Optimal correction of the intrinsic error field (EF) by the correction coils in MAST is numerically studied, based on linear, full MHD plasma response computed in full toroidal geometry. Various optimisation criteria are proposed, and the results are compared with empirical optima from representative error field correction (EFC) experiments. The two best criteria are thus identified, one aiming at minimisation of the net toroidal resonant electromagnetic torque produced on the plasma column by the EF, the other corresponds to the full cancellation of the 2/1 resonant field harmonic at the q=2 surface, including the plasma response. Neither the vacuum field based criterion nor the singular value decomposition (including plasma response) based criteria produce satisfactory predictions for the EFC in MAST.
