I would be very grateful if somebody could tell whether the pressure tensor is evaluated in single or double precision? In the printout there are eight significant figures that perhaps only single precision is used.
Double precision. If you’re talking about the thermo output, you can use thermo_modify to specify a C-style format string for your output and get it to full precision.
perhaps this is a good point to remark that "precision != accuracy".
systematic errors on the pressure tensor are typically much larger
than on forces.