[AMBER] TI solvated ligand simulation "vlimit exceeded for step"

From: <psu4.uic.edu>
Date: Thu, 4 Jul 2013 17:59:49 -0500

Dear Amber community,


 When we try to run TI solvated ligand only simulation from the step 1 ,2,
3 (2 is the softcore potential one), the error “*vlimit exceeded for step*”
always occur in the NPT equilibration step (The complex TI simulation from
step 1 to 3 works perfectly fine with the same command and the complex /
ligand parameters prepared with the same script. ). When the error
messages pop out, the energy terms/temperature/volume/density looks
normal. *From the previous posts, it seems like this error message
suggests there are two atoms overlap to each other or the system is
inhomogeneous.* However, when visualize in the MD movie, we don't see any
overlap atoms.


NSTEP = 17000 TIME(PS) = 67.000 TEMP(K) = 296.94 PRESS =
-532.5

 Etot = -7483.8363 EKtot = 3004.1045 EPtot =
-10487.9408

 BOND = 2158.5737 ANGLE = 18.1624 DIHED =
5.3110

 1-4 NB = 7.5274 1-4 EEL = 16.1451 VDWAALS =
2231.0874

 EELEC = -14924.7478 EHBOND = 0.0000 RESTRAINT =
0.0000

 DV/DL = 2.8211

 EKCMT = 1027.3506 VIRIAL = 1514.5626 VOLUME =
42372.4477


                                 Density = 0.8034

 Ewald error estimate: 0.6122E-03

 ------------------------------------------------------------------------------



  Softcore part of the system: 1 atoms, TEMP(K) =
269.99

 SC_Etot= 1.5887 SC_EKtot= 0.8048 SC_EPtot =
0.7839

 SC_BOND= 0.0034 SC_ANGLE= 0.6469 SC_DIHED =
0.1166

 SC_14NB= 0.0169 SC_14EEL= 0.0000 SC_VDW =
0.0000

 SC_EEL = 0.0000

 SC_RES_DIST= 0.0000 SC_RES_ANG= 0.0000 SC_RES_TORS=
0.0000

 SC_RES_PLPT= 0.0000 SC_RES_PLPL= 0.0000 SC_RES_GEN =
0.0000

 SC_EEL_DER= 0.0000 SC_VDW_DER= -0.2051 SC_DERIV =
      -0.2051

 ------------------------------------------------------------------------------



*vlimit exceeded for step 17068; vmax = 3315.2617*

*
*

The followings are the command we use:



*Minimization*

&cntrl

  imin = 1,

  ntx = 1,

  maxcyc = 2000,

  ntmin = 2,

  ntpr = 100,

  ntf = 1,

  ntc = 1,

  ntb = 1,

  cut = 10.0,

  icfe = 1,

  clambda =0.68392,

  ifsc=1,

  crgmask='${mask0}',

  scmask='${mask0}',

 &end


*Equi NV*

&cntrl

  imin = 0,

  irest = 0,

  ntx = 1,

  ntb = 1,

  cut = 10.0,

  ntr = 1,

  ntc = 1,

  ntf = 1,

  tempi = 0.0,

  temp0 = 300.0,

  ntt = 3,

  gamma_ln = 2.0,

  nstlim = 50000,

  dt = 0.001,

  ntpr = 250,

  ntwx = 250,

  ntwr = 2500,

  icfe = 1,

  clambda =0.68392,

  ifsc=1,

  crgmask='${mask0}',

  scmask='${mask0}',

/

Keep the Ligand fixed with weak restraints

5.0

RES 1 2

/

*Equi NPT*

&cntrl

  imin = 0,

  irest = 1,

  ntx = 5,

  ntb = 2,

  ntp = 1,

  pres0 = 1.0,

  taup = 2.0,

  cut = 10,

  ntr = 0,

  ntc = 1,

  ntf = 1,

  tempi = 300.0,

  temp0 = 300.0,

  ntt = 3,

  gamma_ln = 2.0,

  nstlim = 50000,

  dt = 0.001,

  ntpr = 1,

  ntwx = 1,

  ntwr = 1,

  icfe = 1,

  clambda =0.68392,

  ifsc=1,

  crgmask='${mask0}',

  scmask='${mask0}',

 &end



   Professor Case suggests to run equilibration with icfe = 0 to make the
systemhomogenous. <http://archive.ambermd.org/201109/0543.html> And then
use the homogenous, equilibrated .rst to start over the TI run. So we try
to run 2000 steps minimization, 50 ps NVT and 50ps NPT equilibration with
the group file way, ntwx = ntpr = ntwr =1, but icfe =0. Again, the system
blows up in the NPT equilibration step. The visual inspection of the MD
movie didn’t see anything abnormal. Just some water molecules flow far
away from the system.


NSTEP = 15599 TIME(PS) = 65.599 TEMP(K) = 8276.74 *PRESS =
NaN*

* Etot = NaN* EKtot = 83734.0351 *EPtot
= NaN*

 BOND = 2049.1408 ANGLE = 20.1363 DIHED =
9.4205

 1-4 NB = 8.0200 1-4 EEL = 13.3643 *VDWAALS
= NaN*

 EELEC = NaN EHBOND = 0.0000 RESTRAINT =
0.0000

 EKCMT = 1002.6881 * VIRIAL = NaN* VOLUME =
41816.3476

                                                    Density =
0.8141

* Ewald error estimate: NaN*

 ------------------------------------------------------------------------------



  Our guess is that ntc=1 & ntf =1 together with icfe=1 might be the
possible reason. So we try ntc=2 and ntf=1 with either icfe=1 or icfe=0,
running in the group file fashion. The same error messages ““*vlimit
exceeded for step*” happens when ntc=2 , ntf=1, icfe=1, running in the
group file fashion. *The simulation freezes/not going forward *when ntc=2
, ntf=1, icfe=0, running in group files, running in the group file
fashion.


The following is where the simulation stops/freezes in the ntc=2 , ntf=1,
icfe=0, ntwx = ntpr = ntwr =1, running in the group file fashion.
  Everything just looks normal.


NSTEP = 14037 TIME(PS) = 64.037 TEMP(K) = 297.80 PRESS =
-293.6

 Etot = -8381.6506 EKtot = 2013.5510 EPtot =
-10395.2016

 BOND = 9.3808 ANGLE = 17.8172 DIHED =
5.7128

 1-4 NB = 7.7311 1-4 EEL = 14.5935 VDWAALS =
1599.6602

 EELEC = -12050.0973 EHBOND = 0.0000 RESTRAINT =
0.0000

 EKCMT = 1001.6958 VIRIAL = 1273.5095 VOLUME =
42875.4053

                                                    Density =
0.7940

 Ewald error estimate: 0.3904E-05

 ------------------------------------------------------------------------------



 Another guess is that there might be problems with the solvated ligand
parameter files. So we try to run the regular MD 2000 steps minimization,
50ps NVT, 50ps NPT equilibration and 20ps NPT production with the above
commands (ntc=1 or 2, ntf=1, icfe=0). Everything just works perfectly
fine. *We therefore get confused/puzzled. Could any guru kindly offer any
solution? *


Here is the briefly summary of the several trials of solvated ligand
simulation (Solvated complex TI simulation jobs run perfectly fine.):




*Icfe=1*

*Icfe=0, but run in group files*

*Icfe=1*

*Icfe=0, but run in group files*

*Icfe=0, regular MD*

Minmization

*Fine*

*Fine*

*Fine*

*Fine*

*Fine*

50ps NV

equilibration

ntf=1, ntc=1, *fine*

ntf=1, ntc=1, *fine*

ntf=1, ntc=2, *fine*

ntf=1, ntc=2, *fine*

ntf=1, ntc=1 or2, *Fine*

50ps NPT equilibration

Error messages

ntf=1, ntc=1

* vlimit exceeded for step*

ntf=1, ntc=1

* System potential energy, pressure blows up: NaN.*

ntf=1, ntc=2

* vlimit exceeded for step*

ntf=1, ntc=2

*Simulation freeze around the same steps without any error message.*

ntf=1, ntc=1 or2, *Fine*





  Cheers,

  Henry
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jul 04 2013 - 16:30:02 PDT
Custom Search