Hi Rad. The functional BECKE88_LR can become numerically instable in certain cases. Try to increase the density cutoffs in the XC section (see for example the regtests for MCY3): DENSITY_CUTOFF 1.0E-8 GRADIENT_CUTOFF 1.0E-8 Furthermore, when running GAPW, I suggest to use the FULL_ALL preconditioner in the OT section. Hope this helps... Manuel