University of California at San Diego – Department of Physics – Prof. John McGreevy Quantum Mechanics C (130C) Winter 2014 Assignment 7 Posted March 3, 2014 Due 11am Thursday March 13, 2014 This is the last problem set. Problem Set 7 1. Stationary phase. [extra credit] P inθ ? Consider both the case when θ is an integer multiple of 2π/N , (a) What is N n=1 e and when it is not. P iϕi where {ϕi } are chosen randomly (b) Add together some random phases, i.e. N i=1 e and uniformly in the interval [0, 2π). (The Mathematica command RandomReal is useful here.) How does the magnitude of the answer depend on N as N grows? (c) Evaluate numerically the integral Z I(N ) = dxeiN V (x) with V (x) = (x2 − 1)2 − 12 x3 for several values of N , and compare to the answer from the stationary-phase approximation. [The range of integration shouldn’t matter too much, but you can take x ∈ (∞, ∞). If you are worried about convergence of the integral, you can define it by Z ∞ dxee I(N ) = i( π 2 −) N V (x) −∞ for arbitrarily small positive epsilon.] 2. Path integral for a free particle Consider the path integral description of the quantum mechanics of a free particle in one dimension. The action is Z tf m S[x] = dt ẋ2 . 2 0 (a) What is the equation of motion 0 = δS ? δx(t) (b) Find the classical solution x(t) with x(t = 0) = x0 and x(t = tf ) = xf . 1 (c) Evaluate the action for the classical solution S[x], and evaluate the stationaryphase approximation to the path integral for the quantum propagator Z U (xf , tf ; x0 , 0) = hxf |U(tf )|x0 i = [dx]eiS[x] ' Usc ≡ eiS[x] . R (d) Derive the Hamiltonian associated to the action S = dtL above. [That is, find p = ∂L and eliminate ẋ in H(x, p) = pẋ − L.] ∂ ẋ (e) Treating this Hamiltonian quantum mechanically, evaluate the exact quantum propagator, U (xf , tf ; x0 , 0) = hxf |U(tf )|x0 i = hxf |e−itf H |x0 i. [One way to do this is to use the fact that U here solves the ODE i∂t U (x, t; x0 , 0) = − ~2 2 ∂ U (x, t; x0 , 0) 2m x with the initial condition U (xf , 0; x0 , 0) = δ(xf − x).] Compare with the semiclassical approximation Usc defined above. 3. A charged particle, classically. [Extra credit] This problem is an exercise in calculus of variations, as well as preparation for our discussion of particles in electromagnetic fields. Consider the following action functional for a particle in three dimensions: Z m e ~ S[x] = dt ~x˙ 2 − eΦ(~x) + ~x˙ · A(x) . 2 c (a) Show that the extremization of this functional gives the equation of motion: e δS[x] = −mẍi (t) − e∂xi Φ(x(t)) + ẋj Fij (x(t)) i δx (t) c where Fij ≡ ∂xi Aj − ∂xj Ai . Show that this is the same as the usual CoulombLorentz force law ~ v ~ ~ + ×B F~ = e E c with Bi ≡ ijk Fjk . (b) Show that the canonical momenta are Πi ≡ ∂L e i = m ẋ + Ai (x). ∂ ẋi c R Here S = dtL. (I call them Π rather than p to emphasize the difference from the ‘mechanical momentum’ mẋ.) Show that the resulting Hamiltonian is 2 X 1 e i i H≡ ẋ Π − L = Πi − Ai (x(t)) + eΦ. 2m c i 2 4. Coherent states. Consider a quantum harmonic oscillator. The creation and annihilation operators a† and a satisfy the algebra [a, a† ] = 1 and the vacuum state |0i satisfies a|0i = 0. Coherent states are eigenstates of the annihilation operator: a|αi = α|αi. (a) Show that |αi = e −|α|2 /2 αa† e −|α|2 /2 |0i = e ∞ X αn √ |ni n! n=0 is an eigenstate of a with eigenvalue α. (a is not hermitian, so its eigenvalues need not be real.) (b) Coherent states with different α are not orthogonal. (a is not hermitian, so its 2 eigenstates need not be orthogonal.) Show that |hα1 |α2 i|2 = e−|α1 −α2 | . (c) Compute the expectation value of the number operator n = a† a in the coherent state |αi. (d) Time evolution acts nicely on coherent states. The hamiltonian is H = ~ω a† a + 21 . Show that a coherent state evolves into a coherent state with an eigenvalue α(t): e−iHt |αi = e−iωt/2 |α(t)i where α(t) = e−iωt α. 5. Two coupled spins. [based on Le Bellac problem 6.5.4] Consider a four-state system consisting of two qbits, H = span{|1 i ⊗ |2 i ≡ |1 2 i, =↑z , ↓z }. (a) For each qbit, define σ ± ≡ 21 (σ x ± iσ y ). (These are raising and lowering operators for σ z : [σ z , σ ± ] = ±2σ ± . Show this.) Show that ~1 · σ ~ 2 = 2 σ1+ σ2− + σ1− σ2+ + σ1z σ2z . σ ~1 · σ ~ 2 on the basis states (b) Determine the action of the operator σ | ↑↑i, | ↑↓i, | ↓↑i, | ↓↓i. (c) Show that the four vectors 1 1 |0, 0i = √ (| ↑↓i − | ↓↑i) , |1, 1i ≡ | ↑↑i, |1, 0i ≡ √ (| ↑↓i + | ↓↑i) , |1, −1i ≡ | ↓↓i 2 2 ~1 · σ ~ 2 with eigenvalues 1 or −3. are orthonormal and are eigenvectors of σ 3 ~ 2 )2 and Jz ≡ σ1z + σ2z and (d) Show that they are also eigenvectors of J2 ≡ (~ σ1 + σ find their eigenvalues. (e) Consider the operator 1 ~1 · σ ~ 2) (1 + σ 2 acting on the two spins. Show that P1,2 acts by exchanging the states of the two spins: P1,2 |1 2 i = |2 1 i . P1,2 ≡ 6. Spin chains and spin waves. [Related to Le Bellac problem 6.5.5 on page 200] A one-dimensional ferromagnet can be represented as a chain of N qbits (spin-1/2 particles) numbered n = 0, ...N − 1, N 1, fixed along a line with a spacing ` between each successive pair. It is convenient to use periodic boundary conditions (as in HW 2 problem 2), where the N th spin is identified with the 0th spin: n + N ≡ n. Suppose that each spin interacts only with its two nearest neighbors, so the Hamiltonian can be written as N −1 1 X 1 ~n · σ ~ n+1 . σ H = N J1 − J 2 2 n=0 where J is a coupling constant determining the strength of the interactions. (a) Show that all eigenvalues E of H are non-negative, and that the minimum energy E0 (the ground state) is obtained in the state where all the spins point in the same direction. A possible choice for the ground state |Φ0 i is then |Φ0 i = | ↑z in=0 ⊗ | ↑z in=1 ⊗ ... ⊗ | ↑z iN −1 ≡ | ↑↑ ... ↑i. (b) Show that any state obtained from |Φ0 i by rotating each of the spins by the same angle is also a possible ground state. P ~ n commutes with the Hamilto[Hint: the generator of spin rotations ~J ≡ n σ nian.] [Cultural remark: the phenomenon of a ground state which does not preserve a symmetry of the Hamiltonian is called spontaneous symmetry breaking. ] (c) Now we wish to find the low-energy excitations above the ground state |Φ0 i. Show that H can be written H = N J1 − J N −1 X Pn,n+1 = J n=0 where N −1 X (1 − Pn,n+1 ) . n=0 1 ~n · σ ~ n+1 ) . (1 + σ 2 Using the result of the previous problem, show that the eigenvectors of H are linear combinations of vectors in which the number of up spins minus the number of down spins is fixed. Let |Ψn i be the state in which the spin n is down with all the other spins up. What is the action of H on |Ψn i? Pn,n+1 ≡ 4 (d) We are going to construct eigenvectors |ks i of H out of linear combinations of the |Ψn i. Let N −1 X |ks i = eiks n` |Ψn i n=0 with 2πs , s = 0, 1, ...N − 1 . N` Show that |ks i is an eigenvector of H and determine the energy eigenvalue Ek . Show that the energy is proportional to ks2 as ks → 0. This state describes an elementary excitation called a spin wave or magnon with wave-vector ks . ks = 7. A damped quantum harmonic oscillator [extra credit] [from Preskill] In this problem we extend the amplitude damping channel which we described for a qbit to a harmonic oscillator. This arises if we couple a harmonic oscillator with creation operator a to an environment via an interaction like X ∆H = λi ab†i + h.c. (1) i Here we have modeled the environment by a collection of oscillators whose modes are created by b†i . For the first part of the problem, let’s instead model the environment by two-state system. Consider a situation where evolution over a time dt is given by p √ U|ni ⊗ |0iE = 1 − p|ni ⊗ |0iE + pa|ni ⊗ |1iE . In this expression we suppose that p 1 and we will work only to leading order in p. (a) Note that a|ni ∝ |n − 1i. Find the proportionality constant. (b) Find the Kraus operators M0 , M1 for the evolution above and show that they satisfy the unitary condition 1 = M0 M†0 + M1 M†1 to leading order in p. (c) Find the evolution of the density matrix. To do better than the above, let’s introduce a little more technology. Suppose we are studying an environment which forgets very rapidly. Then we may take the limit where dt → 0 and write a differential equation for the density operator. For sure we have ρ(dt) = Mµ ρ(0)M†µ = ρ(0) + O(dt) and so√one of the Kraus operators is M0 = 1 + O(dt), and the others are small, of order dt, describing transitions that may occur with probability proportional to dt. Let’s make the ansatz √ Mi = dtLi , i = 1, 2, 3 M0 = 1 + dt (−iH + K) 5 where H is the (hermitian) Hamiltonian for the subsystem (H = ~ω a† a + 12 for the SHO) – this term describes the unitary evolution that would happen if the system were closed – and K is some new Hermitian operator describing the damping. Each of H, Li , K is zeroth order in dt. (d) Determine K in terms of the Li using the unitarity condition. You should find that X 1 † 1 † † ∂t ρ = L[ρ] ≡ −i[H, ρ] + Li ρLi − Li Li ρ − ρLi Li . 2 2 i (2) The superoperator on the RHS is called the Lindbladian. (e) Now we return √ to the situation above in (1). In this case there is one lindblad operator, L1 = Γa. (Think of Γ as the rate for one of the environment oscillators b to decay from the first excited state to the groundstate.) Apply the evolution in (2) to the density matrix, and show that the occupation number of the oscillator n ≡ a† a satisfies ∂t hn(t)i = tra† aρ̇ = −Γhn(t)i . Integrate this equation and show that the oscillator is damped. (f) [Note: It is possible to do this part starting from eqn (2), even if you did not do the previous parts.] Now we will think about decoherence. The fact that the oscillator is coupled to the environment via the operator a suggests that it is eigenstates of a that will be the classical states. This turns out to be correct. As we saw in problem 4, eigenstates of a are coherent states. Suppose we begin in a Schrödinger’s cat state of the form √12 (|α1 i + |α2 i), with |α1 − α2 | 1. The initial density matrix is 1 (|α1 i + |α2 i) (hα1 | + hα2 |) . 2 Show that, as a result of the coupling to the environment above, the off-diagonal terms decay rapidly: ρ(0) = ρ(t) = 1 1 1 2 (|α1 ihα1 | + |α2 ihα2 |) + e− 2 Γt|α1 −α2 | (|α2 ihα1 | + h.c.) . 2 2 So the decoherence rate is Γdec = 21 |α1 − α2 |2 Γ, much faster than the damping rate, if the coherent states are very different. [First hint: It may be useful to write X ? † ρ(0) = βi βj eαi a |0ih0|eαj a ij=1,2 with βi = 1 2 √1 e− 2 |αi | , 2 and to make the ansatz X ? † ρ(t) = βij (t)eαi a |0ih0|eαj a . ij=1,2 6 The claim is that for tΓ 1, βii (t) ∼ |αi |2 , but 1 β12 (t) ∝ e− 2 |α1 −α2 | ] 7 2 Γt α1 α2? .