Force convergence using PBE+TS

I am trying to optimize a graphene sheet on Cu(111) using QE v. 6.3, optimized norm-conserving Vanderbilt pseudopotentials from D. R. Hamann and PBE+TS - see attached input and output. In the output it is written that the BFGS optimization converged, but taking a closer look at the forces, this is not the case. The force criterion is 1.0E-03 Ry/Bohr, but atom 4 still has a force component of -0.00164727 (atom 1 and 2 are kept fixed). The same thing happens when using a slightly lower ecutwfc of 65 Ry. Below 65 Ry the BFGS cycle won't converge at all, but the forces instead oscillate between two values.

Are there any known problems with the forces using the TS vdW corrections?

I noticed issue #71 (closed): Tkatchenko-Scheffler does not work with variable cell optimization However, testing the proposed commit did not solve my issue.

Thanks, Mie Andersen Technical University of Munich

log

pw.inp