I am working on transition metal oxides, an antiferromagnetic system. I trying to calculate the activation barrier and transition state of the system. I am using Nudged Elastic Band method implemented in the program Quantum Espresso. Unfortunately, the NEB calculation is not converging. I tried lowering the mixing_beta value and tried adding an intermediate image, but still, it was not converging.

The magnetization values in the reactant and product are different. However, since we cannot specify different starting magnetizations for the reactant and product in the NEB calculation, we have used the same values for both the reactant and the product. I would like to know whether our approach is correct and also would like to know what can be done to achieve convergence of the NEB calculation of such magnetically ordered systems.