[gmx-users] Pres-XX and Pres-YY difference
Johnny Lu
johnny.lu128 at gmail.com
Fri Oct 10 17:38:48 CEST 2014
Sorry, my bad. May be those differences in pressures was normal fluctuation
of pressure.
After waiting for a while, I ran g_energy to get the pressures again.
Parinello Rahman
Statistics over 33869001 steps [ 0.0000 through 33869.0000 ps ], 10 data
sets
All statistics are over 338691 points
Energy Average Err.Est. RMSD Tot-Drift
-------------------------------------------------------------------------------
Pressure 0.82307 0.17 156.886 0.336604 (bar)
Pres-XX 1.28456 0.24 222.754 1.56681 (bar)
Pres-XY -0.171331 0.31 138.127 2.19114 (bar)
Pres-XZ -0.230648 0.36 137.731 1.24648 (bar)
Pres-YX -0.20907 0.3 138.123 2.18304 (bar)
Pres-YY 0.854579 0.28 225.2 1.20945 (bar)
Pres-YZ -0.14902 0.1 138.921 0.238693 (bar)
Pres-ZX -0.271166 0.36 137.719 1.29285 (bar)
Pres-ZY -0.225707 0.11 138.919 0.304902 (bar)
Pres-ZZ 0.330073 0.57 222.786 -1.76645 (bar)
Berendsen
Statistics over 23635301 steps [ 0.0000 through 47270.6000 ps ], 10 data
sets
All statistics are over 236354 points
Energy Average Err.Est. RMSD Tot-Drift
-------------------------------------------------------------------------------
Pressure 1.11418 0.39 140.31 1.76386 (bar)
Pres-XX 1.00041 0.72 212.365 1.26091 (bar)
Pres-XY -0.0733808 0.39 138.148 -1.94821 (bar)
Pres-XZ -0.0559881 0.23 138.047 0.135325 (bar)
Pres-YX -0.0785166 0.39 138.148 -1.99506 (bar)
Pres-YY 1.52045 0.42 212.52 0.2362 (bar)
Pres-YZ 0.142665 0.6 137.845 -1.995 (bar)
Pres-ZX -0.0616126 0.23 138.045 0.141702 (bar)
Pres-ZY 0.135693 0.6 137.843 -1.95285 (bar)
Pres-ZZ 0.821679 0.59 212.735 3.79446 (bar)
After that, I ran fourier transform on the pressure to check for
oscillations in the Pres-XX, Press-YY, and Pres-ZZ.
The Parinello-Rahman simulation has a peak of about 1 bar above baseline in
XX, but not YY or ZZ directions.
http://oi61.tinypic.com/vgrbtd.jpg
In the Berendsen simulation, the XX spectrum was similarly taller than the
YY and ZZ spectrum, despite Box-X and Box-Y has the same length.
http://oi59.tinypic.com/2i0t7kp.jpg
Parinello-Rahman box dimension:
Statistics over 34000001 steps [ 0.0000 through 34000.0000 ps ], 7 data sets
All statistics are over 340001 points
Energy Average Err.Est. RMSD Tot-Drift
-------------------------------------------------------------------------------
Box-X 7.50201 9.8e-05 0.00707842 7.87623e-05 (nm)
Box-Y 7.50201 9.8e-05 0.00707842 7.87623e-05 (nm)
Box-Z 5.30473 6.9e-05 0.0050052 5.56943e-05 (nm)
Volume 298.552 0.012 0.845082 0.00943589 (nm^3)
Box-Vel-XX 2.1668e-06 4.2e-06 0.0198765 5.37524e-06
(nm/ps)
Box-Vel-YY 2.1668e-06 4.2e-06 0.0198765 5.37524e-06
(nm/ps)
Box-Vel-ZZ 1.53216e-06 3e-06 0.0140548 3.80088e-06
(nm/ps)
Berendsen barostat box dimension
Statistics over 23635301 steps [ 0.0000 through 47270.6000 ps ], 4 data sets
All statistics are over 236354 points
Energy Average Err.Est. RMSD Tot-Drift
-------------------------------------------------------------------------------
Box-X 7.50095 6.6e-05 0.00283236 0.000361638 (nm)
Box-Y 7.50095 6.6e-05 0.00283236 0.000361638 (nm)
Box-Z 5.30398 4.7e-05 0.00200278 0.000255718 (nm)
Volume 298.424 0.0079 0.338058 0.0431648 (nm^3)
I guess the volume can't get more stable.
On Fri, Oct 10, 2014 at 10:24 AM, Johnny Lu <johnny.lu128 at gmail.com> wrote:
> I also ran another NPT simulation with Berendsen barostat, after those 300
> ns of NPT with Parrinello-Rahman barostat. So the two simulations in these
> two emails have the same starting point.
> This simulation uses 2 fs time step, while the one above use 1 fs time
> step.
>
> Below is the pressure tensor extracted with g_energy of mixed precision
> gromacs 4.6.7.
>
> Last energy frame read 236353 time 47270.602
>
> Statistics over 23635301 steps [ 0.0000 through 47270.6000 ps ], 9 data
> sets
> All statistics are over 236354 points
>
> Energy Average Err.Est. RMSD Tot-Drift
>
> -------------------------------------------------------------------------------
> Pres-XX 1.00041 0.72 212.365 1.26091 (bar)
> Pres-XY -0.0733808 0.39 138.148 -1.94821 (bar)
> Pres-XZ -0.0559881 0.23 138.047 0.135325 (bar)
> Pres-YX -0.0785166 0.39 138.148 -1.99506 (bar)
> Pres-YY 1.52045 0.42 212.52 0.2362 (bar)
> Pres-YZ 0.142665 0.6 137.845 -1.995 (bar)
> Pres-ZX -0.0616126 0.23 138.045 0.141702 (bar)
> Pres-ZY 0.135693 0.6 137.843 -1.95285 (bar)
> Pres-ZZ 0.821679 0.59 212.735 3.79446 (bar)
>
>
> The shear part of the pressure tensor is smaller. But the normal part
> (XX,YY, and ZZ) are still different.
>
> The XX part of the pressure tensor in both simulations are very close to 1
> bar, the target pressure, while the YY and ZZ part are not as close to 1
> bar.
>
> On Fri, Oct 10, 2014 at 10:03 AM, Johnny Lu <johnny.lu128 at gmail.com>
> wrote:
>
>> Hi. The following seems a bit strange to me. It was extracted from the
>> edr file with g_energy of mixed precision gromacs 4.6.7.
>>
>> Statistics over 31921501 steps [ 0.0000 through 31921.5000 ps ], 9 data
>> sets
>> All statistics are over 319216 points
>>
>> Energy Average Err.Est. RMSD Tot-Drift
>>
>> -------------------------------------------------------------------------------
>> Pres-XX 1.00863 0.23 222.874 0.00505598
>> (bar)
>> Pres-XY -0.218743 0.33 138.143 2.17256
>> (bar)
>> Pres-XZ -0.233072 0.33 137.629 1.36597
>> (bar)
>> Pres-YX -0.262129 0.32 138.137 2.12662
>> (bar)
>> Pres-YY 0.788658 0.23 225.34 0.928867
>> (bar)
>> Pres-YZ -0.0207522 0.23 138.913 1.08012
>> (bar)
>> Pres-ZX -0.276173 0.33 137.619 1.40069
>> (bar)
>> Pres-ZY -0.098337 0.23 138.913 1.14865
>> (bar)
>> Pres-ZZ 0.60576 0.38 222.687 -0.259413
>> (bar)
>>
>> I don't know the meaning of Pres-XX. But I guess it means the pressure at
>> some direction (some kind of tensor.).
>>
>> Why the pressure at XX direction is different from other direction?
>> I thought Pres-XX should be same as Pres-YY and Pres-ZZ.
>>
>> The box is a truncated dodecahedron, the barostat was set to 1 bar, and
>> the simulation was ran with mixed precision gromacs 4.6.7 with Parrinello-Rahman
>> barostat for NPT simulation of about 300 ns before this stretch of NPT
>> simulation.
>>
>> Thank you again.
>>
>
>
More information about the gromacs.org_gmx-users
mailing list