Constraints on the magnitude of alpha in dynamo theory

We consider the backreaction of the magnetic field on the magnetic dynamo coefficients and the role of boundary conditions in interpreting whether numerical evidence for suppression is dynamical. If a uniform field in a periodic box serves as the initial condition for modeling the backreaction on the turbulent EMF, then the magnitude of the turbulent EMF and thus the dynamo coefficient $\a$, have a stringent upper limit that depends on the magnetic Reynolds number $R_M$ to a power of order -1. This is not a dynamic suppression but results just because of the imposed boundary conditions. In contrast, when mean field gradients are allowed within the simulation region, or non-periodic boundary are used, the upper limit is independent of $R_M$ and takes its kinematic value. Thus only for simulations of the latter types could a measured suppression be the result of a dynamic backreaction. This is fundamental for understanding a long-standing controversy surrounding $\alpha$ suppression. Numerical simulations which do not allow any field gradients and invoke periodic boundary conditions appear to show a strong $\alpha$ suppression (e.g. Cattaneo & Hughes 1996). Simulations of accretion discs which allow field gradients and allow free boundary conditions (Brandenburg & Donner 1997) suggest a dynamo $\alpha$ which is not suppressed by a power of $R_M$. Our results are consistent with both types of simulations.