It was 6-8 years ago, sometime around SMOT that some of the FE experimenters were looking at how you could change apparent PM pole polarity by using a ferrite interconnection. If you did not see this or are not familiar with what they were doing I looked up some of my notes on how I tested what they were talking about, which I cover in short below. The reason I bring this up is that I wonder why with all the claims and continued work with PM devices that no one has made use of this as a switching method. The low energy it takes to do the pole switch is far below other methods that seem to be in use.
1) Take two bar or block PM's and one ferrite bar about 6cm long, can be round, square or rectangular. 2) I used a Compass but a Hall device will work. Determine the poles and mark each PM so you can keep track of how they are oriented. 3) Place the ferrite on the S pole of one PM and place the ferrite pointing at the detector just close enough for the detector to show the correct deflection. 4) Now it is evident that if you now slip the second PM (attractive mode with the first PM) onto the ferrite at the detector end that the detector will still indicate the same deflection (polarity). 5) Now remove the second PM and reverse it so that it is in opposition to the PM still attached to the ferrite. Without the ferrite you would not be able to of course get the two PM's to hold together. But with the ferrite you will be able to stick the opposing PM to the ferrite and the indicator will indicate the reverse polarity as in the above first steps. You can now move the PM ferrite combination to see that both exposed PM faces are at the same polarity. You can configure so that you have a SS or a NN on the faces, with a single balancing pole in the center of the ferrite. Now some additional manual checks will easily show that it takes a minimal amount of energy to move the PM's into an opposing orientation with the ferrite in the center, far less that it would take to move the two PM's together in NN or SS without the ferrite. I always thought this might indeed be a simple way (low force requirement) to change a field in a switching operation as indicated in some of the devices being talked about today.