[gmx-users] FW: Lincs all-bonds causing instability in otherwise stable system

[Cara Kreck]

Hi,

I am having trouble constraining all-bonds with lincs in a system that is stable using shake all-bonds or lincs h-bonds in gromacs 4.6.7. The previous step using lincs h-bonds gave these energies:

           Step           Time         Lambda
         200000      200.00000        0.00000

   Energies (kJ/mol)
        G96Bond       G96Angle    Proper Dih.  Improper Dih.          LJ-14
    1.76317e+04    2.43132e+04    1.39308e+04    1.55159e+03   -2.91002e+03
     Coulomb-14        LJ (SR)        LJ (LR)   Coulomb (SR)   Coulomb (LR)
    4.07560e+05   -4.09017e+03   -1.26700e+04   -5.62217e+05   -3.06259e+03
       RF excl.      Potential    Kinetic En.   Total Energy    Temperature
   -1.24406e+05   -2.44368e+05    9.29185e+04   -1.51450e+05    3.02522e+02
 Pressure (bar)   Constr. rmsd
   -1.98735e+02    1.95240e-06

However, when I switch to lincs all-bonds the 0 step kinetic energy, temperature, and pressure are dramatically increased, even though continuation=yes and gen_vel=no. The system then quickly crashes. I tried using GMX_MAXCONSTRWARN=-1 and particle decomposition to get past the errors but then it simply stalls without crashing or outputting anything.

           Step           Time         Lambda
              0        0.00000        0.00000

Grid: 8 x 8 x 14 cells
   Energies (kJ/mol)
       G96Angle    Proper Dih.  Improper Dih.          LJ-14     Coulomb-14
    2.43132e+04    1.39308e+04    1.55159e+03   -2.91002e+03    4.07560e+05
        LJ (SR)        LJ (LR)   Coulomb (SR)   Coulomb (LR)       RF excl.
   -4.09017e+03   -1.26700e+04   -5.62217e+05   -3.06259e+03   -1.24406e+05
      Potential    Kinetic En.   Total Energy    Temperature Pressure (bar)
   -2.62000e+05    4.92862e+05    2.30862e+05    1.93783e+03    1.04858e+04
   Constr. rmsd
    1.44601e-02

When I instead switch to shake all-bonds (with particle decomposition) I get a similar spike in the kinetic energy etc. but the system is able to recover without crashing. Unfortunately gromacs won't let me use pressure coupling with shake due to my twin-range cut-offs though, so I need to find a way to get lincs to work. I tried switching to lincs all-bonds after 100ps with shake all-bonds, and there was no longer the spike in energies except for the pressure:

           Step           Time         Lambda
         100000      100.00000        0.00000

   Energies (kJ/mol)
       G96Angle    Proper Dih.  Improper Dih.          LJ-14     Coulomb-14
    2.46619e+04    1.38286e+04    1.50388e+03   -2.92380e+03    4.11368e+05
        LJ (SR)        LJ (LR)   Coulomb (SR)   Coulomb (LR)       RF excl.
   -3.50689e+03   -1.26695e+04   -5.64100e+05   -2.81505e+03   -1.24427e+05
      Potential    Kinetic En.   Total Energy    Temperature Pressure (bar)
   -2.59079e+05    7.73473e+04   -1.81732e+05    3.04114e+02   -4.16208e+02

           Step           Time         Lambda
              0        0.00000        0.00000

   Energies (kJ/mol)
       G96Angle    Proper Dih.  Improper Dih.          LJ-14     Coulomb-14
    2.46619e+04    1.38286e+04    1.50388e+03   -2.92380e+03    4.11368e+05
        LJ (SR)        LJ (LR)   Coulomb (SR)   Coulomb (LR)       RF excl.
   -3.50690e+03   -1.26695e+04   -5.64100e+05   -2.81533e+03   -1.24427e+05
      Potential    Kinetic En.   Total Energy    Temperature Pressure (bar)
   -2.59079e+05    7.73979e+04   -1.81681e+05    3.04313e+02    5.61153e+02
   Constr. rmsd
    2.45638e-04

However, it still crashes with a domain decomposition error. When I switch back to particle decomposition it stalls again. I then tried decreasing the time-step to 0.1fs and I got lots of these errors before stalling:

WARNING: Listed nonbonded interaction between particles 2485 and 2490
at distance 3f which is larger than the table limit 3f nm.

This is likely either a 1,4 interaction, or a listed interaction inside
a smaller molecule you are decoupling during a free energy calculation.
Since interactions at distances beyond the table cannot be computed,
they are skipped until they are inside the table limit again. You will
only see this message once, even if it occurs for several interactions.

IMPORTANT: This should not happen in a stable simulation, so there is
probably something wrong with your system. Only change the table-extension
distance in the mdp file if you are really sure that is the reason.

There were 156 inconsistent shifts. Check your topology
There were 190 inconsistent shifts. Check your topology


Whereas with domain decomposition and a 0.1 fs time-step it still crashes with a domain decomposition error. I don't understand why a system that is perfectly fine with lincs h-bonds and shake all-bonds can be so problematic with lincs all-bonds. Any suggestions?

Thanks,

Cara