Vous êtes sur la page 1sur 9

Physics 506 Homework Assignment #8 Solutions Textbook problems: Ch. 11: 11.5, 11.13, 11.14, 11.

18

Winter 2008

11.5 A coordinate system K moves with a velocity v relative to another system K . In K a particle has a velocity u and an acceleration a . Find the Lorentz transformation law for accelerations, and show that in the system K the components of acceleration parallel and perpendicular to v are (1 v 2 /c2 )3/2 a (1 + v u /c2 )3 (1 v 2 /c2 ) v a = a + 2 (a u ) 2 3 (1 + v u /c ) c a = Instead of working directly with perpendicular and parallel components, we may start with a particular boost in the x-t direction, and then generalize our results. We thus take a boost of the form x0 = (x0 + x ), x = (x + x0 ), y=y, z=z (1)

Note that = 1/ 1 2 and = v/c are constants specifying the Lorentz boost. In frame K , the path of a particle is specied by the vector function x(x0 ), while in frame K this is instead x (x0 ). Three-velocities and 3-accelerations are then dened in a frame dependent manner frame K : frame K : x , x0 x u =c 0 , x u=c a=c u x0 u a =c 0 x

To transform between the two frames, we need not just the transformation of the 3-vectors, but also the transformation relating times x0 and x0 . Noting from (1) that a particle following a path x (x0 ) yields a time relation x0 = (x0 + x (x0 )) we may write dx0 = (1 + ux /c) dx0 The inverse relation is simply dx0 1 = 0 dx (1 + ux /c)

This useful expression is basically all we need. We start with velocities ux = c and uy = c dx dx0 dx c d u + c = c = (x + x0 ) = x 0 0 0 0 dx dx dx (1 + ux /c) dx 1 + ux /c uy dx0 dy c dy = c = (uy /c) = 0 0 0 dx dx dx (1 + ux /c) (1 + ux /c) (2)

(3)

Writing ux = u , and noting that the x direction is the parallel direction while the y direction is the perpendicular direction, it is easy to see that these velocity transformations may be written as u = u + c 1 + u /c , u = u (1 + u /c)

We now go on to accelerations. From (2), we have ax = c dux c d ux + c = 0 dx (1 + ux /c) dx0 1 + ux /c c (1 + ux /c)(ax /c) (ux + c )(ax /c2 ) = (1 + ux /c) (1 + ux /c)2 ax (1 2 )ax = 3 = 3 (1 + ux /c) (1 + ux /c)3

And from (3) we have ay = c uy duy c d = dx0 (1 + ux /c) dx0 (1 + ux /c) (1 + ux /c)(ay /c) uy (ax /c2 ) c = 2 (1 + ux /c) (1 + ux /c)2 ay + (ux ay uy ax )/c = 2 (1 + ux /c)3

(4)

It is straightforward to convert the expression for ax into one for a . The result is a a = 3 (1 + u /c)3 For the perpendicular direction, we have to be a bit more clever. Noting that x components in (4) are related to ( ), while y components are directly related to the direction, we have a = a + a ( u ) u ( a )/c 2 (1 + u /c)3

Use of the BAC CAB rule nally gives a = a + (a u )/c 2 (1 + u /c)3

11.13 An innitely long straight wire of negligible cross-sectional area is at rest and has a uniform linear charge density q0 in the inertial frame K . The frame K (and the wire) move with a velocity v parallel to the direction of the wire with respect to the laboratory frame K . a) Write down the electric and magnetic elds in cylindrical coordinates in the rest frame of the wire. Using the Lorentz transformation properties of the elds, nd the components of the electric and magnetic elds in the laboratory. The K frame is the rest frame of the wire, while the K frame may be considered as the lab frame. Starting in the K rest frame, we take the wire to be oriented along the z direction. Since the charges are at rest in K , there is no magnetic eld. The electric eld is given by a simple application of Gauss law. Thus (in cylindrical coordinates, and with Gaussian units) E = 2q0 , B =0

We now transform to the lab frame K using a boost along the z axis = (v/c) z. In general, the electric and magnetic elds are related by 2 ( E ) +1 2 B = (B + E ) ( B ) +1 E = (E B ) In this case, since B = 0 and E = 0 we have simply 2q0 2vq0 B = E = c E = E = Finally, noting that the direction is transverse to the boost, we have the simple relation = (ie there is no length contraction in the transverse direction). Thus we nd 2q0 2vq0 E= , B= (5) c b) What are the charge and current densities associated with the wire in its rest frame? In the laboratory?

In the rest frame, and in cylindrical coordinates, the charge density may be expressed as ( ) = q0 2 Since the 3-current vanishes, the 4-current density may be written as J

cq0

( ) ,0 2

Boosting back to the lab frame K gives J = (J 0 , J 0 ) = cq0 () ( ) , vq0 z 2 2

where we once again used the fact that = . The explicit charge and 3-current densities are () () = q0 , J = vq0 z (6) 2 2 c) From the laboratory charge and current densities, calculate directly the electric and magnetic elds in the laboratory. Compare with the results of part a. From (6), the wire in the lab frame has linear charge density q0 and carries a current I = vq0 in the z direction. Gauss law and Amp` eres law then gives E= 2q0 , B= 2vq0 c

in agreement with the result (5) of part a. 11.14 a) Express the Lorentz scalars F F , F F and F F in terms of E and B . Are there any other invariants quadratic in the eld strengths E and B ? Recall that the eld-strength tensors are given by 0 Ex = Ey Ez Ex 0 Bz By Ey Bz 0 Bx Ez By , B x 0 0 Bx = By Bz Bx 0 Ez Ey By Ez 0 Ex Bz Ey Ex 0

while their raised index counterparts are 0 E = x Ey Ez Ex 0 Bz By Ey B z 0 Bx Ez By , Bx 0 0 B = x By Bz B x 0 Ez Ey By Ez 0 E x B z Ey Ex 0

As a result, we nd the scalar contractions to be F F = F F = 2(E 2 B 2 ) F F = 4E B No other independent scalar invariants can be formed that are quadratic in the eld strengths. One way to motivate this is to note that a scalar contraction may be written as s = t F F where t is a rank-4 tensor that can only be created out of Lorentz invariants such as the metric and the antisymmetric tensor . The only two independent possibilities that respect the antisymmetry of the eld strengths are
1 2 [

],

which give rise to the F F and F F invariants. Note, furthermore, that


1 2 1 = 2 [ ]

which explains why F F = F F . b) Is it possible to have an electromagnetic eld that appears as a purely electric eld in one inertial frame and as a purely magnetic eld in some other inertial frame? What are the criteria imposed on E and B such that there is an inertial frame in which there is no electric eld? It is impossible to have an electromagnetic eld that is pure electric in one frame and pure magnetic in another. To see this, we note that the scalar invariant F F = 2(E 2 B 2 ) must be the same in all frames. In particular, a purely electric eld would have F F < 0, while a purely magnetic eld would have F F > 0. This cannot happen, since the invariant cannot have dierent signs in dierent frames. If there is an inertial frame in which there is no electric eld, we would have F F = 2(E 2 B 2 ) 0 and F F = 4E B = 0

where saturation of the inequality corresponds to both E and B vanishing. Hence E and B must satisfy the requirements E2 < B2, EB =0

in any inertial frame. Along with the trivial case (E = B = 0), these are the criteria on the elds such that there is an inertial frame in which the electric eld vanishes.

c) For macroscopic media, E , B form the eld tensor F and D, H the tensor G . What further invariants can be formed? What are their explicit expressions in terms of the 3-vector elds? The macroscopic eld-strength tensor G is given by 0 Dx = Dy Dz Dx 0 Hz Hy Dy Hz 0 Hx Dz Hy Hx 0

This allows us to form the quadratic in G invariants G G = G G = 2(D2 H 2 ) G G = 4D H along with the mixed invariants F G = F G = 2(E D B H ) F G = F G = 2(E H + B D) 11.18 The electric and magnetic elds of a particle of charge q moving in a straight line with speed v = c, given by (11.152), become more and more concentrated as 1, as is indicated in Fig. 11.9. Choose axes so that the charge moves along the z axis in the positive direction, passing the origin at t = 0. Let the spatial coordinates of the observation point be (x, y, z ) and dene the transverse vector r , with components x and y . Consider the elds and the source in the limit of = 1. a) Show that the elds can be written as E = 2q r 2 (ct z ); r B = 2q v r (ct z ) 2 r

where v is a unit vector in the direction of the particles velocity. We take the particle motion to be along the z direction, so that = (v/c) z . In this case, the elds in the rest frame are given by E = qr , r3 B =0

Boosting this back to the lab frame gives E = E 2 ( E ), +1 B =E (7)

In particular, the latter expression indicates that once we compute the electric eld E , the magnetic eld follows from a simple cross product with the velocity. To proceed, we rewrite r in terms of unprimed (lab) quantities. This may be done by considering the explicit boost transformation x = x, This gives r = r + (z vt) z so that E = q (r + (z vt) z) 2 2 2 (r + (z vt) )3/2 y = y, z = (z vt)

Boosting back using (7) gives the lab frame electric eld E= q (r + (z vt) z) 2 2 2 (r + (z vt) )3/2

which is an exact expression, valid for any velocity . We are interested, however, in the limit 1. This is equivalent to taking the limit . Here we may use the fact that the denominator gets highly peaked, and approaches a delta function in this limit. More precisely,
(A2

lim

1 2 = 2 lim = 2 (B ) 2 2 3 / 2 2 2 3 / 2 A (1 + B ) A +B )

for any B and non-zero A. Taking A = r and B = z vt then gives the result E= 2qr 2q (r + (z vt) z) (z vt) = 2 (z vt) 2 r r

Since we have taken the limit v c, we nally arrive at E = 2q r 2 (z ct), r B = E = 2q v r (z ct) 2 r

b) Show by substitution into the Maxwell equations that these elds are consistent with a 4-vector source density J = qcv (2) (r ) (ct z ) where the 4-vector v = (1, v ). Maxwells equations are F = (4/c)J . In particular, the time component gives r 4 0 J = E = 2q (z ct) 2 c r

Using Gauss theorem in two dimensions, it is easy to see that Hence we obtain J 0 = cq (2) (r ) (z ct) In the space directions, we have r 4 1 E r + B = 2q 2 (z ct) + 2q z J = (z ct) 2 c c t r r r r z (z ct) = 2q 2 (z ct) + 2q z 2 r r r r + 2q z ( z ) (z ct) 2 2 r r = 4q z (2) (r ) (z ct) (Note that z = z .) Noting that v =z , this is equivalent to J = cq v (2) (r ) (z ct) Putting together the time and space components gives J = cq (1, v ) (2) (r ) (z ct) c) Show that the elds of part a are derivable from either of the following 4-vector potentials A0 = Az = 2q (ct z ) ln(r ); A = 0 or A0 = 0 = Az ; A = 2q (ct z ) ln(r ) where is an irrelevant parameter setting the scale of the logarithm. Show that the two potentials dier by a gauge transformation and nd the gauge function, . For the rst case (A0 = Az non-vanishing) we have E = A0 2q 1 A = 2q [ (z ct) ln(r )] + z [ (z ct) ln(r )] c t c t = 2q (z ct) ln(r ) + 2q z (z ct) ln(r ) 2q z (z ct) ln(r ) = 2q (z ct) r 2 r r 2 r = 2 (2) (r )

and B = A = 2q [ z (z ct) ln(r )] = 2q [ (z ct) ln(r ) z] v r r (z ct) = 2q (z ct) = 2q 2 z 2 r r For the second case (A non-vanishing) we have A = 2q (ct z ) ln(r ) = 2q (ct z ) This gives E= and B = A = 2q (ct z ) = 2q (ct z ) z = 2q r 1 A = 2q (ct z ) 2 c t r r 2 r r 2 r r 2 r

r 2 2q (ct z ) r

v r (ct z ) 2 r

2 = ln(r ) is curl-free). (since the function r /r

The above two vector potentials are necessarily related by a gauge transformation through A (1) A(2) = To nd the explicit gauge transformation, we write
= A (1) A(2)

= 2q (ct z ) ln(r ), (ct z ) ln(r ) (ct z ) ln(r ) z (ct z ) ln(r ), (ct z ) ln(r )+ z (ct z ) ln(r ) ct z = 2 q , (ct z ) ln(r ) x0 = 2q We thus conclude that the gauge function is = 2q (ct z ) ln(r )