AMBER Archive (2009)

Subject: Re: [AMBER] ntf, ntc with softcore potentials and TI in amber10

Date: Tue May 05 2009 - 22:39:02 CDT


it is correct that you cannot use ntf=2 with ifsc>0. This does not mean
that you can't use SHAKE though. The reason is this: There is no code in
place to handle SHAKED bonds crossing from the common into the softcore
region of the system. Therefore, any such bonds have SHAKE removed from
them. Thats why you need to set ntf=1, to have this interaction evaluated.
This condition is enforced even if no such bond exists, since ntf=1 will
rarely impact performance much anyway. Since this will only involve a few
bonds overall, a timestep of 2 fs can still be used, so this is a
condition in which ntf and ntc need not be equal.

If you are dead set on using ntf=2 you could comment the error-causing
lines out of the code (in mdread.f), recompile and manually make sure that
no SHAKED bond crosses into the softcore region. I doubt it will influence
your simulation much.

Kind Regards,


> I have found that I am unable to use ntf=2, ntc=2 with ifsc=1 during
> thermodynamic intergration (as in step 2 of the TI tutorial). I had
> perhaps naively thought that using scmask would allow ntf=2,ntc=2 and
> that scmask would exclude those atoms with softcore potentials from
> So I guess I am really asking - whether the above is correct? and, do
> I need to use ntf=1,ntc=1 with softcore potentials. Does this matter,
> keeping in mind that ntf=2,ntc=2 are used during charge removal and
> then addition (as in step1 and step3 of the TI tutorial)
> Cheers,
> Matthew
> Matthew CJ Wilce
> NHMRC Senior Research Fellow
> Department of Biochemistry & Molecular Biology
> Building 76, Monash University
> Tel: +61-3-9902-9244
> Email:
> _______________________________________________
> AMBER mailing list

Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

AMBER mailing list