Dear Users and Developers,
I'm trying to simulate an unfolded protein in explicit water at 1 bar
and T = 300K or T > 300K. Minimization is performed successfully. However,
during heating an I found the following error message
*"Error: an illegal memory access was encountered launching kernel
kClearForces"*
*Details of input*
Heat
&cntrl
imin=0,
ntx=1,
irest=0,
nstlim=1000000,
dt=0.002,
ntf=2,
ntc=2,
tempi=0.0,
temp0=298.0,
ntpr=1000,
ntwx=1000,
ntwr=50000,
cut=10.0,
ntb=1,
ntp=0,
ntt=3,
* gamma_ln=2.0,* nmropt=1,
ig=-1,
ntr=1,
restraint_wt =500.0,
restraintmask=":1-57"
I try to simulate the system with different values of collision
frequency *gamma_ln.
*I find that the above error occurs at NSTEP = 979000 with gamma_ln = 2.0.
On reducing the value of gamma_ln to 1.0 this error occurs at NSTEP =
464000. Finally, when I use gamma_ln = 5.0 the error is gone. I proceed
with gamma_ln = 5.0 for equilibration and production run at T = 300K.
After 41 ns production run, I find the same error with gamma_ln = 5.0. I
find that my simulation is stable when using gamma_ln = 10.
It appears that increasing the value of collision frequency resolves the
above error when using Langevin thermostat (not sure for other
thermostats). If this is the case, what would be the optimal choice for
*gamma_ln* when using the Langevin thermostat? Or, is there any other issue
related to the aforementioned error? I'm using the Amber20 version.
Any suggestions would be highly appreciated.
Thank you.
With best regards,
Umesh
*********************************************************
Dr. Umesh C. Roy
School of Computational and Integrative Sciences,
Jawaharlal Nehru University, New Delhi
India-110067
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 27 2021 - 00:30:02 PDT
