the simulations i did seem to confirm the B^4 scaling law. Controlling the magnetic fields strength was by far the most effective way of controlling the fusion rate.
However, a few caveats, as Dan already pointed out:
* the biggest obstacle my simulations revealed is getting the electrons in through the cusps. like trying to get a bullseye through the eye of a tornado. and of course the higher the field strengh, the harder it is. and like the power scaling law, this scales non-linearly (though i don't know if it's b^2 or b^3 or what.)
* increasing the B field strength also impacts the average velocity (or KE, rather) through the core. and there's a KE (albeit very high), beyond which the fusion rate starts dropping off. While this'll probably be more than compensated for by the increased density, it remains true that after this point the scaling is going to be a little under B^4. Though not that bad, really. http://s155.photobucket.com/user/MSimon ... 1.jpg.html
In fact, before that point you're really getting a boost over basic B^4 scaling because your cross-section is increasing pretty quickly. (unless that was already included in? i don't think so, but i don't know.)
* and finally, my simulation didn't model things like Bremsstruhlung loses, etc. only inertial and electromagnetic. fusion rate was approximated based on density and KE. So finer-scale / more subtle things like that, I can't speak to.
But yeah, these caveats aside, the two big factors are size and field strength, and field strength is the much bigger factor. though increasing size means you can increase the field strength more.