Hi everyone,
Another person had emailed Ross about this issue (just saw segfaults), which
I traced back to this disconnect between ATOMS_PER_MOLECULE and NATOM. The
code certainly requires these to match, but in these particular places they
don't. I'm not sure how these prmtops were created in the first place.
The question is, should we add a consistency check that forces this
condition and bombs with a helpful error message, or should we simply say
that modifying the prmtop is inherently risky and we can't possibly catch
all the stupid things people do to them? Of course if this is something you
can trick leap into doing by accident, I'd say introduce the check and patch
leap.
All the best,
Jason
---------- Forwarded message ----------
From: Jason Swails <jason.swails.gmail.com>
Date: Wed, May 18, 2011 at 11:30 PM
Subject: Re: [AMBER] segmentation fault when running sander
To: AMBER Mailing List <amber.ambermd.org>
Hi Chinsu,
I received your topology file and looked at it, and I know why you're
getting the error that you're getting. My question is where and how
*exactly* did you create your topology file? If you tried to hack the
topology file yourself to add stuff, then you need to be more wary of what
you change, and be willing to track down problems like these. If you used
some program or script to build your topology file, then there is a bug with
that program or script (I haven't been able to reproduce this error with
simple leap scripts in either sleap or tleap). If you can provide the exact
method you used to create this topology file it may help other people avoid
similar fates.
The problem is that not every atom is assigned to a molecule. That is, if
you add up every number in the prmtop section ATOMS_PER_MOLECULE, it comes
out to 24550, yet NATOM in that prmtop is defined as 24559. Thus, the last
9 atoms (last 3 water molecules) are not assigned to any molecules. This
will cause a segfault for a large number of compilers (if you're lucky --
otherwise it just spits out bogus results). The actual problem is that the
first "molecule" is actually 2293 atoms large, yet ATOMS_PER_MOLECULE lists
it as 2284. If you make this change, it *should* work, but since I don't
know how this prmtop was created and I already know there's one problem with
it, I wouldn't trust it at all. Note that ONLY constant pressure
simulations use molecule information (to do pressure scaling), so this
segfault will only occur when you try NTP simulations.
HTH,
Jason
On Wed, May 18, 2011 at 10:31 PM, Chinh Su Tran To <chinh.sutranto.gmail.com
> wrote:
> Dear Jason,
>
> The files are too big. Please dont mind if i sent them to your email.
> Thank you.
>
> Chinsu
>
> On Thu, May 19, 2011 at 2:14 AM, Jason Swails <jason.swails.gmail.com
> >wrote:
>
> > Can you attach your input files? This includes the topology file,
> inpcrd,
> > and mdin. Segfaults are indicative of memory errors, but it's not at all
> > specific.
> >
> > There may be something subtly wrong with your topology file, or it may be
> a
> > bug.
> >
> > All the best,
> > Jason
> >
> > --
> > Jason M. Swails
> > Quantum Theory Project,
> > University of Florida
> > Ph.D. Candidate
> > 352-392-4032
> >
> > On May 18, 2011, at 1:00 AM, Chinh Su Tran To <chinh.sutranto.gmail.com>
> > wrote:
> >
> > > Dear Amber users,
> > >
> > > Again i got the same error of "segmentation fault" even whatever i
> > modified
> > > my batch-file. The "ndm_md1.rst" was not even created.
> > >
> > > I compared the "ndm_md1.out" with other successful runs (i.e other
> > > "xx_md1.out"), i found that the process of running wat-md1.in was
> halted
> > in
> > > the first step, as shown below (this is extracted from the
> > "ndm_md1.out"):
> > >
> > > *NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 0.00 PRESS =
> > > 0.0
> > > Etot = -64769.2683 EKtot = 0.0000 EPtot =
> > > -64769.2683
> > > BOND = 79.7941 ANGLE = 500.8945 DIHED =
> > > 1498.9190
> > > 1-4 NB = 453.3799 1-4 EEL = 7256.7348 VDWAALS =
> > > 10279.1689
> > > EELEC = -84838.1594 EHBOND = 0.0000 RESTRAINT =
> > > 0.0000
> > > Ewald error estimate: 0.8471E-04
> > >
> >
> ------------------------------------------------------------------------------
> > > *
> > >
> > > Please help. I really have no idea what is going on. I think the
> > > "ndm_md1.rst" had not been created because the process stopped
> > unexpectedly.
> > >
> > > Thank you.
> > > Chinsu
> > >
> > >
> > >
> > >
> > > On Mon, May 16, 2011 at 7:46 PM, Chinh Su Tran To
> > > <chinh.sutranto.gmail.com>wrote:
> > >
> > >> Dear Jason,
> > >>
> > >> This is my batch file:
> > >>
> > >> */usr/local/amber10/exe/sander -O -i ./file-in/wat-min1.in -o
> > >> ./results/ndm_min1.out -p ./results/ndm-wat.prmtop -c
> > >> ./results/ndm-wat.inpcrd -r ./results/ndm_min1.rst -ref
> > >> ./results/ndm-wat.inpcrd
> > >>
> > >> /usr/local/amber10/exe/sander -O -i ./file-in/wat-min2.in -o
> > >> ./results/ndm_min2.out -p ./results/ndm-wat.prmtop -c
> > ./results/ndm_min1.rst
> > >> -r ./results/ndm_min2.rst
> > >>
> > >>
> > >> /usr/local/amber10/exe/sander -O -i ./file-in/wat-md1.in -o
> > >> ./results/ndm_md1.out -p ./results/ndm-wat.prmtop -c
> > ./results/ndm_min2.rst
> > >> -r ./results/ndm_md1.rst -x ./results/ndm_md1.mdcrd -ref
> > >> ./results/ndm_min2.rst
> > >>
> > >> /usr/local/amber10/exe/sander -O -i ./file-in/wat-md2.in -o
> > >> ./results/ndm_md2.out -p ./results/ndm-wat.prmtop -c
> > ./results/ndm_md1.rst
> > >> -r ./results/ndm_md2.rst -x ./results/ndm_md2.mdcrd
> > >>
> > >> /usr/local/amber10/exe/ambpdb -p ./results/ndm-wat.prmtop <
> > >> ./results/ndm_md2.rst > ./results/ndm_md2.pdb*
> > >>
> > >> And the part below is the error I got:
> > >>
> > >>
> > >> **/var/spool/torque/mom_priv/*
> > >> *jobs/5731.bircibm.SC <http://5731.bircibm.sc/>: line 17: 14885
> > >>> Segmentation fault /usr/local/amber10/exe/sander -O -i
> ./file-in/
> > >>> wat-md1.in -o ./results/ndm_md1.out -p ./results/ndm-wat.prmtop -c
> > >>> ./results/ndm_min2.rst -r ./results/ndm_md1.rst -x
> > >> ./results/ndm_md1.mdcrd
> > >>> -ref ./results/ndm_min2.rst
> > >>>
> > >>> Unit 30 Error on OPEN:
> > >>> ./results/ndm_md1.rst
> > >>> /var/spool/torque/mom_priv/**jobs/5731.bircibm.SC<
> > http://5731.bircibm.sc/>:
> > >> line 25:
> > >>> ./results/ndm_md2.rst: No such file or directory
> > >>
> > >> *
> > >> I double-checked and found that the ndm_md1.rst was not created,
> causing
> > >> the next command was not performed, so the second error led to "could
> > not
> > >> open the ndm_md2.rst"
> > >>
> > >> I viewed other threads from Amber archive titled "segmentation fault"
> > >> saying about "molecular size", but i dont think this error was
> addressed
> > >> because i ran the similar batch file for other larger molecules and
> they
> > ran
> > >> fine.
> > >>
> > >> Is it possible that it was related to the TIP3PBOX of solvent I used?
> I
> > >> used TIP3PBOX 12 for this molecule.
> > >>
> > >> Thanks for your reply, and hope more help.
> > >>
> > >> Chinsu
> > >>
> > >>
> > >> On Mon, May 16, 2011 at 7:27 PM, Jason Swails <jason.swails.gmail.com
> > >wrote:
> > >>
> > >>> On Mon, May 16, 2011 at 6:42 AM, Chinh Su Tran To
> > >>> <chinh.sutranto.gmail.com>wrote:
> > >>>
> > >>>> Dear Amber users,
> > >>>>
> > >>>> Below is the error I encountered when running a batch file of sander
> > for
> > >>> MD
> > >>>> runs following 2-step minimization:
> > >>>>
> > >>>> */var/spool/torque/mom_priv/jobs/5731.bircibm.SC: line 17: 14885
> > >>>> Segmentation fault /usr/local/amber10/exe/sander -O -i
> ./file-in/
> > >>>> wat-md1.in -o ./results/ndm_md1.out -p ./results/ndm-wat.prmtop -c
> > >>>> ./results/ndm_min2.rst -r ./results/ndm_md1.rst -x
> > >>> ./results/ndm_md1.mdcrd
> > >>>> -ref ./results/ndm_min2.rst
> > >>>>
> > >>>> Unit 30 Error on OPEN:
> > >>>> ./results/ndm_md1.rst
> > >>>> /var/spool/torque/mom_priv/jobs/5731.bircibm.SC: line 25:
> > >>>> ./results/ndm_md2.rst: No such file or directory
> > >>>> *
> > >>>>
> > >>>
> > >>> I don't think you copied everything over correctly. Here it says
> that
> > >>> ./results/ndm_md2.rst doesn't exist, but I don't see anywhere above
> > where
> > >>> you used that file.
> > >>>
> > >>>
> > >>>> could anyone please give me a hint about this? I just dont
> understand
> > >>> why
> > >>>> it
> > >>>> happened 'cause it ran fine for other molecules before. And why
> didnt
> > it
> > >>>> occur during minimization?
> > >>>>
> > >>>> Thank you very much, and really looking forward to your reply.
> > >>>>
> > >>>> Regards,
> > >>>> Chinsu
> > >>>>
> > >>>> _______________________________________________
> > >>>> AMBER mailing list
> > >>>> AMBER.ambermd.org
> > >>>> http://lists.ambermd.org/mailman/listinfo/amber
> > >>>>
> > >>>>
> > >>>
> > >>>
> > >>> --
> > >>> Jason M. Swails
> > >>> Quantum Theory Project,
> > >>> University of Florida
> > >>> Ph.D. Candidate
> > >>> 352-392-4032
> > >>> _______________________________________________
> > >>> AMBER mailing list
> > >>> AMBER.ambermd.org
> > >>> http://lists.ambermd.org/mailman/listinfo/amber
> > >>>
> > >>
> > >>
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed May 18 2011 - 21:00:03 PDT
