- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Thomas Cheatham <cheatham_at_chpc.utah.edu>

Date: Wed 12 Jun 2002 12:35:53 -0700 (PDT)

*> I tried to run MD by constraint some atoms (NTR=1). I got an error message
*

*> "vlimit exceeded" at beginning of dynamics. An input and output are as below
*

...

*> Constrain G8
*

*> 5000.0
*

*> ATOM 236 250
*

*> END
*

*> END
*

...

*> vlimit exceeded for step 2; vmax = 271.116268
*

*>
*

*> COORDINATE RESETTING CANNOT BE ACCOMPLISHED,
*

*> DEVIATION IS TOO LARGE
*

*> NITER, NIT, LL, I AND J ARE : 0 1 65 181 182
*

To follow up on Carlos Simmerling's answer with a little more information,

at issue is the time step and high frequency motion. The NTR=1 option

turns on a harmonic potential to restrain coordinates to a reference

frame. The higher the force constant, the tighter the restraint, but also

the higher the frequency of the effective vibration during dynamics

(thermal motion). With force constants greater than ~20-250 kcal/mol

(depending on the system) for positional restraints, it is not possible to

integrate these high frequency motions with a 2 fs time step. In

principle you could simply run with a smaller time step, but this is

impractical. However, large restraint force constants are not necessary

to maintain the structure since the harmonic potential is fairly steep so

large motion away from the reference coordinates is not really possible

without a very large (energetic) push. A value of 1.0-5.0 kcal/mol

for the restraint constant is typically fine for dynamics and will

maintain the structure.

[Note that this issue with dynamics/vibration is irrelevant for

minimization where you can use large force constants like 5000.0]

As a rule of thumb, generally you want the integration time step in MD

simulation, dt, to be dt ~< T / 10 where T is the shortest period (or

fastest vibration). So for example, H- stretches are in the > 3000 cm-1

range, whereas bends (N-H, C-H, etc) are in the 1000-1600 cm-1 range.

v ~ 3000 cm-1 x (3x10**10 cm/sec) = 90x10**12/sec so

T = 1/v = 0.011 ps or 11 fs

--> dt ~1 fs for bond stretches (unless SHAKE is applied)

v ~ 1600 cm-1 x (3x10**10 cm/sec) = 48x10**12/sec so

T = 1/v = 0.021 ps or 21 fs

--> dt ~ 2 fs for bends

Note that the force constants for bond vibration in parm94.dat are

~200-600 kcal/mol and for bending are ~25-75 kcal/mol. With positional

restraints due to the (effectively) larger instantaneously violations,

smaller force constants are typically recommended.

I hope this helps.

Thomas E. Cheatham, III

Department of Medicinal Chemistry & Center for High Performance Computing

University of Utah INSCC 418 / BPRP 295A

30 South 2000 East, Room 201 155 South 1452 East

Salt Lake City, Utah 84112-5820 Salt Lake City, Utah 84112-0190

http://www.chpc.utah.edu/~cheatham

phone: (801) 587-9652 FAX: (801) 585-9119

Received on Wed Jun 12 2002 - 12:35:53 PDT

Date: Wed 12 Jun 2002 12:35:53 -0700 (PDT)

...

...

To follow up on Carlos Simmerling's answer with a little more information,

at issue is the time step and high frequency motion. The NTR=1 option

turns on a harmonic potential to restrain coordinates to a reference

frame. The higher the force constant, the tighter the restraint, but also

the higher the frequency of the effective vibration during dynamics

(thermal motion). With force constants greater than ~20-250 kcal/mol

(depending on the system) for positional restraints, it is not possible to

integrate these high frequency motions with a 2 fs time step. In

principle you could simply run with a smaller time step, but this is

impractical. However, large restraint force constants are not necessary

to maintain the structure since the harmonic potential is fairly steep so

large motion away from the reference coordinates is not really possible

without a very large (energetic) push. A value of 1.0-5.0 kcal/mol

for the restraint constant is typically fine for dynamics and will

maintain the structure.

[Note that this issue with dynamics/vibration is irrelevant for

minimization where you can use large force constants like 5000.0]

As a rule of thumb, generally you want the integration time step in MD

simulation, dt, to be dt ~< T / 10 where T is the shortest period (or

fastest vibration). So for example, H- stretches are in the > 3000 cm-1

range, whereas bends (N-H, C-H, etc) are in the 1000-1600 cm-1 range.

v ~ 3000 cm-1 x (3x10**10 cm/sec) = 90x10**12/sec so

T = 1/v = 0.011 ps or 11 fs

--> dt ~1 fs for bond stretches (unless SHAKE is applied)

v ~ 1600 cm-1 x (3x10**10 cm/sec) = 48x10**12/sec so

T = 1/v = 0.021 ps or 21 fs

--> dt ~ 2 fs for bends

Note that the force constants for bond vibration in parm94.dat are

~200-600 kcal/mol and for bending are ~25-75 kcal/mol. With positional

restraints due to the (effectively) larger instantaneously violations,

smaller force constants are typically recommended.

I hope this helps.

Thomas E. Cheatham, III

Department of Medicinal Chemistry & Center for High Performance Computing

University of Utah INSCC 418 / BPRP 295A

30 South 2000 East, Room 201 155 South 1452 East

Salt Lake City, Utah 84112-5820 Salt Lake City, Utah 84112-0190

http://www.chpc.utah.edu/~cheatham

phone: (801) 587-9652 FAX: (801) 585-9119

Received on Wed Jun 12 2002 - 12:35:53 PDT

Custom Search