Re: Re: Termination between Minimization to Equilibration

From: Peter Freddolino (petefred_at_umich.edu)
Date: Wed Dec 02 2020 - 20:07:58 CST

Well, I suspect that the problem that you're encountering is that the unit
cell size is probably changing quite quickly, which can lead to large
forces if you have fixed atoms (since their coordinates don't get rescaled
along with everything else) or 'constraints' (using NAMD terminology),
which can give large forces upon large changes in periodic cell size.
During the early stages of equilibration, you might shorten the time step
and/or increase your piston period and decay constant to slow down the
volume fluctuations and let the system keep up (fourfold higher than your
current values is pretty typical anyway, even for the production part of
simulations).
Best,
Peter

On Wed, Dec 2, 2020 at 8:48 PM Ramon Mendoza Uriarte <ramendoza_at_uchicago.edu>
wrote:

> Yes, I am using this pressure control in my config file with a timestep of
> 1 fs:
>
> if {1} {
> useGroupPressure yes ;# needed for 2fs steps
> useFlexibleCell yes ;# no for water box, yes for membrane
> useConstantRatio yes ;# keeps the ratio of the unit cell in the x-y
> plane constant A=B
>
> langevinPiston on
> langevinPistonTarget 1.01325 ;# in bar -> 1 atm
> langevinPistonPeriod 50.0
> langevinPistonDecay 25.0
> langevinPistonTemp $temperature
> }
>
> Should no pressure control be used in the first minimization? Or does this
> need to be changed into something like this:
>
> if {1} {
> useGroupPressure yes ;# needed for 2fs steps
> useFlexibleCell yes ;# no for water box, yes for membrane
> useConstantRatio yes ;# keeps the ratio of the unit cell in the x-y
> plane constant A=B
> }
>
> Thank you,
> Ramon
>
> ------------------------------
> *From:* Peter Freddolino <petefred_at_umich.edu>
> *Sent:* Wednesday, December 2, 2020 6:35 PM
> *To:* <namd-l_at_ks.uiuc.edu> <namd-l_at_ks.uiuc.edu>; Ramon Mendoza Uriarte <
> ramendoza_at_uchicago.edu>
> *Subject:* Re: namd-l: Re: Termination between Minimization to
> Equilibration
>
> Constraints/restraints can play poorly with pressure control... you may
> need to tone them down or use a shorter timestep until the system volume
> stabilizes. Are you running a constant pressure simulation?
> Best,
> Peter
>
>
> On Wed, Dec 2, 2020 at 8:24 PM Ramon Mendoza Uriarte <
> ramendoza_at_uchicago.edu> wrote:
>
> I think the termination was occurring due to the simulation taking up too
> much memory and I seem to have fixed this by increasing restartfreq and
> dcdfreq. After, fixing this I get the error that I was initially trying to
> fix:
>
> ERROR: Constraint failure in RATTLE algorithm for atom XXXX!
> ERROR: Constraint failure; simulation has become unstable.
> ------------------------------
> *From:* owner-namd-l_at_ks.uiuc.edu <owner-namd-l_at_ks.uiuc.edu> on behalf of
> Ramon Mendoza Uriarte <ramendoza_at_uchicago.edu>
> *Sent:* Wednesday, December 2, 2020 12:43 PM
> *To:* NAMD Listserve <namd-l_at_ks.uiuc.edu>
> *Subject:* namd-l: Termination between Minimization to Equilibration
>
> Hello NAMD community,
>
> I am having an issue with a system containing a protein, lipid, water, and
> ions (totaling ~50,000 atoms). I built the system using CHARMM-GUI and I am
> having difficulty with the simulation failing in the transition from
> minimization to equilibration. Consequently, I could use a bit of guidance.
>
> I first attempted to minimize the system with the protein backbone and
> ions held fixed, except for a small portion of the protein which was not
> held fixed. This was done with fixedAtoms and the minimization was
> performed for 2000 steps at a 2 fs timestep and at a temperature of 310 K.
> However, the simulation failed at the start of the equilibration with the
> error 'ERROR: Constraint failure in RATTLE algorithm for atom XXXX!'. I
> checked the output and the atoms in question seem to be fine. The NAMD
> archives suggested that this error was caused by insufficient minimization
> and/or not using a 1 fs timestep (Or maybe due to rapid heating?).
>
> As such, I restarted the simulation with an additional 5000 steps of
> minimization (all other simulation parameters unchanged). Again, the
> minimization successfully completes but fails at the start of the
> equilibration step this time with the following:
>
> namd2:11843 terminated with signal 11 at PC=110c1ea SP=7eba4bffd2c0.
> Backtrace:
>
> namd2(_ZN12PmeRealSpace14compute_forcesEPKPKfPK11PmeParticleP6Vector+0x620)[0x110c1ea]
> namd2[0xc68efc]
>
> namd2(_ZN19CkIndex_WorkDistrib29_call_enqueuePme_LocalWorkMsgEPvS0_+0xe)[0x11fe2fe]
> namd2(_Z15_processHandlerPvP11CkCoreState+0x543)[0x132dba3]
> namd2(CsdScheduler+0x59f)[0x14c0b5f]
> namd2[0x149c5ff]
> namd2[0x1497f63]
> /lib64/libpthread.so.0(+0x7dd5)[0x7efe03b65dd5]
> /lib64/libc.so.6(clone+0x6d)[0x7efe0136b02d]
>
> This error was not as clear, at least to me, and my search into this
> termination seems to suggest the issue is with the program trying to access
> unassigned memory.
>
> So, I next decided to restart the simulation with an additional 3000 steps
> however with a 1 fs timestep (all other simulation parameters unchanged). Again,
> the minimization successfully completes but fails at the start of the
> equilibration step with a similar termination as above:
>
> namd2:1714 terminated with signal 11 at PC=110c1ea SP=7efd3c76a1c0.
> Backtrace:
>
> namd2(_ZN12PmeRealSpace14compute_forcesEPKPKfPK11PmeParticleP6Vector+0x620)[0x110c1ea]
> namd2[0xc68efc]
>
> namd2(_ZN19CkIndex_WorkDistrib29_call_enqueuePme_LocalWorkMsgEPvS0_+0xe)[0x11fe2fe]
> namd2(_Z15_processHandlerPvP11CkCoreState+0x543)[0x132dba3]
> namd2(CsdScheduler+0x59f)[0x14c0b5f]
> namd2[0x149c5ff]
> namd2[0x1497f63]
> /lib64/libpthread.so.0(+0x7dd5)[0x7f04c6b61dd5]
> /lib64/libc.so.6(clone+0x6d)[0x7f04c436702d]
>
> At this point I am unsure how to proceed and so any guidance will be
> appreciated!
>
> Thank you,
> Ramon
>
>

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:10 CST