<div dir="ltr">Dear cp2k users, I'm trying to run a MD simulation with NPT_I ensemble of liquid water with BLYP XC functional and dispersion. Without XC functional smoothing there's a small energy drift, but the density is very close to well established values(~1.04). But when I added smoothing, although the conserved quantity is constant, the density then changed to a substantially value(~1.2). The only difference between the 2 runs is the second one has the smoothing section: &XC_GRID XC_SMOOTH_RHO NN10 XC_DERIV SPLINE2_SMOOTH &END XC_GRID I'm wondering if the smoothing I used was inappropriate. I also attached the input file with smoothing. Thanks very much for any suggestions! Best, Yifei </div>