[gmx-users] Differences in default values for nstpcouple and cmap atomtypes between versions 4.0 and 4.5.4

Mark Abraham Mark.Abraham at anu.edu.au
Tue Apr 12 23:31:15 CEST 2011


On 13/04/2011 2:10 AM, Anna Duncan wrote:
> Hi,
>
> I've started using version 4.5.4 of gromacs, having previously been 
> using version 4.0 on an older server.
>
> When I run a simulation on v4.0 everything seems to run fine. 
>  However, when I run the same simulation on v4.5.4, mdrun gives me the 
> error message:
>
> ########################################################################
>
> Making 3D domain decomposition 4 x 3 x 2
> starting mdrun 'PROTEIN IN BILAYER'
> 5000000 steps, 150000.0 ps.
> step 0
> Step 11  Warning: pressure scaling more than 1%, mu: 1.02292 1.02292 
> 1.0122
>
> Step 11  Warning: pressure scaling more than 1%, mu: 1.02292 1.02292 
> 1.0122
>
> .
> .
> .
> .
> Step 11  Warning: pressure scaling more than 1%, mu: 1.02292 1.02292 
> 1.0122
>
> Step 21  Warning: pressure scaling more than 1%, mu: 1.09061 1.09061 
> 1.02015
>
> Step 21  Warning: pressure scaling more than 1%, mu: 1.09061 1.09061 
> 1.02015
> .
> .
> .
> Step 21  Warning: pressure scaling more than 1%, mu: 1.09061 1.09061 
> 1.02015
>
> Step 25, time 0.75 (ps)  LINCS WARNING
> relative constraint deviation after LINCS:
> rms 0.000004, max 0.000015 (between atoms 147 and 148)
> bonds that rotated more than 30 degrees:
>  atom 1 atom 2  angle  previous, current, constraint length
>     178    179   34.7    0.2600   0.2600      0.2600
>
> Step 28, time 0.84 (ps)  LINCS WARNING
> relative constraint deviation after LINCS:
> rms 19.241184, max 126.849724 (between atoms 35 and 36)
> bonds that rotated more than 30 degrees:
>  atom 1 atom 2  angle  previous, current, constraint length
>
> Step 28, time 0.84 (ps)  LINCS WARNING
> relative constraint deviation after LINCS:
> rms 1.739422, max 8.610986 (between atoms 496 and 497)
> bonds that rotated more than 30 degrees:
>  atom 1 atom 2  angle  previous, current, constraint length
>      45     46   90.0    0.2650   0.5083      0.2650
>      35     36   90.0    0.3100  39.6334      0.3100
>      42     43   90.0    0.2600   1.2129      0.2600
>      45     46   90.0    0.2650   0.5083      0.2650
>     482    483   90.0    0.2600   0.8649      0.2600
>     496    497   90.0    0.2700   2.5950      0.2700
>     496    498   90.0    0.2700   1.4511      0.2700
>     497    498   90.0    0.2700   2.5822      0.2700
> Wrote pdb files with previous and current coordinates
> Wrote pdb files with previous and current coordinates
> Segmentation fault

This looks like a standard case of 
http://www.gromacs.org/Documentation/Terminology/Blowing_Up, which can 
be quite situation dependent. In the absence of further information, I'd 
guess that 4.0 got lucky with the numerical integration, and 4.5.4 didn't.

> ########################################################################
>
>
> I've run g_gmxdump/gmxdump to get the input parameters for the 
> simulations using the 2 different versions,

gmxcheck is a good tool for comparing various file types.

> which I  can't attach because they're too big but which differ in the 
> lines shown below (v4.0 on the left and v4.5.4 on the right):
>
>    nstcomm              = 1      |   nstcomm              = 10
>    nstcalcenergy        = 1      |   nstcalcenergy        = 10
>    nsttcouple           = 1      |   nsttcouple           = 10
>    nstpcouple           = 1      |   nstpcouple           = 10

The heuristics by which these are set by default have changed. These 
values are not of themselves indicative of a problem.

>    rgbradii             = 2      |   rgbradii             = 1
>    sa_surface_tension   = 2.092      |   sa_surface_tension   = 2.05016
>    sc_sigma_min         = 0      |   sc_sigma_min         = 0.3
>    nstdhdl              = 1      |   nstdhdl              = 10

These only matter if you're using the relevant algorithms, for which the 
defaults have apparently changed.

> and also for cmap atomtypes values:
> in v4.0:
>       atomtype[  0]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  1]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  2]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  3]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  4]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  5]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  6]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  7]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  8]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[  9]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 10]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 11]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 12]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 13]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 14]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 15]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 16]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>       atomtype[ 17]={radius= 0.00000e+00, volume= 0.00000e+00, 
> gb_radius= 0.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, S_hct= 
> 0.00000e+00)}
>
> and in v4.1
>       atomtype[  0]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  1]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  2]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  3]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  4]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  5]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  6]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  7]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  8]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[  9]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 10]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 11]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 12]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 13]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 14]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 15]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 16]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}
>       atomtype[ 17]={radius=-1.00000e+00, volume=-1.00000e+00, 
> gb_radius=-1.00000e+00, surftens=-1.00000e+00, atomnumber=  -1, 
> S_hct=-1.00000e+00)}

Since I don't think any native 4.0 forcefield used CMAP at all, I infer 
you aren't, so this doesn't matter either.

> I haven't set any of these values and I don't understand why the 
> default values for parameters such as nsttcouple differ by a factor of 
> 10.  I know that in v4.5.4 parameters such as nsttcouple are by 
> default set to be equal to nstlist (which =10 in my simulations) but I 
> don't understand how the default is being set in v4.0.
>
> The larger nstpcouple default values seem to me like the cause of my 
> warning messages and eventual crash of the system when run in v4.5.4 
> but I am also worried that I don't know how the cmap atomtypes values 
> are being set and how these differences will affect the simulations. 
>  Do you have any pointers?

We don't know whether you're even using T-coupling, P-coupling or CMAP. 
How about a quick description of your objective, simulation system, 
equilibration protocol and maybe an .mdp file?

In any case, until you've exhausted the diagnosis steps on the above 
link, the smart money is betting that you have a somewhat unhappy 
starting configuration.

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110413/9c6909c8/attachment.html>


More information about the gromacs.org_gmx-users mailing list