<p>The idea of model predictive control of nonlinear Lipschitz models based on discrete model has been studied by several studies. The mainstream of these research need solving a sum of squares (SOS) problem to find the best region of dominance (RoD) or equivalently the best Lipschitz gain matrix. This eventually leads to maximizing the intersection between the RoD and the region of convergence (RoC). This paper proposes an automatic way for finding the Lipschitz gain matrix by introducing a special structure for this matrix. The proposed method causes simplification of the Lipschitz gain matrix in the LMI and reduces the effect of the Lipschitz gain matrix to a scalar tunable gain. The resulting scalar gain is equivalent to the maximum singular value of the Lipschitz gain matrix. Therefore, the proposed idea reduces the difficulty of finding a Lipschitz gain matrix (by solving SOS problems) to finding the maximum singular value of the matrix. The proposed special structure for the Lipschitz gain matrix has eigenvectors in the direction of the matrix of elliptic RoC. This causes, the resulting Lipschitz gain matrix has the most overlap with the defined RoC. Furthermore, an algorithm is proposed to manually find the optimal gain value (maximum singular value) which maximizes the RoC. Theoretical and simulation results show the significance of the approach whilst the outcome reaches higher computational performance.</p>