Ok, so I apply a force to a magnet to accelerate it to speed V within my frame of reference. Then I accelerate another magnet to the same speed. Both magnets have the same mass and react with each other to leave one stationary and the other moving at 2v in my frame of reference.

How does that work?

This could be a valid explanation for the Donald Duck motor generator OU system at
http://www.youtube.com/watch?v=V-MQvzOCNSI

Different frame of reference for the "pushes"

??

...
Is it even possible to define reference frames in electromagnetic terms?
...

It is not possible if you refer to fields. For example if you are stationary between the very large plates of a big capacitor moving along the plane of the plates, you can't see a moving field.
A field is just scalar values representing some space characteristics at every point (Feynman's definition) and when the values are constant it doesn't a matter if the source is moving. It is the same thing with a disk magnet rotating around its magnetic axis: the field is constant everywhere therefore the field doesn't rotate.
It follows that we can't take a field as "reference frame".

Meaningless. What means "fast enough"? How much? What means "passing extra path"? Where is the impulse applied from? Where are the equations? Those in the pdf are wrong or misinterpreted. It has to be honestly rewritten.
...

I do remember my engineering classes.  Units were very important.  Remember Hubble?

Not just English to Metric units; we could always check our answers for gross errors by checking the MLT, (Mass, Length & Time,) characteristics of our answer against what was expected: If our answer was in meters per second, (length over time = L/T,) and we were expecting an L/T^2, (acceleration,) answer, then we knew we made a mistake somewhere in our math because we got velocity, L/T.

I got stuck/lost on page three of the PDF and completely lost interest after that.  They proposed transferring energy using a force, while on the previous page they state momentum as the transfer unit.

mass = M
length = L
time = T
velocity = L/T
acceleration = L/T^2
momentum = mass*velocity = M*L/T
force = mass*acceleration = M*L/T^2
work = energy = force over a distance = M*L/T^2*L = M*L^2/T^2
power = work done over time = M*L^2/T^2/T = M*L^2/T^3

So, instead of using momentum, M*L/T, he uses a force, M*L/T^2, or M*L/T/T.  This is why he needs to state that it is an impact force, he needs to eliminate the "over time" part of the force, (/T), to get back to the momentum transfer.

Then there were all of these extra stipulations or definitions of how the force was always constant everywhere and I started to doze-off.  I wasted about an hour typing a response here and then I deleted it without posting.  I felt like I was getting sucked into another lead-out theory discussion around all the overunity spewing out of a simple pendulum because someone was twisting the basic rules of the game.  I was not up to tearing down the window dressing hung up around the force about how this force shall act and how this force shall be applied, again.

The whole set-up on the top of page three just blew me away.

...
Milkovic's pendulum is a usefull device for some applications but under unity and perfectly explainable with current known physics laws. Nothing more. Or prove it by making it selfrunning instead of providing misleading math; this work would be much more profitable. If we had "12 times more output than input" as stipulated, this should be very very easy...

Yep.