#23363 - 02/05/10 09:28 PM
Re: Problem of Periodic Box MD Simulation
[Re: rmv]
|
Forum Member
Registered: 04/28/09
Posts: 133
Loc: U.S.A.
|
I am wondering if I could use two different methods for pressure and temperature at the same time in the production run? For example, can I use extended system method to control pressure, while using the 3-step Verlet T scaling to control temperature? Or should I only use extended system method for both temperature and pressure together? For NPT, one should use PGAMMA 0. and disable T scaling via IEQFRQ.
You've enabled both the HOOVER thermostat, and the 3-step Verlet T scaling via "IEQFRQ 2000 ICHECW 1"; you have dueling T control methods, and the latter one is fairly crude.
DCNTRL is the dynamics control subroutine name within CHARMM.
It appears that the HOOVER thermostat info is not echoed in the dynamics startup output.
|
|
Top
|
|
|
|
|
#23365 - 02/05/10 10:20 PM
Re: Problem of Periodic Box MD Simulation
[Re: mooctopus]
|
Forum Member
Registered: 09/17/03
Posts: 5604
Loc: 39 03 48 N, 77 06 54 W
|
It's possible to use IEQFRQ instead of HOOVER, but then it's not really NPT; the velocity adjustments of IEQFRQ are not a rigorous thermostat like HOOVER.
HOOVER can be used with a fixed volume for NVT via the CPT extended system. Especially useful for replica exchange, where one needs to quickly re-equilibrate at a new T value.
NVE does not regulate the T in any way, but for a well equilibrated system, the T should not drift too much.
The code does not echo the REFT and TBATH values; you don't see that info because the program does not provide it.
_________________________
Rick Venable
computational chemist
|
|
Top
|
|
|
|
|
#23366 - 02/05/10 10:29 PM
Re: Problem of Periodic Box MD Simulation
[Re: rmv]
|
Forum Member
Registered: 04/28/09
Posts: 133
Loc: U.S.A.
|
Just to confirm, so you mean the TPcontrol method when using NTV in production run? It's possible to use IEQFRQ instead of HOOVER, but then it's not really NPT; the velocity adjustments of IEQFRQ are not a rigorous thermostat like HOOVER.
HOOVER can be used with a fixed volume for NVT via the CPT extended system. Especially useful for replica exchange, where one needs to quickly re-equilibrate at a new T value.
NVE does not regulate the T in any way, but for a well equilibrated system, the T should not drift too much.
The code does not echo the REFT and TBATH values; you don't see that info because the program does not provide it.
Edited by mooctopus (02/05/10 10:30 PM)
|
|
Top
|
|
|
|
|
#23367 - 02/05/10 10:41 PM
Re: Problem of Periodic Box MD Simulation
[Re: mooctopus]
|
Forum Member
Registered: 09/17/03
Posts: 5604
Loc: 39 03 48 N, 77 06 54 W
|
No, TPCONTROL (tpcntrl.doc) is another code set for const T and/or const P that uses the VV2 integrator; it is not compatible with the CPT code described in pressure.doc. The VV2 integrator is required for using polarizable force fields with the Drude oscillator in CHARMM.
_________________________
Rick Venable
computational chemist
|
|
Top
|
|
|
|
|
#23375 - 02/08/10 01:10 PM
Re: Problem of Periodic Box MD Simulation
[Re: mooctopus]
|
Forum Member
Registered: 09/17/03
Posts: 5604
Loc: 39 03 48 N, 77 06 54 W
|
Suggested reading on the CPT implementation, from pressure.doc--
Feller, Zhang, Pastor & Brooks, J.Chem.Phys. 103, 4613 (1995)
Zhang, Feller, Brooks & Pastor, J.Chem.Phys. 103, 10252 (1995)
There are NVT examples with CPT and HOOVER in pressure.doc
Standard dynamics with the LEAPfrog integrator provides NVE dynamics; CPT or TPCONTROL codes are not used or needed.
_________________________
Rick Venable
computational chemist
|
|
Top
|
|
|
|
|
#23379 - 02/08/10 06:19 PM
Re: Problem of Periodic Box MD Simulation
[Re: mooctopus]
|
Forum Member
Registered: 09/17/03
Posts: 5604
Loc: 39 03 48 N, 77 06 54 W
|
When PMASS is used, all 9 elements of the piston mass tensor are set to that value, according to source/dynamc/prssre.src
_________________________
Rick Venable
computational chemist
|
|
Top
|
|
|
|
|
|
Previous Topic
Index