Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: field transformations



On 04/02/2003 10:02 PM, Joe Heafner wrote:

What I've been looking for is a way to derive this transformation
without using tensors or four vectors. Today, I found a way on pp.
231-232 of Eyges' text _The Classical Electromagnetic Field_ (Dover,
1980). The method involves Lorentz transforming the components of \/ x
E = -(1/c)dB/dt and imposing form invariance. I'm in the process of
carrying out the whole derivation to make sure I understand it.

Please allow me to nit-pick the term "form
invariance". The point is that the expression
E = -(1/c)dB/dt
is not invariant as to form when we change from
one frame to another. Presumably everybody knows
this; it is the topic of this thread.

To say the same thing in other words: The physics
described by the Maxwell equations _collectively_ is
the same in the new frame, but if you consider one
Maxwell equation separately it will go haywire.

E and B are the components (in a given frame) of a higher-
grade object F, the electromagnetic field. Suppose Bob
is moving relative to Alice. Then a field F that is purely
electrical according to Alice will be partly magnetic
according to Bob. The real physics (F) will be the same,
but the decomposition into components (E and B) will be
different.

This is entirely analogous to the behavior of Cartesian
coordinates. Suppose Doug's frame is rotated relative
to Carol's. Then they will disagree as to the x-coordinate
and y-coordinate of any given object. The physics will
be the same, but the decomposition into components will
be different.

Something is "invariant" if it doesn't change. We don't
expect components to be invariant. They might be covariant
or contravariant, but not invariant.

I agree what geometric methods are probably the most elegant, but I
don't understand them. I'm just now beginning to think I understand
tensors.

Every bivector can be represented as a tensor, so if
you understand tensors you get bivectors for free.

At the next level of detail: every bivector can be
represented as an antisymmetric 2nd-rank tensor, and
vice versa. So bivectors are equivalent to a proper
subset of tensors, namely the antisymmetric ones.

The tensors that correspond to bivectors are by far
the easiest tensors to visualize. Just as you can represent
a vector using a stick (with a direction marked on it),
you can represent a bivector using a piece of cardboard
(with a direction of circulation marked on it). You
can make models and wave them around.

Thinking of F (as opposed to E and/or B) can demystify
some things that would otherwise be very mysterious.
http://www.monmouth.com/~jsd/physics/pierre-puzzle.htm


> What I've been looking for is a way to derive this transformation
> without using tensors or four vectors.

I still don't understand why this would be considered
desirable.

I vividly remember my first physics TA saying "our goal
is not to teach you to do Lorentz transformations; the
goal is to _avoid_ doing Lorentz transformations." The
point is that more-or-less any real-world problem worth
doing can be done most easily using four-vectors. Were
not talking about elegance for the sake of elegance, or
elegance at the expense of simplicity. Were talking
about just plain easier. Better results with less effort.

To me, Lorentz transformations are just mathematics,
whereas four-vectors are real physics. I can visualize
four-vectors. I can make models. I can make diagrams.

For example, if you make a space-time diagram of the
infamous traveling twins, there's no paradox. If
you just write out the mathematics
a) students are quite likely to get the math wrong, and
b) even if the math is right it's mysterious.

http://www.monmouth.com/~jsd/physics/twins.htm

Matter & Interactions, presents a very nice discussion of
reference frames and fields. The authors specifically demonstrate how a
static E field in one inertial frame is seen as a combined E and B
field in another inertial frame. They do so using the Lorentz
transformation of E and B, but they do not present an actual derivation
of the transformation. They don't present a derivation of the standard
Lorentz transformation either.

I worry that such an approach might tend to promote
rote learning without imparting any real feel for
the physics.