p) = dp (p)V

(12-23)

Consequently, the representation of the operator for the PE in momentum space is =V V Massachusetts Institute of Technology ih p , (12-24) XII-4

8.04 Quantum Physics

Lecture XII

where a function V of an operator is dened in terms of its Taylor expansion, Eq. (12-22). It follows that the Hamiltonian is =T +V H 2 h 1 = + V (x) 2m i x p2 + V ih = p 2m The SE equation is always the same: ih (x, t) (x, t) = H t (p, t) ih (p, t) = H t (x) = E (x) H (p) = E(p) H time-dependent SE time-dependent SE time-independent SE time-independent SE (12-28) (12-29) (12-30) (12-31) (12-25) in position space in momentum space (12-26) (12-27)

Example. For the harmonic oscillator, the SE (in appropriately chosen units) looks the same in position and momentum space: 1. linear potential V (x) = Ax h 2 (x) + Ax (x) = E (x) 2m p2 (p) + ihA (p) = E(p) 2m in position space (12-32) simpler equation in momentum space (12-33) 2. harmonic oscillator: V (x) = 1 m 2 x2 2 h 2 1 (x) + m 2 x2 (x) = E (x) 2 2m p2 1 2 (p) h m 2 (p) = E(p) 2m 2 (12-34) (12-35)

If we know the solutions in one space, we know the solutions in the other. The HO is symmetric in position and momentum. Massachusetts Institute of Technology XII-5

8.04 Quantum Physics

Lecture XII

Time evolution of the wave function

Consider a particle in the innite box with a wavefunction at t = 0, (x, t = 0). Expansion into eigenstates (x, t = 0) = c1 u1 (x) + c2 u2 (x) + = n=1 cn un (x). Since each eigenstate un (x, t) evolves at a rate given by its eigenenergy En ,
h un (x, t) = un (x, t = 0)eiEn t/ h = un (x)eiEn t/

(12-36) (12-37)

the wavefunction (x, t) at later time t is simply given by the linear superposition (x, t) =
n=1 h cn un (x)eiEn t/

(12-38)

where the expansion coecients cn are calculated at t = 0: cn = dxu n (x)(x, t = 0)

(12-39)

Hence the importance of energy eigenstates and eigenvalues: The eigenvalues represent not only the only possible outcomes of individual energy measurements, but the combination of eigenstates and eigenvalues allows one to write down the time evolution of an arbitrary initial state.

How does a particle move?

Example. (x, t = 0) = and rst excited state.
1 (u1 (x) 2

+ u2 (x)). Particle in equal superposition of ground

1 h (x, t) = u1 (x)eiE1 t/ + u2 (x)eiE2 t/h 2 1 iE1 t/ h = e u1 (x) + u2 (x)ei(E2 E1 )t/h 2 1 h 2 |(x, t)|2 = |u1 (x) + u2 (x)ei(E2 E1 )t/ | 2

(12-40) (12-41) (12-42)

At any xed position, interference term between u1 and u2 oscillates from constructive to destructive interference with angular frequency 21 = E2 E1 h (12-43)

The energy dierence determines the oscillation of the particle between the halves of the box. Massachusetts Institute of Technology XII-6

8.04 Quantum Physics

Lecture XII

Figure I: A particle in a superposition of the ground state an the rst excited state oscillates from left to right at the frequency corresponding to the energy dierence between the two states. Note. If (x, t = 0) is an eigenstate, (x, t = 0) = un (x), then |(x, t)|2 = |(x, 0)|2 , i.e. the probability density does not change in time: Bohrs stationary states are energy eigenstates. An oscillating electron (particle) is in a superposition of at least two energy eigenstates. An electron in a Bohr atom that emits a Lyman photon is in a superposition of the ground (E1 ) and the rst excited state (E2 ). It oscillates in space at the freE1 quency E2 , exactly the frequency of the emitted Lyman photon. h Our box example also shows: The more localized the initial spatial distribution (x, 0), the more eigenstates are involved, and the more complicated the time evolution will be (there will be interference terms oscillating at (E2 E1 )/h , (E3 E1 )/h , (E3 E2 )/h , . . . ) All motion of particles involves oscillating interference.

What is the relation between the SE and CM

QM should reproduce CM as limiting case CM p = mv = m dx dt We expect this (and other) classical equation(s) to hold for the QM expectation values (average position, momentum), at least in some limiting case. Calculate m dx : the only time variation arises from the time variation of wave dt function, x is coordinate, not particle position in the SE. Massachusetts Institute of Technology XII-7

8.04 Quantum Physics

Lecture XII

dx dt

=m

d x dt d =m dx (x, t)x(x, t) dt = m dx x + x t t 2 1 h 2 h 2 2 1 = m dx + V (x) x + x ih 2m x2 ih 2m x2 = h 2i

(12-44) (12-45) (12-46) + V (x) (12-47)

dx

2 2 1 x x 2 2 x x ih

dx [ V (x)x xV (x)] (12-48)

The second term is zero, the rst term can be integrated by parts: 2 x = x x2 x x = x x x Similarly, 2 x 2 = x

(x) x x x x x x

(12-49)

2 x x2

(12-50)

A= x + x x x x x = x x ( ) + 2 x x x x x h 2 2 + V (x) SE 2m x2

(12-51) (12-52)

ih = t

(12-53)

1 For the wavefunction to be normalizable, it has to vanish at faster than . x Consequently, the integral over the rst two terms in A yields zero and we are left

Massachusetts Institute of Technology

XII-8

8.04 Quantum Physics with d h m x = dx2 dt 2i x h = dx (x, t) (x, t) i x = dx (x, t) p(x, t) = p

Lecture XII

(12-54) (12-55) (12-56)

So it follows from the SE that the expectation value of momentum is equal to the particle mass times the rate of change of the expectation value of particle position: m d x = p dt (12-57)

This equation follows from the SE in combination with the position representation of the momentum operator p = h . Does the appearance of 1 mean that momentum is i x i complex (imaginary)? Let us calculate the complex conjugate p of the expectation values of p in some arbitrary state (x, t) . . .

Massachusetts Institute of Technology

XII-9