AMBER Archive (2006)

Subject: RE: AMBER: ff02 + POL3 problem !

From: Yong Duan (
Date: Tue May 16 2006 - 15:06:54 CDT

Hi Pradipta,

The "LINMIN FAILURE" signals difficulty to continue minimization. This was
because the forces in sander was not "conditioned" for the constraints. At
this point, my suggestion is not to worry about LINMIN FAILURE, unless you
want to add a module to shake force (you are very welcome to do so, though

Please do not try MD with SHAKE off. As I said early, all these water models
were developed under the assumption of rigid bonds. If you let the bond
changes, you need to re-do the parameterization of the water models.

If you run MD with NTC=1, as the manual says, you have no contrain on the
bonds. Sander does not automatically apply SHAKE (because user says no); it
really tries very hard to follow the instruction from the users :).

Bottom line, continue MD with NTC=2 from the minimized structure (w/ SHAKE).


> -----Original Message-----
> From:
> [] On Behalf Of Pradipta Bandyopadhyay
> Sent: Tuesday, May 16, 2006 7:08 AM
> To:
> Subject: RE: AMBER: ff02 + POL3 problem !
> Hello Yong,
> Thanks a lot for your help.
> I did with shake on but still the problem remains
> I tried the following protocol
> (1) first IPOL=0 and shake off
> (2) then IPOL=1 and shake off
> (3) then IPOL=1 and shake on
> however, whenever I turn on shake with IPOL=1, it is stopping with
> I tried one MD with shake off (from the minimized geometry of
> IPOL=1 and
> shake=off) - it crashes in 50 steps (since, some parts of the
> system is
> getting large forces).
> I would appreciate any suggestion.
> regards,
> Pradipta
> p.s. one thing is not clear - it seems we must set NTC=2 for
> fixing rigid
> water models. what happens if someone does NTC=1 in MD for
> TIP3P+protein -
> does AMBER by default fix the rigid water models? Otherwis if
> models like
> TIP3P is not kept rigid then it is no longer a TIP3P model strictly
> speaking.
> >
> > Pradipta,
> >

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to