Β§4.1–4.2The Hamiltonian Operator; Normal Modes of a String

Part II Phillips pp. 59–63 Β· ~15 min read

  • Hamiltonian operator
  • eigenvalue problem
  • eigenfunction
  • separation of variables

Energy is the one observable whose operator also runs the clock. Everything odd about this chapter follows from Δ€ doing two jobs that no other operator does at once.

Chapter 3 handled position and momentum. Chapter 4 does energy β€” and energy turns out to be different in kind, because its operator is not merely another observable. It is also what drives time forward.

The chapter states its destination in its first paragraph, and it is worth holding onto:

the observable properties of a quantum state with a sharply defined energy never change.

4.1 The Hamiltonian Operator

Β§3.5 established that observables are described by operators and gave two of them:

r^=randp^=βˆ’iβ„βˆ‡(4.1)\hat{\mathbf r} = \mathbf r \quad\text{and}\quad \hat{\mathbf p} = -i\hbar\nabla\tag{4.1}

The energy observable gets an operator too β€” the Hamiltonian operator , written H^\hat H. Phillips builds it by assumption, and says so: the relation between the operators for energy, momentum and position is taken to mirror the relation between the classical quantities. Classically E=p2/2m+V(r)E = p^2/2m + V(\mathbf r), so

H^=p^22m+V(r^)(4.2)\hat H = \frac{\hat{\mathbf p}^2}{2m} + V(\hat{\mathbf r})\tag{4.2}

and substituting Eq. (4.1) turns that into something you can actually apply:

H^=βˆ’β„22mβˆ‡2+V(r)(4.3)\hat H = -\frac{\hbar^2}{2m}\nabla^2 + V(\mathbf r)\tag{4.3}

The dual role

Here is the observation the whole chapter turns on. H^\hat H does two different jobs.

First, it is the energy observable. The recipe of Β§3.5 β€” sandwich the operator between Ξ¨βˆ—\Psi^* and Ξ¨\Psi, integrate β€” gives the energy expectation value directly:

⟨E⟩=βˆ«Ξ¨βˆ—(r,t) H^ Ψ(r,t) d3r(4.4)\langle E\rangle = \int \Psi^*(\mathbf r, t)\,\hat H\,\Psi(\mathbf r, t)\,\mathrm d^3\mathbf r\tag{4.4}

Second, it governs time evolution, because the SchrΓΆdinger equation, Eq. (2.17), is nothing more than

iβ„βˆ‚Ξ¨βˆ‚t=H^Ξ¨(4.5)i\hbar\frac{\partial\Psi}{\partial t} = \hat H\Psi\tag{4.5}

Equation (4.5) β€” the most compact statement in the book

symbol
is
how the state changes in the next instant. First order in time, so Ξ¨ now determines Ξ¨ for ever after β€” the evolution is completely deterministic.
units
type
complex

Click any symbol to see what it is, what units it carries, and what kind of object it is once you put it in an array.

That is the fundamental connection between energy and time, and everything in this chapter is a consequence of it.

How we will solve it

Equation (4.5) is a partial differential equation in four variables. Phillips announces the strategy plainly, and notes it is the same one used for the classical wave equation or the diffusion equation: seek a separable solution, then solve an eigenvalue problem.

Rather than introduce that abstractly, he does something better β€” he solves a problem you can already picture.

4.2 Normal Modes of a String

Let Ξ¨(x,t)\Psi(x,t) be the transverse displacement of a stretched string. It obeys the classical wave equation

βˆ‚2Ξ¨βˆ‚x2βˆ’1c2βˆ‚2Ξ¨βˆ‚t2=0(4.6)\frac{\partial^2\Psi}{\partial x^2} - \frac{1}{c^2}\frac{\partial^2\Psi}{\partial t^2} = 0\tag{4.6}

with cc the wave speed, and if the ends are fixed at x=0x = 0 and x=ax = a,

Ξ¨(x,t)=0Β atΒ x=0Β andΒ atΒ x=a,Β forΒ allΒ timeΒ t(4.7)\Psi(x,t) = 0 \text{ at } x = 0 \text{ and at } x = a, \text{ for all time } t\tag{4.7}

Separating the variables

From Eq. (4.6) to two ordinary differential equations

step 1 of 5

This is the whole technique. Watch what forces each step β€” nothing here is a guess.

  1. 1ASSUME a solution in which every point of the string moves with the same time dependence, differing only in how much. That is what a normal mode is, and it is the only assumption in the derivation.

    Equation (4.8). ψ(x) is the shape, T(t) is the common time factor.

The separated equation itself is Eq. (4.9); equating each side to βˆ’Ο‰2-\omega^2 splits it into Eq. (4.10) for the time factor, whose solution T(t)=Acos⁑ωt+Bsin⁑ωtT(t) = A\cos\omega t + B\sin\omega t is Eq. (4.11), and an equation for the spatial shape:

d2ψdx2=βˆ’k2ψ,wherek=Ο‰c(4.12)\frac{\mathrm d^2\psi}{\mathrm dx^2} = -k^2\psi,\quad\text{where}\quad k = \frac{\omega}{c}\tag{4.12} ψ(0)=ψ(a)=0(4.13)\psi(0) = \psi(a) = 0\tag{4.13}

Where quantization comes from

The general solution of Eq. (4.12) is ψ(x)=Mcos⁑kx+Nsin⁑kx\psi(x) = M\cos kx + N\sin kx. Now apply the boundary conditions, and watch what happens:

  • ψ(0)=0\psi(0) = 0 kills the cosine, so M=0M = 0.
  • ψ(a)=0\psi(a) = 0 then requires sin⁑ka=0\sin ka = 0, so kaka must be a whole multiple of Ο€\pi: k=Ο€/a,Β 2Ο€/a,Β 3Ο€/a,…k = \pi/a,\ 2\pi/a,\ 3\pi/a,\dots

kk can no longer take just any value. There are infinitely many solutions, one per integer:

ψn(x)=Nsin⁑knx,withkn=nΟ€a(4.14)\psi_n(x) = N\sin k_nx,\quad\text{with}\quad k_n = \frac{n\pi}{a}\tag{4.14}

for n=1,2,3,…n = 1, 2, 3, \dots, and combining with Eq. (4.11) at Ο‰n=ckn\omega_n = ck_n gives the complete normal mode:

Ξ¨n(x,t)=[Ancos⁑ωnt+Bnsin⁑ωnt]sin⁑knx(4.15)\Psi_n(x,t) = \left[A_n\cos\omega_nt + B_n\sin\omega_nt\right]\sin k_nx\tag{4.15}

A vibrating string β€” one mode keeps its shape, a mixture does not

00.20.40.60.81.0-2-1012position x (units of a)displacement Ξ¨(x, t)

One mode. The curve only scales between the dashed envelope and its mirror image β€” every point shares the single factor cos Ο‰β‚™t, and the nodes never move. This is what makes Ξ¨(x, t) = ψ(x)Β·T(t) possible, and it is the classical shadow of a stationary state.

Released from rest, so Eq. (4.15) has Bβ‚™ = 0 and the sliders are its Aβ‚™. The walls are the boundary condition ψ(0) = ψ(a) = 0 β€” the only reason k cannot take just any value. Every mode here is a sine that fits a whole number of half-wavelengths between them, which is the entire content of kβ‚™ = nΟ€/a.

Figure 4.1 β€” the four shapes

Fig. 4.1 β€” the first four eigenfunctions, with a = 1

00.20.40.60.81.0-1.5-1.0-0.500.51.01.5position xeigenfunctionn = 1
00.20.40.60.81.0-1.5-1.0-0.500.51.01.5position xeigenfunctionn = 2
00.20.40.60.81.0-1.5-1.0-0.500.51.01.5position xeigenfunctionn = 3
00.20.40.60.81.0-1.5-1.0-0.500.51.01.5position xeigenfunctionn = 4

Exactly the book's figure. The peak value is √2 β‰ˆ 1.41 because these are normalized to ∫|ψ|Β² dx = 1 with N = √(2/a) β€” the same constant Β§3.4 derived. Count the interior zeros: n βˆ’ 1 of them, every time. Phillips' caption is worth re-reading: in classical physics these are the shapes of a vibrating string; in quantum physics they are the wave functions of a particle in a box with definite energy. The same curves, twice.

The general motion, and a change of vocabulary

Because Eq. (4.6) is a homogeneous linear partial differential equation, any superposition of normal modes is also a solution β€” and in fact the general motion of the string is exactly that:

Ξ¨(x,t)=βˆ‘n=1,2,3…[Ancos⁑ωnt+Bnsin⁑ωnt]sin⁑knx(4.16)\Psi(x,t) = \sum_{n=1,2,3\dots}\left[A_n\cos\omega_nt + B_n\sin\omega_nt\right]\sin k_nx\tag{4.16}

Given the initial displacement and velocity of every point, ordinary Fourier series techniques find every AnA_n and BnB_n.

Phillips then does something quietly clever: he renames what he has just done. No new mathematics, only quantum vocabulary attached to steps you have already taken.

The same eigenfunctions, one at a time

00.20.40.60.81.0-1.5-1.0-0.500.51.01.5position xψ(x)
  • Re ψ
interior nodes: 0 β€” always n βˆ’ 1

Οˆβ‚(x) = √(2/a)Β·sin(Ο€x/a)

Drag n and count the nodes. This is the widget the rest of the book uses for every bound state β€” the harmonic oscillator of chapter 5 and the hydrogen radial functions of chapter 9 are the same picture with a different potential.

What we didWhat it is called
Solved Eq. (4.12) subject to Eq. (4.13)solved an eigenvalue problem
Found solutions exist only for k=nΟ€/ak = n\pi/afound the eigenvalues knk_n
ψn(x)=Nsin⁑knx\psi_n(x) = N\sin k_nxthe eigenfunction belonging to knk_n
Wrote the general motion as Eq. (4.16)expanded in a superposition of eigenfunctions

What comes next

Everything on this page was classical physics. Not one line of it mentioned ℏ\hbar, and a vibrating string is as concrete an object as physics has.

Β§4.3 now runs this exact procedure on the SchrΓΆdinger equation β€” same separation, same eigenvalue problem, same superposition at the end. Only one thing changes: because Eq. (4.5) is first order in time rather than second, the time factor comes out as eβˆ’iEt/ℏe^{-iEt/\hbar} instead of cos⁑ωt\cos\omega t.

That one difference is the whole chapter.

Check yourself

0 / 7 answered

  1. 1.What is meant by the Hamiltonian's "dual role"?

  2. 2.In the separation of variables, why must both sides of equal a constant?

  3. 3.What actually quantizes for the string?

  4. 4.In Fig. 4.1 the eigenfunctions peak at about 1.41 rather than 1. Why?

  5. 5.A single normal mode keeps its shape as it vibrates. What does that correspond to mathematically?

  6. 6.Discretized, separation of variables turns out to be which linear-algebra operation?

  7. 7.The string's frequencies go as , but the box's energies will go as . Where does the difference come from?