Re: Simulation of large bilayer crashes with "ERROR: Stray PME grid charges detected"

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Fri Sep 14 2007 - 07:19:38 CDT

Hi Philip,
don't worry; I just wanted to clarify things.
Thanks,
Peter

Philip Peartree wrote:
> Apologies Peter! My bad!
>
> Philip Peartree
>
>
> Quoting Peter Freddolino <petefred_at_ks.uiuc.edu>:
>
>> Hi Maria,
>> you mentioned that you saw several atoms drift out of the box but
>> maintain bonds to atoms in the box -- were you viewing the file using a
>> pdb as the starting structure, or a psf? The most likely cause for what
>> you're describing is just a visual artifact stemming from not using a
>> structure file that includes bond information. Because atoms should only
>> be wrapped across periodic boundaries as a complete bonded chain, you
>> should not be getting bonds split across the periodic boundary.
>>
>> Out of curiosity, have you tried this run on the same hardware with a
>> single processor? Stray PME grid charge errors are not possible with one
>> processor, so this might help to find some underlying problem with the
>> system. And have you performed other namd runs using the same
>> computer(s) in the past?
>>
>> Wrapnearest effectively changes the shape of the unit cell you're using,
>> and should be avoided unless you're certain that's what you want. It has
>> no effect on a unit cell with orthogonal axes, but can be used to get
>> unusual shapes (such as hexagonal cells) in other cases.
>>
>> Best,
>> Peter
>> Philip Peartree wrote:
>>> I think that using wrapNearest (i.e. setting wrapNearest to on) as
>>> well as
>>> wrapAll might solve this problem... maybe others on the list can
>>> confirm this?
>>>
>>>
>>> Phil Peartree
>>>
>>>
>>> Quoting maria goranovic <mariagoranovic_at_gmail.com>:
>>>
>>>> Hi,
>>>>
>>>> Yes, I am using full PBC. Below is the configuration script. The
>>>> constraint
>>>> file just has harmonic constraints (0.1) on all POPC atoms. Water is
>>>> free
>>>>
>>>> #############################################3
>>>> set MOL popc-big
>>>>
>>>> set xbox 114
>>>> set ybox 114
>>>> set zbox 168
>>>>
>>>> structure ${MOL}.psf
>>>> coordinates ${MOL}.pdb
>>>> temperature 0
>>>> parameters par_all27_prot_lipid.prm
>>>> paraTypeCharmm on
>>>>
>>>> outputEnergies 5000
>>>> outputTiming 5000
>>>> xstFreq 5000
>>>> dcdFreq 5000
>>>> wrapAll on
>>>>
>>>> timestep 1
>>>> nonBondedFreq 1
>>>> fullElectFrequency 1
>>>> stepsPerCycle 20
>>>>
>>>> exclude scaled1-4
>>>> 1-4scaling 1.0
>>>> switching on
>>>> switchDist 12
>>>> cutoff 14
>>>> pairlistdist 17
>>>>
>>>> cellBasisVector1 $xbox 00.00 00.00
>>>> cellBasisVector2 00.00 $ybox 00.00
>>>> cellBasisVector3 00.00 00.00 $zbox
>>>>
>>>> cellOrigin 0 0 0
>>>>
>>>> Pme on
>>>> PmeGridsizeX 120
>>>> PmeGridsizeY 120
>>>> PmeGridsizeZ 180
>>>>
>>>> constraints on
>>>> consRef ${MOL}-cons.pdb
>>>> consKFile ${MOL}-cons.pdb
>>>> consKCol B
>>>>
>>>> #########################################
>>>> # PRESSURE AND TEMPERATURE CONTROL
>>>> #########################################
>>>>
>>>> # temperature
>>>> langevin on
>>>> langevinDamping 5
>>>> langevinTemp $temp
>>>> langevinHydrogen no
>>>>
>>>> useflexiblecell yes
>>>> useconstantratio yes
>>>> langevinPiston on
>>>> langevinPistonTarget 1.01325
>>>> langevinPistonPeriod 100
>>>> langevinPistonDecay 50
>>>> langevinPistonTemp $temp
>>>>
>>>> binaryoutput on
>>>> outputname ${MOL}-equil
>>>>
>>>> # INITIAL ENERGY
>>>> minimize 0
>>>>
>>>> minimize 10000
>>>> output ${MOL}-min1
>>>>
>>>> minimize 10000
>>>> output ${MOL}-min2
>>>>
>>>> minimize 10000
>>>> output ${MOL}-min3
>>>>
>>>> #########################################
>>>> # HEAT WITH PC constrained
>>>> #########################################
>>>>
>>>> set tem 1
>>>> while { $tem <= 313 } {
>>>> langevinTemp $tem
>>>> run 3000
>>>> output ${MOL}-heat-$tem
>>>> set tem [expr $tem + 3]
>>>> }
>>>>
>>>> run 100000
>>>> output ${MOL}-dyna-1
>>>> #############################################3
>>>>
>>>>
>>>>
>>>> On 9/13/07, Philip Peartree <P.Peartree_at_postgrad.manchester.ac.uk>
>>>> wrote:
>>>>>
>>>>> I assume you're running in full periodic boundary conditions?
>>>>> Could you
>>>>> post the
>>>>> input script you're using?
>>>>>
>>>>> Philip Peartree
>>>>>
>>>>> Quoting maria goranovic <mariagoranovic_at_gmail.com>:
>>>>>
>>>>> > Hi,
>>>>> >
>>>>> > I have a 110 x 110 x 170 A bilayer (build using vmd plugins),
>>>>> which I am
>>>>> > trying to minimize using a PME grid of 120 x 120 x 180. However,
>>>>> starting
>>>>> > from minimization step 80, I get several errors similar to the
>>>>> following
>>>>> in
>>>>> > the simulation
>>>>> >
>>>>> > ##############################
>>>>> > ..
>>>>> > ERROR: Stray PME grid charges detected: 28 sending to 25 for
>>>>> planes 92
>>>>> 93 94
>>>>> > 95
>>>>> > ERROR: Stray PME grid charges detected: 28 sending to 10 for
>>>>> planes 36
>>>>> 37 38
>>>>> > 39
>>>>> > ##############################
>>>>> >
>>>>> > Eventually the simulation crashes with the log:
>>>>> >
>>>>> > ##############################
>>>>> > ..
>>>>> >
>>>>> > ERROR: Stray PME grid charges detected: 28 sending to 23 for
>>>>> planes 84
>>>>> 85 86
>>>>> > 87
>>>>> > Stack Traceback:
>>>>> > [0]
>>>>> _ZN17ComputeHomeTuplesI8BondElem4bond9BondValueE10loadTuplesEv+0x95e
>>>>> > [0x81311ae]
>>>>> > [1]
>>>>> _ZN17ComputeHomeTuplesI8BondElem4bond9BondValueE6doWorkEv+0x56b
>>>>> > [0x8133b7b]
>>>>> > [2] _ZN11WorkDistrib12enqueueBondsEP12LocalWorkMsg+0x19
>>>>> [0x82de0ed]
>>>>> > [3]
>>>>> >
>>>>> _ZN19CkIndex_WorkDistrib31_call_enqueueBonds_LocalWorkMsgEPvP11WorkDistrib+0x11
>>>>>
>>>>>
>>>>> > [0x82de0cd]
>>>>> > [4] CkDeliverMessageFree+0x29 [0x8322f7d]
>>>>> > [5] _Z15_processHandlerPvP11CkCoreState+0x417 [0x83225bf]
>>>>> > [6] CmiHandleMessage+0x1d [0x837c885]
>>>>> > [7] _ZN7BackEnd4initEiPPc+0x295 [0x80e2f89]
>>>>> > [8] main+0x30 [0x80deff4]
>>>>> > [9] __libc_start_main+0xdc [0xf7d3c87c]
>>>>> > [10] _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_c+0x59
>>>>> > [0x80dba71]
>>>>> > ##############################
>>>>> >
>>>>> > Looking at the trajectory (3 frames, 5000 steps of minimization
>>>>> each),
>>>>> > several atoms have moved out of the simulation box, and have very
>>>>> long
>>>>> bonds
>>>>> > with atoms inside the box. These are both POPC and water atoms.
>>>>> >
>>>>> > Can someone please tell me what the issue might be ?
>>>>> >
>>>>> > Thank you very much,
>>>>> > --
>>>>> > Maria G.
>>>>> > Technical University of Denmark
>>>>> > Copenhagen
>>>>> >
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Maria G.
>>>> Technical University of Denmark
>>>> Copenhagen
>>>>
>>>
>>>
>>
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:15 CST