C++ Compiler issue

196 views
Skip to first unread message

AKASH BANERJEE

unread,
Jul 8, 2021, 6:09:31 PM7/8/21
to vo...@googlegroups.com
Hi Votca developers, 

My question: C++ Compiler issue: I am trying to install VOTCA on a supercomputer using spack. I am getting a C++ error. Could you please have a look at the attached log file? I am trying to figure out what my next steps are. 

Response from Christoph:  The issue you found is already fixed (see
https://github.com/votca/tools/pull/361), but not part of a release
yet.
You can install "votca@stable" using spack to get the fix.

Apologies for the test email. 

Thank you 
Akash
output.out

Mohsen Farshad

unread,
Jul 9, 2021, 5:42:12 PM7/9/21
to vo...@googlegroups.com
Hi Votca developers, 

Is there a way to solve this issue without the need for installation of "votca@stable" using spack?

Thanks,
Mohsen
output.out

Christoph Junghans

unread,
Jul 9, 2021, 5:53:11 PM7/9/21
to vo...@googlegroups.com
On Fri, Jul 9, 2021 at 3:42 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Hi Votca developers,
>
> Is there a way to solve this issue without the need for installation of "votca@stable" using spack?
Not really, we really have to make a 2021.1 release and bump the spackage.
Alternatively, you could apply the code change of the patch
(https://github.com/votca/tools/commit/6bb7e35ba7d1a31247eafb323be2777ec0439cfe)
manually.
I also made a quick fix in spack: https://github.com/spack/spack/pull/24815
Once that is merged, you should be good to go.

Christoph

>
> Thanks,
> Mohsen
> ---------- Forwarded message ---------
> From: ab1...@scarletmail.rutgers.edu <Unknown>
> Date: Thursday, July 8, 2021 at 6:09:31 PM UTC-4
> Subject: C++ Compiler issue
> To: vo...@googlegroups.com <Unknown>
>
>
> Hi Votca developers,
>
> My question: C++ Compiler issue: I am trying to install VOTCA on a supercomputer using spack. I am getting a C++ error. Could you please have a look at the attached log file? I am trying to figure out what my next steps are.
>
> Response from Christoph: The issue you found is already fixed (see
> https://github.com/votca/tools/pull/361), but not part of a release
> yet.
> You can install "votca@stable" using spack to get the fix.
>
> Apologies for the test email.
>
> Thank you
> Akash
>
> --
> Join us on Slack: https://join.slack.com/t/votca/signup
> ---
> You received this message because you are subscribed to the Google Groups "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/4ee1f111-852d-4526-9920-55cfe9b110e9n%40googlegroups.com.



--
Christoph Junghans
Web: http://www.compphys.de

Christoph Junghans

unread,
Jul 10, 2021, 10:01:06 AM7/10/21
to vo...@googlegroups.com
On Fri, Jul 9, 2021 at 3:52 PM Christoph Junghans <jung...@votca.org> wrote:
>
> On Fri, Jul 9, 2021 at 3:42 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
> >
> > Hi Votca developers,
> >
> > Is there a way to solve this issue without the need for installation of "votca@stable" using spack?
> Not really, we really have to make a 2021.1 release and bump the spackage.
> Alternatively, you could apply the code change of the patch
> (https://github.com/votca/tools/commit/6bb7e35ba7d1a31247eafb323be2777ec0439cfe)
> manually.
> I also made a quick fix in spack: https://github.com/spack/spack/pull/24815
> Once that is merged, you should be good to go.
The fix is part of spack now, just pull the latest version of spack
and everything should be fine.

Mohsen Farshad

unread,
Jul 15, 2021, 4:34:14 PM7/15/21
to vo...@googlegroups.com
Hi Christoph,

After cloning Spack I inserted the following commands on my own machine and the cluster, but neither machines were able to successfully complete the installation after a relatively long time. 

git clone https://github.com/spack/spack.git spack
source spack/share/spack/setup-env.sh
spack install votca-csg@stable

Do you think there is something wrong with the compiler or I am missing some commands and installation is more complicated than I am thinking of?

I have made sure the Spack is updated.

Looking forward to hearing from you,

Thanks,
Mohsen

Christoph Junghans

unread,
Jul 15, 2021, 4:38:24 PM7/15/21
to vo...@googlegroups.com
On Thu, Jul 15, 2021 at 2:34 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Hi Christoph,
>
> After cloning Spack I inserted the following commands on my own machine and the cluster, but neither machines were able to successfully complete the installation after a relatively long time.
>
> git clone https://github.com/spack/spack.git spack
> source spack/share/spack/setup-env.sh
> spack install votca-csg@stable
>
> Do you think there is something wrong with the compiler or I am missing some commands and installation is more complicated than I am thinking of?

Can you post the error message?

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DpAJGCt-_E9RLcOQOA8QQ0jDcsm4AyHtHVfGN4JMZiqdA%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 15, 2021, 4:42:45 PM7/15/21
to vo...@googlegroups.com
On the cluster, the run is not ending now. In my own machine, it says:

==> Error: votca-csg-stable-t3g3evics7o4vvqtxxunsnx2p67u56bg: Package was not installed

==> Error: Installation request failed.  Refer to reported errors for failing package(s).


The last comment on the cluster is following: 

[+] /home/tuo98996/spack/opt/spack/linux-centos7-haswell/gcc-4.8.5/gromacs-2019.6-xtfitykzurvzarvflpj2v6cu3wywr5fc



Christoph Junghans

unread,
Jul 15, 2021, 7:02:23 PM7/15/21
to vo...@googlegroups.com
Can you run "spack cd -b votca-csg@stable" and look for a .log file in there?

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Dq6vsiEjqxsJGoVLawnA4VwPzLaCbO7zn3Avn2xJoa5rg%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 19, 2021, 5:04:17 PM7/19/21
to vo...@googlegroups.com
Hi Christoph,

I was able to install the Votca using Spack. However, going through the tutorial I confronted an error after executing the following command:

csg_call table integrate CG-CG.force CG-CG.pot

The last lines of the error is as follows:

  #000: /tmp/tuo98996/spack-stage/spack-stage-hdf5-1.10.7-3mcrng3jtezm72xs2i2fbwpwzogxsupf/spack-src/src/H5T.c line 1708 in H5Tcopy(): not a datatype or dataset

    major: Invalid arguments to routine

    minor: Inappropriate type

Running subscript 'table_integrate.pl CG-CG.force CG-CG.pot' (from tags table integrate) dir /home/tuo98996/spack/opt/spack/linux-centos7-broadwell/gcc-9.3.0/votca-csg-stable-7slgbunnjcw6657pwdjeshozxpfzi3d5/share/votca/scripts/invers


The resulting CG table potential is zero for all the distances.


Looking forward to your feedback,


Thanks,

Mohsen




Christoph Junghans

unread,
Jul 19, 2021, 6:34:30 PM7/19/21
to vo...@googlegroups.com
On Mon, Jul 19, 2021 at 3:04 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Hi Christoph,
>
> I was able to install the Votca using Spack. However, going through the tutorial I confronted an error after executing the following command:
>
> csg_call table integrate CG-CG.force CG-CG.pot
>
> The last lines of the error is as follows:
>
> #000: /tmp/tuo98996/spack-stage/spack-stage-hdf5-1.10.7-3mcrng3jtezm72xs2i2fbwpwzogxsupf/spack-src/src/H5T.c line 1708 in H5Tcopy(): not a datatype or dataset
>
> major: Invalid arguments to routine
>
> minor: Inappropriate type
>
> Running subscript 'table_integrate.pl CG-CG.force CG-CG.pot' (from tags table integrate) dir /home/tuo98996/spack/opt/spack/linux-centos7-broadwell/gcc-9.3.0/votca-csg-stable-7slgbunnjcw6657pwdjeshozxpfzi3d5/share/votca/scripts/invers
>
>
> The resulting CG table potential is zero for all the distances.
The integrate script will only act on the first valid point, usually
marked with "i" in the last column of the force table.
So that might be the problem. Otherwise send us the force table and we
can have a look as well.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DqTUn3QnuN1QFwb6UBnouvgbFzAJWfb9YeuL8sPNc7LvQ%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 19, 2021, 7:47:16 PM7/19/21
to vo...@googlegroups.com
Here are the potential, force, and output potentials. Would you please let me know whether they are correct? 

Thank you,
Mohsen



--
Mohsen Farshad
Ph.D. Chemistry
University of Maine
CG-CG.pot
CG-CG.force
table_CG_CG.xvg

Christoph Junghans

unread,
Jul 19, 2021, 10:38:16 PM7/19/21
to vo...@googlegroups.com
On Mon, Jul 19, 2021 at 5:47 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Here are the potential, force, and output potentials. Would you please let me know whether they are correct?
The outcomes are actually correct.

Your forces and the potential ends at 0.9nm and so is the xvg file.
The values smaller than 0.9nm are not zero.
We shift the potential in such a way that it is zero at the cutoff,
hence all values for r> 0.9nm are zero.
We put the values in the C12 column of the gromacs xvg table, hence
columns 2-5 are empty and only columns 6+7 have numbers.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Dr9PXU%3D7j4%3D0TP-azBAHucna2WoWpZxz%3DCGqReCrGnaFA%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 20, 2021, 10:47:16 AM7/20/21
to vo...@googlegroups.com
Thanks for the explanation. I am a bit confused. Is there any repulsion term in the Votca potential? Would you please explain what kind of potential is that?

Also in SPCE tutorial, to run CG simulation, it is suggested to use the files for ibi example. I have a few questions in regards to this part of the tutorial:

1. What command should I use to run coarse grain:
"gmx grompp"  and then "gmx mdrun -v"
or "csg_inverse --options settings.xml"
1. Would you explicitly explain how should I adjust the settings.xml to make it compatible with the potential table that comes from FM. The current IBI setting.xml uses IBI interactive algorithm. Is that needed to run CG simulation using FM potential? --- if setting.xml is not needed please ignore this question.

3. For CG simulation the CG configuration should be used or atomistic one?

Thank you,
Mohsen


Christoph Junghans

unread,
Jul 20, 2021, 5:07:31 PM7/20/21
to vo...@googlegroups.com
On Tue, Jul 20, 2021 at 8:47 AM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Thanks for the explanation. I am a bit confused. Is there any repulsion term in the Votca potential? Would you please explain what kind of potential is that?

Gromacs just has the C6 and C12 tables columns, but in the case of
VOTCA we just add the C6 and C12 parts together and you can make one
of them zero.
(see https://manual.gromacs.org/documentation/2019.5/reference-manual/special/tabulated-interaction-functions.html#user-specified-potential-functions)
Basically because it is not easy to separate the C6 from the C12 part
and also because you can do some funky things by using the C6 part for
something else.

>
> Also in SPCE tutorial, to run CG simulation, it is suggested to use the files for ibi example. I have a few questions in regards to this part of the tutorial:
>
> 1. What command should I use to run coarse grain:
>
> "gmx grompp" and then "gmx mdrun -v"
This runs a MD simulation directly only use that after coarse-graining
or for atomistic simulation.

>
> or "csg_inverse --options settings.xml"
This is the right command to do IBI

>
> 1. Would you explicitly explain how should I adjust the settings.xml to make it compatible with the potential table that comes from FM. The current IBI setting.xml uses IBI interactive algorithm. Is that needed to run CG simulation using FM potential? --- if setting.xml is not needed please ignore this question.

FM is a different CG method than IBI. FM is one shot and done, while
IBI is iterative.
You could use the FM potential as a starting potential for IBI, but
this is kind of pointless as FM matches the forces, while IBI matches
the structure.

>
> 3. For CG simulation the CG configuration should be used or atomistic one?
The CG one.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Doyh8wWXaPFoYxHtMPdnrMDA%2BjtBhnPmY3nE9X60asS2g%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 21, 2021, 11:56:30 AM7/21/21
to vo...@googlegroups.com
Thanks for the explanation. I have still some difficulty understanding the process, and I have the following questions:

1. Are C6 and C12 are literally summed up ("+") or they are just combined during the FM process --- essentially they have not ever gotten separated?

2. Does the FM technique aim to extract a table potential for CG system from the atomistic system with matching the forces? Can we use this extracted FM potential table for CG simulations --- not IBI, just normal CG MD simulation? If so would you please let me know how can I use this potential table to run CG simulations and what files are needed for that? 

Thank you,
Mohsen

Mohsen Farshad

unread,
Jul 21, 2021, 2:45:52 PM7/21/21
to vo...@googlegroups.com
Hi Christoph,

Is this correction answer to my question number 2, "Does the FM technique aim to extract a table potential for CG system from the atomistic system with matching the forces? Can we use this extracted FM potential table for CG simulations --- not IBI, just normal CG MD simulation? If so would you please let me know how can I use this potential table to run CG simulations and what files are needed for that?":

gmx grompp -f grompp.mdp - p topol.top -c confout.gro
gmx mdrun -table table_CG_CG.xvg

where confout.gro is GC configuration file, and table_CG_CG.xvg is the potential table extracted from FM technique. 

Thanks,
Mohsen

Christoph Junghans

unread,
Jul 21, 2021, 5:08:08 PM7/21/21
to vo...@googlegroups.com
On Wed, Jul 21, 2021 at 9:56 AM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Thanks for the explanation. I have still some difficulty understanding the process, and I have the following questions:
>
> 1. Are C6 and C12 are literally summed up ("+") or they are just combined during the FM process --- essentially they have not ever gotten separated?
Yes, FM has no mechanism to distinguish them.

>
> 2. Does the FM technique aim to extract a table potential for CG system from the atomistic system with matching the forces? Can we use this extracted FM potential table for CG simulations --- not IBI, just normal CG MD simulation? If so would you please let me know how can I use this potential table to run CG simulations and what files are needed for that?

See my answer in the next email.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Dp%2BeLuCARbwoHJZQm3yLTEQ57S%2BTkb%2ByjXsGdm37kQKdQ%40mail.gmail.com.

Christoph Junghans

unread,
Jul 21, 2021, 5:09:53 PM7/21/21
to vo...@googlegroups.com
On Wed, Jul 21, 2021 at 12:45 PM Mohsen Farshad
<mohsen....@maine.edu> wrote:
>
> Hi Christoph,
>
> Is this correction answer to my question number 2, "Does the FM technique aim to extract a table potential for CG system from the atomistic system with matching the forces? Can we use this extracted FM potential table for CG simulations --- not IBI, just normal CG MD simulation? If so would you please let me know how can I use this potential table to run CG simulations and what files are needed for that?":
>
> gmx grompp -f grompp.mdp - p topol.top -c confout.gro
> gmx mdrun -table table_CG_CG.xvg
>
> where confout.gro is GC configuration file, and table_CG_CG.xvg is the potential table extracted from FM technique.
Yes that that is correct, but can also use the mapped atomistic
conf.gro instead of confout.gro

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DpT89%2BR7m3%3DoFnT_JiEtV6Kx%3DuNSqohEZ5AW8bEBiFygg%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 21, 2021, 5:15:45 PM7/21/21
to vo...@googlegroups.com
Thank you very much! I have still a single remaining confusion.

Does this mean if I use atomistic configuration and topology files with GC potential table, the system automatically runs coarse-grained simulation using the following commands:

gmx grompp -f grompp.mdp - p topol.top -c conf.gro
gmx mdrun -table table_CG_CG.xvg

where conf.gro and topol are atomistic configuration and topology files, and table_CG_CG.xvg is the potential table extracted from FM technique.

Thanks,
Mohsen 



Christoph Junghans

unread,
Jul 21, 2021, 5:25:18 PM7/21/21
to vo...@googlegroups.com
On Wed, Jul 21, 2021 at 3:15 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> Thank you very much! I have still a single remaining confusion.
>
> Does this mean if I use atomistic configuration and topology files with GC potential table, the system automatically runs coarse-grained simulation using the following commands:

No you will have to use a cg topology and configuration file.
For the former you can csg_ gmxtopol, which will generate a good starting point.
For the latter, you can just map the atomistic configuration file.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DpX7MN5NOxQC5SBbQfKESFyVt0iL_jb0sD4%3DVb3%2BMh%2B6w%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 21, 2021, 9:20:00 PM7/21/21
to vo...@googlegroups.com
This means I have to use the following command to make the CG topology file:

"csg_gmxtopol --top topol.top --cg water.xml --out topol_CG.top"

where water.xml and topol.top can be found in the SPC/E tutorial in the atomistic folder.

and for making the configuration file I can use the center of mass of water molecules which is made by mapping command from atomistic SPC/E using the following commands:

echo "Mapping confout.gro to get the starting configuration for coarse-grain

csg_map --top topol.tpr --trj conf.gro --cg water.xml --out conf_cg.gro


In the same way, it is asking the user to do it in the tutorial.

Please let me know if any of the above commands or relevant explanation to create the topology and configuration of the CG system is wrong.

Thanks,
Mohsen

Christoph Junghans

unread,
Jul 21, 2021, 9:32:38 PM7/21/21
to vo...@googlegroups.com
On Wed, Jul 21, 2021 at 19:20 Mohsen Farshad <mohsen....@maine.edu> wrote:
This means I have to use the following command to make the CG topology file:

"csg_gmxtopol --top topol.top --cg water.xml --out topol_CG.top"

where water.xml and topol.top can be found in the SPC/E tutorial in the atomistic folder.

and for making the configuration file I can use the center of mass of water molecules which is made by mapping command from atomistic SPC/E using the following commands:

echo "Mapping confout.gro to get the starting configuration for coarse-grain

csg_map --top topol.tpr --trj conf.gro --cg water.xml --out conf_cg.gro


In the same way, it is asking the user to do it in the tutorial.

Please let me know if any of the above commands or relevant explanation to create the topology and configuration of the CG system is wrong.
All correct, you might have to check and hand-edit the generated topol.top.

For spc/e you could also use the topology from the ibi tutorial.

Christoph 

Mohsen Farshad

unread,
Jul 21, 2021, 10:26:13 PM7/21/21
to vo...@googlegroups.com
After I run the following command:

csg_gmxtopol --top topol.tpr --cg water.xml --out topol_cg

the following topol_cg is generated:

  SOL 3

   

  [atoms]

  ; nr type resnr residue atom cgnr charge mass

  1 CG 1 RES CG 1 0.000000 18.015400 

Is this incomplete? Is the attached spc/e topology file from the IBI folder belong to the coarse-graining spc/e? If so do you have any insight in regards to why the above topology file is not the same as the attached file?

Thanks,
Mohsen



topol.top

Christoph Junghans

unread,
Jul 22, 2021, 9:05:33 AM7/22/21
to vo...@googlegroups.com
On Wed, Jul 21, 2021 at 8:26 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> After I run the following command:
>
> csg_gmxtopol --top topol.tpr --cg water.xml --out topol_cg
>
> the following topol_cg is generated:
>
> SOL 3
>
>
>
> [atoms]
>
> ; nr type resnr residue atom cgnr charge mass
>
> 1 CG 1 RES CG 1 0.000000 18.015400
>
> Is this incomplete? Is the attached spc/e topology file from the IBI folder belong to the coarse-graining spc/e? If so do you have any insight in regards to why the above topology file is not the same as the attached file?
Yes, it is incomplete, the help of csg_gmxtopol says it all:
"Create skeleton for gromacs topology based on atomistic topology
and a mapping file. File still needs to be modified by the user."

IBI is basically running the cg simulation over and over again without
knowing the potential, hence the topology from the tutorials is just
the cg one.

csg_gmxtopol can only deal with one molecule type right now (patches
welcome), hence the generated topology is missing some block incl.
defaults, atomtypes, system, molecules and maybe other funky things
you want to do in the cg simulation.

Please have a look at the gromacs manual on how to write a topology
file, it is really not that hard.
https://manual.gromacs.org/documentation/current/reference-manual/topologies/topology-file-formats.html#topology-file
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DoZRS_df_i-rwGQn0wwz9gXz3UmKt8gtnzib%3DrPpHZTBQ%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 22, 2021, 12:39:03 PM7/22/21
to vo...@googlegroups.com
Thank you very much. I am running the CG simulation using the FM potential table obtained through the force-matching process in the force_matching folder and grompp.mdp from the atomistic folder. I receive the below error after the following commands:

gmx grompp -f grompp.mdp -p topol.top -c conf_cg.gro

gmx mdrun -table table_CG_CG.xvg


Fatal error:

An atom moved too far between two domain decomposition steps

This usually means that your system is not well equilibrated


I attach all the files here.

Thanks,
Mohsen

 

md.log
grompp.mdp
conf_cg.gro
table_CG_CG.xvg
topol.top

Christoph Junghans

unread,
Jul 22, 2021, 2:41:18 PM7/22/21
to vo...@googlegroups.com
On Thu, Jul 22, 2021 at 10:39 AM Mohsen Farshad
<mohsen....@maine.edu> wrote:
>
> Thank you very much. I am running the CG simulation using the FM potential table obtained through the force-matching process in the force_matching folder and grompp.mdp from the atomistic folder. I receive the below error after the following commands:
>
> gmx grompp -f grompp.mdp -p topol.top -c conf_cg.gro
>
> gmx mdrun -table table_CG_CG.xvg
>
>
> Fatal error:
>
> An atom moved too far between two domain decomposition steps
>
> This usually means that your system is not well equilibrated
Yeah, this error basically means your system exploded, I would try to
run a minimization first.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DphKMo202XFNH3L7Ywh5r%3D6JP%2BBnB_TzhRmfVQPpVzVmg%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 23, 2021, 11:03:17 AM7/23/21
to vo...@googlegroups.com
Hi Christoph, 

I use the following .mdp gromacs script adjusted from the gromacs tutorial to minimize the our CS system:

  1 ; minim.mdp - used as input into grompp to generate em.tpr

  2 ; Parameters describing what to do, when to stop and what to save

  3 integrator  = steep         ; Algorithm (steep = steepest descent minimization)

  4 emtol       = 1000.0        ; Stop minimization when the maximum force < 1000.0 kJ/mol/nm

  5 emstep      = 0.001          ; Minimization step size

  6 nsteps      = 100000         ; Maximum number of (minimization) steps to perform

  7 

  8 ; Parameters describing how to find the neighbors of each atom and how to calculate the interactions

  9 nstlist         = 1        ; Frequency to update the neighbor list and long range forces

 10 ns_type         = grid      ; Method to determine neighbor list (simple, grid)

 11 coulombtype     = Reaction-Field      ; Treatment of long range electrostatic interactions

 12 rcoulomb        = 1.0       ; Short-range electrostatic cut-off

 13 rvdw            = 1.0       ; Short-range Van der Waals cut-off

 14 pbc             = xyz       ; Periodic Boundary Conditions in all 3 dimensions


and I get the following error:


Steepest Descents converged to machine precision in 8003 steps,

but did not reach the requested Fmax < 1000.


Is this the right way to minimize the system? If so is there something wrong with the initial configuration?

Thanks,
Mohsen


Christoph Junghans

unread,
Jul 23, 2021, 9:04:00 PM7/23/21
to vo...@googlegroups.com
On Fri, Jul 23, 2021, 09:03 Mohsen Farshad <mohsen....@maine.edu> wrote:
Hi Christoph, 

I use the following .mdp gromacs script adjusted from the gromacs tutorial to minimize the our CS system:

  1 ; minim.mdp - used as input into grompp to generate em.tpr

  2 ; Parameters describing what to do, when to stop and what to save

  3 integrator  = steep         ; Algorithm (steep = steepest descent minimization)

  4 emtol       = 1000.0        ; Stop minimization when the maximum force < 1000.0 kJ/mol/nm

  5 emstep      = 0.001          ; Minimization step size

  6 nsteps      = 100000         ; Maximum number of (minimization) steps to perform

  7 

  8 ; Parameters describing how to find the neighbors of each atom and how to calculate the interactions

  9 nstlist         = 1        ; Frequency to update the neighbor list and long range forces

 10 ns_type         = grid      ; Method to determine neighbor list (simple, grid)

 11 coulombtype     = Reaction-Field      ; Treatment of long range electrostatic interactions

 12 rcoulomb        = 1.0       ; Short-range electrostatic cut-off

 13 rvdw            = 1.0       ; Short-range Van der Waals cut-off

 14 pbc             = xyz       ; Periodic Boundary Conditions in all 3 dimensions


and I get the following error:


Steepest Descents converged to machine precision in 8003 steps,

but did not reach the requested Fmax < 1000.


Is this the right way to minimize the system? If so is there something wrong with the initial configuration?
The mdp looks right! The force limit isn't super important as long as it is converted. I would go ahead now and give the cg run another try with the minimized configuration as starting point.

Christoph


Mohsen Farshad

unread,
Jul 24, 2021, 1:51:25 PM7/24/21
to vo...@googlegroups.com
Thanks. I tried to do simulation with the minimized structure using FM potential table. But the problem remains with the same following error:

Fatal error:

An atom moved too far between two domain decomposition steps

This usually means that your system is not well equilibrated



I tried different things including using IBI potential table, minimizing the minimized structure, changing the parameters in the .mdp, but none of them solved the problem. I must mention that I pretty new to the Gromacs and Votca. I attach the files here, that would be a great if you could look at the files. Are you able to run this?

em.gro is the minimized structure and conf_cg.gro is the initial structure of spc/e model.

Thanks,
Mohsen



topol.top
grompp.mdp
conf_cg.gro
em.gro
table_CG_CG.xvg

David Rosenberger

unread,
Jul 25, 2021, 12:37:24 PM7/25/21
to votca
Hi Mohsen,

can you please change the integrator in the mdp file from "md" to "sd'?

It seems you don't have a thermostat.
Your integrator is "md", and you set "Tcouple=no". This means you don't have a thermostat. You don't need to define a thermostat if you use the "sd" integrator.
"SD" has some friction term in the equations of motions, this has the same effect as setting thermostat, given a proper value for tau_t is set. For more please see the GROMACS manual.

General remark: For coarse grained systems it is better to use Langevin-type equations of motions (e.g. the SD integrator in GROMACS), as you loose friction between particles due to the coarsening.

Best,
David

Mohsen Farshad

unread,
Jul 25, 2021, 2:47:38 PM7/25/21
to vo...@googlegroups.com
Hi David,

Thanks for the informative explanation on the equation of motion, I had not realized that. 

I changed the integrator from md to sd and confronted a similar error:

Fatal error:

An atom moved too far between two domain decomposition steps

This usually means that your system is not well equilibrated


Thanks,

Mohsen

David Rosenberger

unread,
Jul 26, 2021, 4:01:14 AM7/26/21
to votca
Hi Mohsen,
by having a look at the md.log file, I think the issue is the following:
--------
generated table with 1186 data points for RF-zero.
Tabscale = 500 points/nm
Generated table with 1186 data points for LJ6.
Tabscale = 500 points/nm
Generated table with 1186 data points for LJ12.
Tabscale = 500 points/nm
--------
Gromacs does not read your table properly, but insteads generates a generic one.

Could you copy the table.xvg file from e.g. the spce/ibi example to your working directory and run "mdrun -table", instead of "mdrun -table table_CG_CG.xvg?
Note: You need both files: table.xvg and table_CG_CG.xvg.

I tested your files with gromacs 2019v2, and it runs without any problems.

Though I had to generate an index file containing the "CG" atoms,  and I had to add "cutoff-scheme=group" to the mdp file.

Best,
David

Mohsen Farshad

unread,
Jul 26, 2021, 10:34:34 AM7/26/21
to vo...@googlegroups.com
Hi David,

Thank you. I did not realized that those lines indicate the Gromacs is not reading the potential table.

Did you use the following command to generate the index file:

gmx make_ndx -f em.gro -o index.ndx

I added table.xvg from IBI to and changed the command to gmx mdrun with both tables in the current folder. I added the cutoff_scheme=group to grompp.mpd but still got the same error with different versions of gromacs. Would you please send me the files that you used to run the system?

Also, for clarification, I want to use force-matching potential to run the system rather than IBI potential.

Looking forward to hearing from you,

Thanks,
Mohsen

David Rosenberger

unread,
Jul 26, 2021, 11:16:43 AM7/26/21
to votca
Hi Mohsen,

I used your files. Did you re-run grompp?

Here's what I did:
gmx make_ndx -f conf_cg.gro and then for selection:  a 'CG'.

Then gmx grompp -f conf_cg.gro -n index.ndx.

Then gmx mdrun -table.

If you do that, then check what is written in the md.log file and see which table file is used.

If it crashes, it doesn't use the correct table file. 'table.xvg' is just a dummy file, which gromacs is looking for.
It then automatically takes table_CG_CG.xvg for the CG_CG interaction defined in energy groups.

Best,
David

Mohsen Farshad

unread,
Jul 26, 2021, 11:45:42 AM7/26/21
to vo...@googlegroups.com
When I do this "gmx make_ndx -f conf_cg.gro" I receive the following in the command line without any index file generated. Do you know the reason for this? Also I do not understand what does "select CG mean".

  gmx make_ndx -f conf_cg.gro



Reading structure file

Going to read 0 old index file(s)

Analysing residue names:

There are:  2180      Water residues


  0 System              :  2180 atoms

  1 Water               :  2180 atoms

  2 SOL                 :  2180 atoms


 nr : group      '!': not  'name' nr name   'splitch' nr    Enter: list groups

 'a': atom       '&': and  'del' nr         'splitres' nr   'l': list residues

 't': atom type  '|': or   'keep' nr        'splitat' nr    'h': help

 'r': residue              'res' nr         'chain' char

 "name": group             'case': case sensitive           'q': save and quit

 'ri': residue index


>


However, I copied the index file from IBI folder to the current folder and ran the following commands:

gmx grompp -f grompp.mdp -c em.gro -n index.ndx 

gmx mdrun -table


Would you please let me know how should I look for table in md.log. I could not find any.xvg file in the md.log. I attach it here.

Thanks,
Mohsen


md.log

Mohsen Farshad

unread,
Jul 26, 2021, 11:47:10 AM7/26/21
to vo...@googlegroups.com
I forgot to mention, the system crashed with the same error. I must be missing something very simple yet frustrating.

David Rosenberger

unread,
Jul 26, 2021, 11:51:53 AM7/26/21
to votca
Gromacs 2020 doesn't support tabulated potentials. You have to switch to a older version.
Maybe it's back in Gromacs 2021, but 2020 cannot be used.

Mohsen Farshad

unread,
Jul 26, 2021, 11:53:32 AM7/26/21
to vo...@googlegroups.com
Okay, I figured how to save the index file sorry for a simple question. I will test it now with an older version of gromacs.

You received this message because you are subscribed to a topic in the Google Groups "votca" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/votca/vdLCQArut2g/unsubscribe.
To unsubscribe from this group and all its topics, send an email to votca+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/971a3961-d790-4833-b14a-5302b6df336en%40googlegroups.com.

David Rosenberger

unread,
Jul 26, 2021, 11:57:46 AM7/26/21
to votca
For the index file: You have to make a selection. If you press 'q' the index file will be generated, with the three selections presented above.
You have to type 'a CG' to add the CG atoms to the index file, if you then press 'q' a index file with  4 selctions will be generated.

Maybe it's advisory for you to go carefully through a gromacs tutorial and read the corresponding sections in the manual
before continuing.

Christoph Junghans

unread,
Jul 26, 2021, 12:16:50 PM7/26/21
to vo...@googlegroups.com
gromacs-2020 and never should error out if you are using cutoff_scheme=group.
When using csg_inverse VOTCA will automatically add that line to the
mdp file, hence it is not in the spce tutorials mdp file.

Christoph

On Mon, Jul 26, 2021 at 9:57 AM 'David Rosenberger' via votca
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/8aa80bd5-a159-4bd8-80f0-3b25b998addan%40googlegroups.com.

Mohsen Farshad

unread,
Jul 26, 2021, 12:20:10 PM7/26/21
to vo...@googlegroups.com
Thank you very much for your replies. The system is running. I checked what potential is the system using, here is a portion of .log file:

 361 Read user tables from table_CG_CG.xvg with 2501 data points.

 362 Tabscale = 1000 points/nm

 363 Read user tables from table.xvg with 1501 data points.

 364 Tabscale = 500 points/nm

 365 

 366 NOTE: All elements in table table.xvg are zero


Does "All elements in table table.xvg are zero" The table.xvg file is treated like a dummy file and the system is actually using "table_CG_CG.xvg"?



Christoph Junghans

unread,
Jul 26, 2021, 1:02:00 PM7/26/21
to vo...@googlegroups.com
On Mon, Jul 26, 2021 at 10:20 AM Mohsen Farshad
<mohsen....@maine.edu> wrote:
>
> Thank you very much for your replies. The system is running. I checked what potential is the system using, here is a portion of .log file:
>
> 361 Read user tables from table_CG_CG.xvg with 2501 data points.
>
> 362 Tabscale = 1000 points/nm
>
> 363 Read user tables from table.xvg with 1501 data points.
>
> 364 Tabscale = 500 points/nm
>
> 365
>
> 366 NOTE: All elements in table table.xvg are zero
>
>
> Does "All elements in table table.xvg are zero" The table.xvg file is treated like a dummy file and the system is actually using "table_CG_CG.xvg"?

table_CG_CG.xvg is used for all atoms between the CG and CG groups.
And table.xvg for everything else, but as everything is covered under
the CG-CG part you should be good.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Dr8TqRGpcxp7Va%3Dxdjai_B6SJ5jpkfZOJwO9hFJ06UPqw%40mail.gmail.com.

Mohsen Farshad

unread,
Jul 26, 2021, 1:31:02 PM7/26/21
to vo...@googlegroups.com
Hi Christoph,

Thanks. Would you please let me know how you realize the system is using table_CG_CG.xvg rather than table.xvg for the CG system by looking at those lines in .log file?

Christoph Junghans

unread,
Jul 26, 2021, 1:51:23 PM7/26/21
to vo...@googlegroups.com
On Mon, Jul 26, 2021 at 11:31 AM Mohsen Farshad
<mohsen....@maine.edu> wrote:
>
> Hi Christoph,
>
> Thanks. Would you please let me know how you realize the system is using table_CG_CG.xvg rather than table.xvg for the CG system by looking at those lines in .log file?
You can usually see that when you create the index file, the number of
atoms in CG should be the same as in the "system" group.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4DpxnNcK2AOntm-EYQUOzGjq40QeF1u%3D-%3D38KqGd%3Dshw-Q%40mail.gmail.com.

David Rosenberger

unread,
Jul 26, 2021, 1:56:39 PM7/26/21
to votca
Here's a more extended explanation for tables and gromacs: https://www.gromacs.org/@api/deki/files/94/=gromacs_nb.pdf
It's for a very old version, but still contains all important information.

Mohsen Farshad

unread,
Jul 26, 2021, 4:27:27 PM7/26/21
to vo...@googlegroups.com
David and Christoph,

I read the relevant parts of the pdf that David sent us. Thank you for that. My brain neurons are getting wired up, however, I need to clarify my understandings, therefore that would be great if you correct me if my following interpretations are incorrect or imprecise:

Potential table coordinate with both .gro and .ndx files. Looks at index file for separating the potential tables and also looks at .gro for specifying the none_bonded interactions by finding the names of particles in .gro. 

The user tabulated table should have a specific name like table_CG_CG that which in this case determines the interaction between CG and CG in .gro file. If the system has more atoms than defined in CG list then the Gromacs looks for parameters from a default potential table called table.xvg which is in my case nothing as the whole system is CG. It is not clear in PDF where this table.xvg comes from if it is not defined by USER.

The following two lines in mdp file

Read user tables from table_CG_CG.xvg with 2501 data points.

Tabscale = 1000 points/nm

Read user tables from table.xvg with 1501 data points.

Tabscale = 500 points/nm


The first table is specific to the CG particles and the second one is for system particles, and because the whole system in my case is CG elements the table.xvg will remain without any use. Therefore, it does not matter what it contains as long as it is following the format of a Gromacs potential table.

Thanks,
Mohsen










Christoph Junghans

unread,
Jul 26, 2021, 5:26:57 PM7/26/21
to vo...@googlegroups.com
On Mon, Jul 26, 2021 at 2:27 PM Mohsen Farshad <mohsen....@maine.edu> wrote:
>
> David and Christoph,
>
> I read the relevant parts of the pdf that David sent us. Thank you for that. My brain neurons are getting wired up, however, I need to clarify my understandings, therefore that would be great if you correct me if my following interpretations are incorrect or imprecise:
>
> Potential table coordinate with both .gro and .ndx files. Looks at index file for separating the potential tables and also looks at .gro for specifying the none_bonded interactions by finding the names of particles in .gro.
Not quite, really only the index group name matters, but one could
name the group FOO, then the table would be called table_FOO_FOO.xvg.
We just happen to name the index group that contains all CG atoms
"CG". In other tutorials you will see more table, e.g. with groups A
and B you will see table_A_A.xvg, table_B_B.xvg and table_A_B.xvg.

>
> The user tabulated table should have a specific name like table_CG_CG that which in this case determines the interaction between CG and CG in .gro file. If the system has more atoms than defined in CG list then the Gromacs looks for parameters from a default potential table called table.xvg which is in my case nothing as the whole system is CG. It is not clear in PDF where this table.xvg comes from if it is not defined by USER.

It might be that newer gromacs versions don't need table.xvg anymore,
but back in gromacs-5 times, gromacs could not figure out if all
combinations of energy groups were covered by the tables given. Just
imagine you have N index groups, then you need N*(N-1)/2 + N tables,
so table.xvg is the fallback table for uncovered groups and is usually
just a normal C6 /C12 LJ potential.


>
> The following two lines in mdp file
>
> Read user tables from table_CG_CG.xvg with 2501 data points.
>
> Tabscale = 1000 points/nm
>
> Read user tables from table.xvg with 1501 data points.
>
> Tabscale = 500 points/nm
>
>
> The first table is specific to the CG particles and the second one is for system particles, and because the whole system in my case is CG elements the table.xvg will remain without any use. Therefore, it does not matter what it contains as long as it is following the format of a Gromacs potential table.
Correct.

Christoph
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/CAAbE4Drv-RM0rV9mcabnLXS-Eh9L5LXNF5uLcwj_iDk%3D7Z9h0w%40mail.gmail.com.

Mohsen Farshad

unread,
Aug 2, 2021, 10:53:04 AM8/2/21
to vo...@googlegroups.com
Hi Christoph,

Thank you for your clrafications!

I ran the course-grained system of SPC/Model using the FM potential, and the attached RDF is generated which is compared with atomistic simulation. Do my plots look normal and represent the efficacy of FM potential correctly? Is there anything that we can do to match the FM with the All-atom simulation even better?

Thanks,
Mohsen

RDF_SPCE.pdf

Christoph Junghans

unread,
Aug 2, 2021, 2:48:27 PM8/2/21
to vo...@googlegroups.com
On Mon, Aug 2, 2021, 09:53 Mohsen Farshad <mohsen....@maine.edu> wrote:
Hi Christoph,

Thank you for your clrafications!

I ran the course-grained system of SPC/Model using the FM potential, and the attached RDF is generated which is compared with atomistic simulation. Do my plots look normal and represent the efficacy of FM potential correctly? Is there anything that we can do to match the FM with the All-atom simulation even better?
Have a look at the original votca paper, the result there is similar to what you saw.

In a nutshell, FM reproduces the force not the structure, if you need a cg model that reproduces the structure, you will need use a structure-based method like IBI or IMC.

Mohsen Farshad

unread,
Aug 2, 2021, 3:22:24 PM8/2/21
to vo...@googlegroups.com
Thank you, I had looked earlier at your dissertation and noticed that on page 88, also in the paper of Christoph Scherer.

I inferred that maybe FM potential inherently cannot mimic the distances between the water molecules.

Thanks,
Mohsen

Christoph Junghans

unread,
Aug 2, 2021, 9:22:56 PM8/2/21
to vo...@googlegroups.com


On Mon, Aug 2, 2021, 14:22 Mohsen Farshad <mohsen....@maine.edu> wrote:
Thank you, I had looked earlier at your dissertation and noticed that on page 88, also in the paper of Christoph Scherer.
I would also have a look at the FM (aka multi-scale coarse-graining) papers by the Voth group.

Christoph 
Reply all
Reply to author
Forward
0 new messages