I skipped a lot of steps, so the questions I asked could be rather
misleading, sorry for that. I'll try to explain what I did in more details.
The main idea is to calculate simple radical in a water box.
There were the following steps/considerations made:
1. create a radical structure in tleap and
--------------------------------------------
SaveOff r r.lib
SavePdb r.pdb
--------------------------------------------
2. use qspcfw ff for water and solvate the radical
--------------------------------------------
source leaprc.ff99SB
loadOff solvents.lib
loadOff qspcfwbox.off
loadOff r.lib
WAT = SPG
set WAT.1 name "SPG"
loadAmberParams frcmod.qspcfw
set default FlexibleWater on
r = loadPdb r.pdb
solvateBox r QSPCFWBOX 8.0 0.5
saveAmberParm r Rqspcfw.prmtop Rqspcfw.inpcrd
--------------------------------------------
3. create LES, addles commands are:
--------------------------------------------
file rprm name=(Rqspcfw.prmtop) read
file rcrd name=(Rqspcfw.inpcrd) read
file wprm name=(Rqspcfw.LES.prmtop) wovr
file wcrd name=(Rqspcfw.LES.inpcrd) wovr
action
omas
space numc=32 pick chem prtc H done <-- this is H
atoms for the radical
space numc=16 pick chem prtc H1 | chem prtc H2 done <--- this is
water H atoms
*EOD
--------------------------------------------
here I first ran into problems, that I described earlier:
"LES was applied to solvent, atom This is not allowed."
to avoid this, I edit %FLAG SOLVENT_POINTERS in Rqspcfw.prmtop (to the
form 0 NNN 1), and after addles is done change this flag in
Rqspcfw.LES.prmtop back to original setting.
And another problem: in that case I getting Rqspcfw.LES.inpcrd file
without the last line (i.e. without box size), so I need to copy it from
the original Rqspcfw.inpcrd
How that could be avoided?
4. to equilibrate the system I need to run constant pressure dynamics:
--------------------------------------------
&cntrl
ipimd = 2, ntb = 2, ig = -1, ntx = 1, irest = 0, jfastw = 4,
temp0 = 300.0, temp0les = -1, ntt = 4, nchain = 4,
ntp = 1, pres0 = 100.0, comp = 43.9, taup = 1.0,
dt = 0.0002, nstlim = 10000,
cut = 7.0, ntpr = 1, ntwr = 20, ntwx = 1, ntwv = 0, ntwe = 0
/
--------------------------------------------
$AMBERHOME/bin/sander.LES.MPI -O -i Rqspcfw.LES.in -p Rqspcfw.LES.prmtop
-c Rqspcfw.LES.inpcrd -o Rqspcfw.LES.out -r Rqspcfw.LES.rst -x
Rqspcfw.LES.crd -v Rqspcfw.LES.vel -inf Rqspcfw.LES.info -e
Rqspcfw.LES.ext > sander_Rqspcfw.LES.out
here is another problem: sander stops with the following error:
--------------------------------------------
* NB pairs 229 213708 exceeds capacity ( 213809) 6
SIZE OF NONBOND LIST = 213809
SANDER BOMB in subroutine nonbond_list
Non bond list overflow!
check MAXPR in locmem.f
--------------------------------------------
OK, there is probably too much of local enhancement (that is strange,
could that be avoided without recompiling the program?), trying to run
with smaller numbers:
--------------------------------------------
space numc=32 pick chem prtc H done
space numc=8 pick chem prtc H1 | chem prtc H2 done
--------------------------------------------
In that case it was crushing with segfault:
--------------------------------------------
forrtl: severe (174): SIGSEGV, segmentation fault occurred
--------------------------------------------
So I thought that this could happen because the system is far from
equilibrium, and decided to equilibrate the system before the nuclei
quantization. That is how the question about classical simulation arose.
As for "exploding", maybe you are right about failure to image the
resulting trajectory back to the primary unit cell. I'm not very
familiar with how that could be done - where can I find the information
about that?
Right now I think I found the reason for the last crush: it looks like
addles starts to generate LES configuration with all beads for each
quantized nucleus with the same coordinates! When I tried it on a very
simple system it was creating beads with small variation in coordinates
(as I think it should). So now I'm trying to understand what had changed
since then.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jan 16 2013 - 09:00:03 PST
