AMBER Archive (2004)

Subject: Re: AMBER: Regarding Dielectric Constant in AMBER8

From: pande.vineet_at_fc.up.pt
Date: Fri Nov 26 2004 - 12:59:19 CST


Dear Prof. Simmerling:
Thanks for reply,
> >
> you say that you usually set use_pme to 0 to "save time", but you
> should also consider the quality of the calculation, not just the speed.
> many studies have shown that long-range interactions can be important.

Sorry I didnīt make my self clear...I specify use_pme to 0 only during initial
phases of MD to save time, like during initial water minimization..i did not
mean I donīt use the reciprocal part of the PME at all!
> >
> yes, this is a distance dependent dielectric. sander should
> tell you this, please read the output files carefully.
>
I am pasting a part of output:

 Number of triangulated 3-point waters found: 0

     Sum of charges from parm topology file = 3.49997982

--------------------------------------------------------------------------------
   4. RESULTS
--------------------------------------------------------------------------------

 ---------------------------------------------------

     eedmeth=5: Using 1/r dielectric

 ---------------------------------------------------
| Local SIZE OF NONBOND LIST = 1294586
| TOTAL SIZE OF NONBOND LIST = 1294586

   NSTEP ENERGY RMS GMAX NAME NUMBER
      1 2.8405E+04 1.9081E+03 1.3421E+05

**********
**********

eedmeth=5: Using 1/r dielectric ŧŧ does not look like as specified in the page
109 of manual, a distance-dependent dielctric should be 1/r2 (?) rather than
1/r...??

I repeat my input flags:
 &cntrl
 ntx = 1, ntrx = 1, ntpr = 10,
 ntb = 0,
 igb = 0,
 dielc = 4,
 cut = 12.0,
 scnb = 2.0, scee = 1.2,
 imin = 1, ntr = 1, restraint_wt=32, restraintmask=':* & !@H=', maxcyc = 5000,
ncyc = 500,
 &end
 &ewald
 eedmeth=5,
 &end
EOF

What could be the problem?
With thanks and kind regards
Vineet

Quoting Carlos Simmerling <carlos_at_ilion.bio.sunysb.edu>:

> pande.vineet_at_fc.up.pt wrote:
>
> >
> >The calculation runs beyond perfection, structure is quite reasonable :)...
> This
> >means that with eedmeth=5 in the ewald part and dielc = 4, in the upper
> section
> >we are actually implementing a distance-dependent dielctric 4r(i,j), right?
> One
> >
> >
> yes, this is a distance dependent dielectric. sander should
> tell you this, please read the output files carefully.
>
> >another question.. In this non-periodic case of ewald sums, how shall i
> decide
> >for example on other PME flags like USE_PME (which I usually put 0 for PBCs
> to
> >
> >
> use_pme won't matter, your sander output should already say that pme isn't
> used since it isn't periodic.
>
> >save time..), and VDWMETH..? Shall they remain defaults, like I used in this
> >test?
> >
> >
> you say that you usually set use_pme to 0 to "save time", but you
> should also consider the quality of the calculation, not just the speed.
> many studies have shown that long-range interactions can be important.
> carlos
>
> >Thanks again,
> >Kind Regards
> >Vineet
> >
> >
> >
> >omo_at_scripps.edu
> >
> >
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

-------------------------------------------------------------
A FCUP utiliza o sistema de webmail Horde/IMP (www.horde.org)

Visite: http://www.fc.up.pt/

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu