[gmx-users] surprising: what's the reason?
Chetan Mahajan
chetanvm10 at gmail.com
Mon Mar 17 09:56:22 CET 2014
I am a bit perplexed at this data about the box vector error I mentioned in
first email:
nstxout 1000 10 1 10 1000 1000 1000 1 nstvout 1000 10 1 10 1000 1000 1000
1 nstlog 1000 10 1 10 1000 1000 1000 1 nstenergy 1000 10 1 10 1000 1000
1000 10 nstxtcout 1000 10 1 10 1000 1000 1000 1 nstlist 10 10 1 1 1
5 2 10
error yes yes no
no no yes
no no
Columns represent values chosen in different runs for the options on
extreme left. Row at the bottom indicate if error occurred in that row. As
we see, for the runs for which nstlist was 1 or 2, no error was observed.
Now, when nstlist is 10 (column 2 and column 8), conflicting results were
obtained. From column 2 and 8, we observe that just making nstxout,
nstvout, nstlog, nstxtcout = 1, no error is observed. Why should error
depend on these options?
Thanks
Chetan
On Mon, Mar 17, 2014 at 1:22 AM, Chetan Mahajan <chetanvm10 at gmail.com>wrote:
> Hi Mark,
>
> Could you clarify what is meant by "gentle" and why does it help solve the
> issue?
>
> Thanks
> Chetan
>
>
> On Sun, Mar 16, 2014 at 8:29 AM, Mark Abraham <mark.j.abraham at gmail.com>wrote:
>
>> Ja. Rebuilding the neighbor list more often is more gentle, which often
>> helps. The output period has nothing directly to do with it.
>>
>> Mark
>> On Mar 16, 2014 8:18 AM, "Tsjerk Wassenaar" <tsjerkw at gmail.com> wrote:
>>
>> > Hi Chetan,
>> >
>> > nstenergy = 1 requires nstcalcenergy = 1, which requires nstlist = 1.
>> You
>> > might want to try setting only the latter, and check nstpcoupl.
>> >
>> > Hope it helps ,
>> >
>> > Tsjerk
>> > On Mar 16, 2014 7:17 AM, "Chetan Mahajan" <chetanvm10 at gmail.com> wrote:
>> >
>> > > Hi Mark, Tsjerk and others interested,
>> > >
>> > > I changed the frequency of nstxout, nstvout, nstlog, nstenergy,
>> nstxtcout
>> > > in equilibration run from 1000 to 1 and Whoa! the entire code (even if
>> > the
>> > > frequency of these in production run is still 1000) is running fine
>> now.
>> > > That box vector error is gone. I wonder what's going on, but I am
>> happy
>> > > code is working. I have confirmed this by reproducing this run yet
>> again.
>> > > ( Code still gives box vector error when output frequency is 1000
>> > instead
>> > > of 1 ).
>> > >
>> > > Thanks
>> > > Chetan
>> > >
>> > >
>> > > On Wed, Mar 12, 2014 at 4:20 AM, Mark Abraham <
>> mark.j.abraham at gmail.com
>> > > >wrote:
>> > >
>> > > > Sounds like a normal case of being too rough with a system during
>> > > > equilibration, which can lead to
>> > > > http://www.gromacs.org/Documentation/Terminology/Blowing_Up if you
>> get
>> > > > unlucky. There's general advice there.
>> > > >
>> > > > Mark
>> > > >
>> > > >
>> > > > On Wed, Mar 12, 2014 at 7:05 AM, Chetan Mahajan <
>> chetanvm10 at gmail.com
>> > > > >wrote:
>> > > >
>> > > > > Hi Tsjerk,
>> > > > >
>> > > > > Thanks. It is on the same machine Stampede, same number of nodes
>> and
>> > > > > processors. It does go through minimization part (5000 steps) well
>> > and
>> > > > > gives this error near the start of equilibration run, which comes
>> > after
>> > > > > minimization.
>> > > > >
>> > > > > Thank you.
>> > > > >
>> > > > > regards
>> > > > > Chetan
>> > > > >
>> > > > >
>> > > > > On Wed, Mar 12, 2014 at 12:50 AM, Tsjerk Wassenaar <
>> > tsjerkw at gmail.com
>> > > > > >wrote:
>> > > > >
>> > > > > > Hi Chetan,
>> > > > > >
>> > > > > > Was it on the same machine/node(s)? Same number of processors?
>> How
>> > > many
>> > > > > > steps until the error? Any difference may cause a different
>> > outcome.
>> > > > > >
>> > > > > > Cheers,
>> > > > > >
>> > > > > > Tsjerk
>> > > > > > On Mar 12, 2014 5:42 AM, "Chetan Mahajan" <chetanvm10 at gmail.com
>> >
>> > > > wrote:
>> > > > > >
>> > > > > > > Hi
>> > > > > > >
>> > > > > > > I am testing one MD run using gromacs. 2 days back it ran
>> fine.
>> > > Today
>> > > > > > when
>> > > > > > > I ran the "exactly same" code again, it gave following error:
>> > > > > > >
>> > > > > > >
>> > > > > > > Increasing the number of cell to communicate in dimension X
>> to 3
>> > > for
>> > > > > the
>> > > > > > > first time
>> > > > > > >
>> > > > > > > -------------------------------------------------------
>> > > > > > > Program mdrun_mpi, VERSION 4.6.3
>> > > > > > > Source code file:
>> > > > > > >
>> > > /admin/build/admin/rpms/stampede/BUILD/gromacs-4.6.3/src/mdlib/ns.c,
>> > > > > > line:
>> > > > > > > 2730
>> > > > > > >
>> > > > > > > Fatal error:
>> > > > > > > One of the box vectors has become shorter than twice the
>> cut-off
>> > > > length
>> > > > > > or
>> > > > > > > box_yy-|box_zy| or box_zz has become smaller than the cut-off.
>> > > > > > >
>> > > > > > > I am confused as to why same run should give different
>> answers on
>> > > > > > different
>> > > > > > > occasions. Is it pointing to something wrong with the gromacs
>> > > > > executable
>> > > > > > I
>> > > > > > > have or something else?
>> > > > > > >
>> > > > > > > Thank you.
>> > > > > > >
>> > > > > > > regards
>> > > > > > > Chetan
>> > > > > > > --
>> > > > > > > Gromacs Users mailing list
>> > > > > > >
>> > > > > > > * Please search the archive at
>> > > > > > >
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_Listbefore
>> > > > > > > posting!
>> > > > > > >
>> > > > > > > * Can't post? Read
>> http://www.gromacs.org/Support/Mailing_Lists
>> > > > > > >
>> > > > > > > * For (un)subscribe requests visit
>> > > > > > >
>> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-usersor
>> > > > > > > send a mail to gmx-users-request at gromacs.org.
>> > > > > > >
>> > > > > > --
>> > > > > > Gromacs Users mailing list
>> > > > > >
>> > > > > > * Please search the archive at
>> > > > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_Listbefore
>> > > > > > posting!
>> > > > > >
>> > > > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> > > > > >
>> > > > > > * For (un)subscribe requests visit
>> > > > > >
>> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-usersor
>> > > > > > send a mail to gmx-users-request at gromacs.org.
>> > > > > >
>> > > > > --
>> > > > > Gromacs Users mailing list
>> > > > >
>> > > > > * Please search the archive at
>> > > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_Listbefore
>> > > > > posting!
>> > > > >
>> > > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> > > > >
>> > > > > * For (un)subscribe requests visit
>> > > > >
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-usersor
>> > > > > send a mail to gmx-users-request at gromacs.org.
>> > > > >
>> > > > --
>> > > > Gromacs Users mailing list
>> > > >
>> > > > * Please search the archive at
>> > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> > > > posting!
>> > > >
>> > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> > > >
>> > > > * For (un)subscribe requests visit
>> > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-usersor
>> > > > send a mail to gmx-users-request at gromacs.org.
>> > > >
>> > > --
>> > > Gromacs Users mailing list
>> > >
>> > > * Please search the archive at
>> > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> > > posting!
>> > >
>> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> > >
>> > > * For (un)subscribe requests visit
>> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> > > send a mail to gmx-users-request at gromacs.org.
>> > >
>> > --
>> > Gromacs Users mailing list
>> >
>> > * Please search the archive at
>> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> > posting!
>> >
>> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> >
>> > * For (un)subscribe requests visit
>> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> > send a mail to gmx-users-request at gromacs.org.
>> >
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>>
>
>
More information about the gromacs.org_gmx-users
mailing list