The final issue we are going to check is convergence w.r.t. is the size of the displacement. The small displacement method is based on the basic assumption that the forces are proportional to the displacement. This is true if the displacements are small, but eventually the approximation breaks down if one increased the size of the displacement. It would seem advisable to use a very small displacement then, and in principle there should be no limit on how small that can be made. However, if the displacement is very small then also the induced forces are small, and at some point they become so small that they start to be affected by numerical convergence as they are calculated by vasp. There is therefore an ideal range for the size of the displacement: not too large to leave the harmonic limit and not small to introduce numerical errors. We can test this by repeating the calculations with a wide range of displacements and see how this affects the results.
