PHY{396 L. Solutions for homework set #6. Problem 1(a): Think of box's walls as mirrors. Every time the particle bounces o a wall, let us look on its reection instead. The reected image y(t) of the particle's actual trajectory x(t) is described by dy = dx ( 1)# bounces thus far: (S:1) dt dt Thus, before the rst bounce y = x, after the rst bounce but before the second y = 2a x or y = x (depending on which wall the particle bounces o of), after the second bounce but before the third y = x or y = x 2a, etc., etc. At the end of the path, y0 x0 modulo 2a for an even number of reections and y0 x0 modulo 2a for an odd number of reections. Let 0 x0; x0 a; it is easy to see that for any imagined trajectory y(t) from y0 = x0 to y0 = 2na x0 there is precisely one real trajectory x(t) from x0 to x0 that is conned to the box at all times. Therefore 0 Z Zx x0 D[x(t)]box = and since S [x(t)] = 0 +1 X 2naZZx X n= 1 Zt t0 x0 D[y(t)]free (S:2) 0 m x_ 2 dt 2 = S [y(t)]; (S:3) it follows that 0 ZZx Ubox (t0; x0; t0; x0) = D[x(t)]box eiS[x(t)]=h = x0 0 +1 X 2naZZx X D[y(t)]freeeiS[y(t)]=h n= 1 x0 +1 X X = Ufree t0; (2na x0); t0; x0 ; n= 1 1 (S:4) cf. eq. (1), Q:E :D: Problem 1(b): As explained in class, +1 X n= 1 Ufree t0; (na + x0); t0; x0 = +1 X 1 e2i`(x0 x0)=a e 22 ihT`2 =ma2 `= 1 a (S:5) where T = t0 t0. Let us apply this formula to eq. (1) by substituting a 7! 2a and x0 7! x0. Thus, +1 X X 1 e2i`(x0x0)=2a e 22ihT`2 =4ma2 0 0 Ubox (t ; x ; t0; x0) = `= 1 2a +1 2 X 1 = 2a + 2a e+2i`x0=2a + e 2i`x0=2a `=1 e 22 ihT`2 =4ma2 = +1 X `=0 0 ` (x0 ) ` (x ) e e 2i`x0 =2a + e+2i`x0 =2a iTE` =h (S:6) where h `)2 ; E` = (2ma 2 0 (x) r 1 and a ` (x) = r 2 cos `x for ` > 0: a a (S:7) In other words, the evolution kernel (1) describes a quantum particle in a box with Neumann boundary conditions. Problem 1(c): We already mentioned that when the particle bounces o the walls an even number of times 2 y0 = 2na + x0 while for an odd number of bounce-os t0 = 2na x0. Therefore, 0 ZZx 0 (t0 ; x0 ; t0; x0 ) = D[x(t)]box eiS [x(t)]=h ( 1)#bounces Ubox = = x0 +1 X X 2naZZx0 n= 1 +1 X X x0 () D[y(t)]freeeiS[y(t)]=h ()Ufree t0; (2na x0 ); t0; x0 ; n= 1 +1 X X 1 e2i`(x0x0)=2a e 22ihT`2 =4ma2 = () `= 1 2a +1 1 X 0 0 = 2a e+2i`x0=2a e 2i`x0 =2a e 2i`x =2a + e+2i`x =2a `=1 e 22ihT`2 =4ma2 = +1 X `=1 0 `(x0 ) ` (x ) e (S:8) iTE` =h where the energies E` are exactly as in eq. (S.7) | but excluding the E0 = 0 | and ` (x) = r 2 sin `x : a a (S:9) And this is of course precisely the spectrum of the box with Dirichle boundary conditions. Problem 2(a): Let x(t) = xcl (t)+ y(t) where xcl (t) satises the Newton equation mxcl (t) = f (t) m!2xcl(t) as well as the initial and nal conditions xcl (t0) = x0 and xcl (t0) = x0 . Then clearly x(tZZ0 )=x0 D[x(t)] = x(t0 )=x0 3 y(ZtZ0 )=0 D[y(t)] : y(t0 )=0 (S:10) At the same time, Zt 0 1 2 2 2 m! x + fx S [x(t)] = dt 12 mx_ 2 t0 Zt (S:11) 0 = S [xcl(t)] + 0 + dt 21 my_ 2 12 m!2y2 t0 and hence U (t0 ; x0; t0; x0) = = x(tZZ0 )=x0 D[x(t)] eiS[x(t)]=h x(t0 )=x0 eiS[xcl (t)]=h eiS cl =h A(t0 y(tZZ0 )=0 t0 Z im D[y(t)] exp 2h dt (y_ 2 !2 y2) (S:12) t0 y(t0 )=0 t0); eq. (4). Note that the factor A(t0 t) depends only on the elapsed time but not on the external force f (t), the classical trajectory xcl(t) or even the initial and nal points x0 and x0. cf. Problem 2(b): According to eq. (S.12), A(t0 t0) = where y(tZZ0 )=0 D[y(t)] eiS [y(t)]=h 0 (S:13) y(t0 )=0 Zt 0 S0[y(t)] = m2 dt (y_ 2 !2 y2): t0 (S:14) is a quadratic functional of the function y(t). To diagonalize this quadratic functional, we expand y(t) into a Fourier series; in light of the initial and nal conditions y(t0) = y(t0) = 0, 4 we write y(t) = 1 X n=1 yn sin n (t t0) (S:15) where = =(t0 t0). Consequently, 1 0 t0 ) X m ( t 2 2 ! 2 )y 2 ( n S0 = n 4 n=1 (S:16) and therefore A(t0 t0) = J = J = J Y 1Z n=1 1 Z Y dyne n=1 s 1 Y n=1 dyn 1 0 t0 ) X im ( t 2 2 ! 2 )y 2 exp ( n n 4h n=1 im(t0 t0 ) (n2 2 4 h m(t0 !2 )yn2 (S:17) 4ih t0) (n2 2 !2 ) Q where J is the Jacobian of the linear variable transform from D[y(t)] to dyn . Note that both the Jacobian J and the innite product on the last line of eq. (S.17) are badly divergent, so what we really have here is N Y s 4ih A(t0 t0) = Nlim J N 0 !1 n=1 m(t t0 ) (n2 2 ! 2 ) (S:18) for some `partial' Jacobians JN . By construction of the path integral, these Jacobians may depend on the elapsed time t0 t0 and the particle's mass m, but they cannot possibly depend on the potential energy V (x) of the harmonic oscillator. Consequently, for a free particle we would have N Y s 4ih Afree(t0 t0) = Nlim J N 0 !1 n=1 m(t t0 ) (n2 2 0) 5 (S:19) with exactly same Jacobians JN and therefore N Aosc(t0 t0) = Y Afree(t0 t0) n=1 s n2 2 = n2 2 !2 " Y 1 !(t0 t0) n 2 ! 0 = sin!(!t(0 t t t)0) 0 # 1=2 : (6) Please note that this is the correct eq. (6); the printed homework version distributed in class had this eq-n wrong! p On the other hand, for a free particle, A(t0 t0) = m=2ih (t0 t0), hence Aosc(t0 t0) = r m! 2ih sin !(t0 t0) : (7) Q:E :D: Problem 2(c): Without an external force f (t), the classical trajectory of an un-forced oscillator is simply xcl (t) = A cos !(t t0) + B sin !(t t0) (S:20) where the initial and the nal conditions impose 0 x0 cos ! (t0 t0 ) B = x sin !(t0 t0) : A = x0 ; (S:21) Consequently, 0 Zt Scl = dt 12 mx_ 2cl t0 1 m! 2 x2 cl 2 m! = 2 sin !T 2x0x1 + (x20 + x02) cos !T (S:22) where T = t0 t0 and hence 0 x U^ (T ) jx0i = r m! exp 2ih sin !T im! 2x x 2h sin !T 0 1 6 (x20 + x02 ) cos !T : (S:23) Now consider the trace of the evolution operator: Z tr U^ (T ) = dx hxj U^ (T ) jxi r Z m! im! 2 = 2ih sin !T dx exp 2h sin !T 2x (1 cos !T ) s 1 i = = 2(1 cos !T ) 2 sin( 21 !T ) = 1 X n=0 e i(n+ 12 )!T : In light of eq. (8), this trace implies En = h !(n + 12 ), Q:E :D: 7 (S:24)