Since the equilibration after field changes is obtained in part by the motion of domain walls, we should be able to inhibit equilibration by restricting the domain wall motion. "neel" tries to accomplish this by introducing a random distribution of local fields at the spin sites with rms value sigma. These "impurities" help pin the spins at various sites to particular orientations and prevent motion of the domain walls. The result is to produce hysteresis in the M(H) curves.
Run this preset at zero field to let the domain structure become established. With the program running, click the slot of the field slider to a field of 2 or 3 and then back to zero. Now what is the magnetization? Do the same in the opposite direction. Now what is the magnetization at zero field?
A repetition at T = 2.5 shows much less hysteresis. The added thermal fluctuations help depin the domain walls and allow approach to a more nearly equilibrium state.