I see it that way, with three coils. Two small end coils, and a larger one in the middle, as shown in the picture published by LM in previous page (EDIT: and by DeltaV just above my present post).
Let's have an electric current running in the same direction inside the two smaller coils, like a Helmholtz coil configuration in a classic magnetic mirror machine. The magnetic field is uniform in the middle. But there is a magnetic pressure gradient Pm = B^2 / 2µ0 that doesn't confine the plasma well in such a system. That magnetic pressure even completely blows the electric discharges away under low pressure air.
Then,
the trick: Add this larger coil in between the two coils, and run the current inside it the opposite way. The magnetic field is then minium at the center, and the magnetic pressure gradient makes a potential well directed inwards, confining the plasma in the middle. The magnetic structure:
This 3-coil confinement system has been published for the first time in peer-reviewed literature in 1977, as a way to confine a plasma near a wall, despite the magnetic pressure, which has a tendency to blow electric glow discharges away from the device. At that time it was intended not as a fusion reactor, but as an air-breathing magnetoplasmadynamic accelerator (MHD aerodyne). The confinement configuration is similar though.
The original paper:
New MHD converters: induction machines
In 2009 I copublished with that same author a paper detailing the non-linear confinement behavior due to magnetic pressure, explaining it with the scalar electrical conductivity σs and the Hall parameter β.[1] The paper was entitled
Non-Equilibrium Plasma Instabilities.
The concept was further detailed with experimental results confirming this theoretical point of view in 2012 in the paper
Wall confinement technique by magnetic gradient inversion. (Actually those experiments had been already done in 1975 but not filmed nor peer-reviewed at that time. Mistake fixed 37 years laters… better late than never)
I think LM reuses this idea for their reactor. They just got rid of the walls and the technique involving eddy currents for MHD acceleration… and kept the confinement coils, inside a hollow cylinder.
The magnetic field lines below (just remove the walls of the aerodyne where the plasma is confined in this picture, and then the plasma would be confined in the middle of the device (in the bottom of this half axis-cut picture):
[1] If you want to understand how this works, check the article about the
electrothermal instability in magnetohydrodynamics that I wrote in Wikipedia in 2008. Is sums up the calculation about this effect.