[gmx-users] Grompp checkpoint bug?

[Dries Van Rompaey]

Hi everyone,

I’m experiencing some discrepancies when I pass the checkpoint file to
grompp using the -t checkpoint flag:

grompp -f md.mdp -o md.tpr -p topol.top -c md_res_1.gro -t md_res_1.cpt

This command results in a tpr file without any variables for barostat and
thermostat:
box (3x3):
   box[    0]={ 6.70916e+00,  0.00000e+00,  0.00000e+00}
   box[    1]={ 0.00000e+00,  6.70916e+00,  0.00000e+00}
   box[    2]={ 3.35458e+00,  3.35458e+00,  4.74410e+00}
box_rel (3x3):
   box_rel[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   box_rel[    1]={ 0.00000e+00,  1.00000e+00,  0.00000e+00}
   box_rel[    2]={ 5.00000e-01,  5.00000e-01,  7.07108e-01}
boxv (3x3):
   boxv[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   boxv[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   boxv[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
pres_prev (3x3):
   pres_prev[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   pres_prev[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   pres_prev[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
svir_prev (3x3):
   svir_prev[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   svir_prev[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   svir_prev[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
fvir_prev (3x3):
   fvir_prev[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   fvir_prev[    1]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   fvir_prev[    2]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
nosehoover_xi: not available

These values are, however, present in the cpt file:
box (3x3):
   box[    0]={ 6.69977e+00,  0.00000e+00,  0.00000e+00}
   box[    1]={ 0.00000e+00,  6.69977e+00,  0.00000e+00}
   box[    2]={ 3.34989e+00,  3.34989e+00,  4.73746e+00}
box-rel (3x3):
   box-rel[    0]={ 0.00000e+00,  0.00000e+00,  0.00000e+00}
   box-rel[    1]={ 0.00000e+00,  1.00000e+00,  0.00000e+00}
   box-rel[    2]={ 5.00000e-01,  5.00000e-01,  7.07108e-01}
box-v (3x3):
   box-v[    0]={ 3.46479e-02,  0.00000e+00,  0.00000e+00}
   box-v[    1]={ 0.00000e+00,  3.46479e-02,  0.00000e+00}
   box-v[    2]={ 1.73240e-02,  1.73240e-02,  2.44998e-02}
pres_prev (3x3):
   pres_prev[    0]={ 5.48752e+02,  2.40244e+02, -2.50854e+02}
   pres_prev[    1]={ 2.39676e+02, -8.38463e+01, -4.59079e+02}
   pres_prev[    2]={-2.52490e+02, -4.60009e+02,  7.15652e-01}
nosehoover-xi:   -0.265991    0.015571

Running the tpr file and additionally providing the checkpoint through the
‘mdrun -cpi’ command produces a different output than running mdrun using
the same tpr (even though the checkpoint has already been passed into
grompp).
This seems to be contrary to the intended behaviour, as the manual states:
'If you wish to change the ensemble or things like output frequency, then
supplying the checkpoint file to gmx grompp with -t along with a new .mdp
<http://manual.gromacs.org/online/mdp.html> file with -f is the recommended
procedure.’
If what I’m seeing is, in fact, the intended behaviour, it might be useful
to explicitly mention the need to supplement grompp with the energy files
as well in order to keep the barostat variables.


Kind regards

Dries