To add comments or start new threads please go to the full version of: Galilean transformation
PhysForum Science, Physics and Technology Discussion Forums > General Sci-Tech Discussions > Homework Help

plasma
1) Show that the electromagnetic wave equation,
d^2(phi)/dx^2 + d^2(phi)/dy^2 + d^2(phi)/dz^2 (1/c^2)( d^2(phi)/dt^2) = 0
is not invariant under Galilean transformation.
Note: here d is a partial differential operator.

I have the solution but I couldnt understand one particular step. The solution is as follows:
The equation will be invariant if it retains the same form when expressed in terms of the new variables x,y,z,t. From Galilean transformation we have,
dx/dx=1, dx/dt=-v, dt/dt=dy/dy=dz/dz=1, dx/dy= dx/dz= dy/dx= dt/dx=0
From chain rule and using the above results we have,
d(phi)/dx= [d(phi)/dx][dx/dx] + [d(phi)/dy][dy/dx] + [d(phi)/dz][dz/dx] + d(phi)/dt][dt/dx] = d(phi)/dx
And,
d^2(phi)/dx^2= d^2(phi)/dx^2
Similarly,
d^2(phi)/dy^2= d^2(phi)/dy^2 &
d^2(phi)/dz^2= d^2(phi)/dz^2

Moreover,
d(phi)/dt= -v[d(phi)/dx] + d(phi)/dt
Differentiating the above equation with respect to t ,
d^2(phi)/dt^2 = d^2(phi)/dt^2 -2v[d^2(phi)/dxdt] + v^2[d^2(phi)/dx^2]
This is where I have a doubt. I differentiated in the following way:

d^2(phi)/dt^2= -v[d^2(phi)/dxdt] - [d(phi)/dx][dv/dt] + [d^2(phi)/dt^2]
= -v[(d^2(phi)/dx^2)(dx/dt)] + [d^2(phi)/dt^2] ------> {dv/dt=0}
= (v^2)[ d^2(phi)/dx^2] + [d^2(phi)/dt^2]
I am able derive 2 of the terms but how to derive the third term -2v[d^2(phi)/dxdt] . Could somebody please help me with this derivation?
mr_homm
I think you tried to shortcut the process too much toward the end. When you went through the full process earlier, like
QUOTE
d(phi)/dx= [d(phi)/dx'][dx'/dx] + [d(phi)/dy'][dy'/dx] + [d(phi)/dz'][dz'/dx] + d(phi)/dt'][dt'/dx]

that was the right way to go.

If you apply the same process to v[d(phi)/dx'] + d(phi)/dt', the first term gives d(v[d(phi)/dx'])/dt =
[d(v[d(phi)/dx'])/dx'][dx'/dt] + [d(v[d(phi)/dx'])/dy'][dy'/dt] + [d(v[d(phi)/dx'])/dz'][dz'/dt] + d(v[d(phi)/dx'])/dt'][dt'/dt] =
v[d^2(phi)/dx'^2]v + v[d^2(phi)/dx'dt']1.

The second term gives something similar, and both terms turn out to contribute one copy of v[d^2(phi)/dx'dt'], which gives you what you want.

For problems like this, there is generally an easier way to do the calculation. Just factor out the original function and look at what the derivative operator does. d/dx'(phi) = d/dx(phi), so d/dx' = d/dx. Similarly, d/dy' = d/dy and d/dz' = d/dz. However, d/dt(phi) = -v[d(phi)/dx'] + d(phi)/dt'. Therefore, d/dt = v*d/dx' + d/dt'.

Now usually you have to be very careful combining differential operators, because the operator on the left will differentiate any functions that appear in the operator on the right. But in this case, the only function in the d/dt operator is v, which is a constant.

Therefore, you can do it like this:
d^2/dx^2 =
(d/dx)(d/dx) = (d/dx')(d/dx') = d^2/dx'^2,
and likewise,
d^2/dy^2 = d^2/dy'^2
and
d^2/dz^2 = d^2/dz'^2.
But
d^2/dt^2 =
(d/dt)(d/dt) =
(v*d/dx' + d/dt')(v*d/dx' + d/dt') =
v^2*(d/dx'^2) + 2v*(d/dx')(d/dt') + (d/dt')^2 =
v^2*d^2/dx'^2 + 2v*d^2/dx'dt' + d^2/dt'^2.

Now you just put the (phi) back in everywhere.

Hope this helps!

--Stuart Anderson
plasma
I tried your first method. But I got 4 extra terms other than d^2(phi)/dt^2 -2v[d^2(phi)/dxdt] + v^2[d^2(phi)/dx^2]
For the first term,
-d(v[d(phi)/dx'])/dt = v^2[d^2(phi)/dx^2] v[d^2(phi)/dxdy][dy/dt] v[d^2(phi)dxdz][dz/dt] v[d^2(phi)/dxdt]

For the second term,
(d/dt)[d(phi)/dt'] = -v[d^2(phi)/dxdt] + [d^2(phi)/dydt][dy/dt] + [d^2(phi)/dzdt][dz/dt] + d^2(phi)/dt^2
How to eliminate the terms containing dy/dt & dz/dt ?
turin
From the derivatives that you gave to define the Galilean transformation, it looks like you are boosting in the x-direction. You forgot to list the time derivatives of y' and z'. They should be part of the Galilean transformation, and they should vanish if you are boosting in the x-direction.
plasma
QUOTE (turin+Mar 19 2007, 04:59 AM)
From the derivatives that you gave to define the Galilean transformation, it looks like you are boosting in the x-direction.  You forgot to list the time derivatives of y' and z'.  They should be part of the Galilean transformation, and they should vanish if you are boosting in the x-direction.

But the electromagnetic wave equation
d^2(phi)/dx^2 + d^2(phi)/dy^2 + d^2(phi)/dz^2 (1/c^2)( d^2(phi)/dt^2) = 0
is defined for 3 dimensions, isn't it? So can we take the velocity in the y and z direction as zero?
turin
If by "the velocity" you mean "the boost velocity", then you are free to choose any direction you like. Since your wave equation is isotropic, you only need to choose one direction, and the isotropy guaruntees the same result for all directions (ah, the beauty of symmetry). And, this idea can be carried a step further in that, you are free to choose the x-direction as your boost direction. So, in short, you may set dy/dt and dz/dt to zero in your Galilean transformation.
PhysOrg scientific forums are totally dedicated to science, physics, and technology. Besides topical forums such as nanotechnology, quantum physics, silicon and III-V technology, applied physics, materials, space and others, you can also join our news and publications discussions. We also provide an off-topic forum category. If you need specific help on a scientific problem or have a question related to physics or technology, visit the PhysOrg Forums. Here youll find experts from various fields online every day.
To quit out of "lo-fi" mode and return to the regular forums, please click here.