Dear Sir/Madame,
I wanted to run MD simulation of chromatin filament using LAMMPS. However the simulation ended by the Segmentation fault. Could you help me to recognise what is wrong with options I used or my system. I checked the molecule in the VMD and it looks ok, I did not found any extremely long bond. Below is the input file and the output of LAMMPS:
Input:
irina,
this is something that is difficult to debug and assist you without being able to look over your shoulder.
there are a few warnings that you should look into whether they indicate a serious problem or not.
they are placed in the code for a reason and when you are not certain, that they are harmless, you need to investigate.
also, i would recommend, to separate the “editing” and “production” part of your simulation.
so write out a data file with write_data after you have done all system manipulations (like deleting atoms/bonds).
and then you should also consider using the “reset_ids” command (and again write out a data file).
then you can check out those data files (and visualize them) for correctness and then you can attempt to start your simulation in a new input with the final data file.
Axel.
Dear Axel,
Thank you for fast reply and your suggestions, but when I hashed deleting lines, I still have the same error. If you have any other suggestions, I would would be very grateful.
Best regards,
Irina.
as i have already mentioned, this kind of issue is difficult and time consuming to debug from remote.
the warnings you get hint at some possible issues with your data file. however, what exactly it can be, there are just too many options to list them all (and then there some i just can’t think of at the moment).
the general advice for issues like these is to reduce and simplify the simulation and then use tools like valgrind, or compiler instrumentation to identify possible locations of memory issues.
it is also a good idea to compile and test with the latest patch release in case you are running into a bug that has been fixed since.
but the same flexibility that makes LAMMPS an attractive tool for simulations also makes it difficult to protect people from doing things that can lead to crashes. the saying “garbage in, garbage out” applies to LAMMPS more than for other software, because what may be considered bad under some circumstances would be a good idea under others.
for the same reason, we are still cashing bugs regularly (and not only in newly added code), but after several years of systematic use of code analysis tools, automated integration and regression testing and (as of recent) some initial unit testing (although a large part should technically be considered a simplified form of regression testing), we are getting increasingly confident that bugs are mostly in new features and rarely used features for which we - as of yet - don’t have sufficient tests.
axel.
Thank you very much for your answer. I generated a file with the coordinates of my system based on the Monte Carlo simulation pdb file. I used VMD to convert my system to a LAMMPS file. But I think VMD does not produce relevant data. Can you advise any software to generate a LAMMPS file and then visualize and analyze the LAMMPS trajectory for novice LAMMPS users?
Could you also advise where I can find the description of the LAMMPS system file? I am interested what three columns after the coordinates in the lines of atoms describe?
Best regards,
Irina.
Please note that as a novice you will need in person tutoring and supervision in matters of MD simulation best practices and workflows.
This cannot be done effectively via e-mail and is mostly beyond the scope of this mailing list.
Thank you very much for your answer. I generated a file with the coordinates of my system based on the Monte Carlo simulation pdb file. I used VMD to convert my system to a LAMMPS file. But I think VMD does not produce relevant data.
it cannot produce it, because the PDB file does not contain it. creating a proper data file with proper atom typing and bond topology is not something that can be done fully automatically with a file with limited information like a PDB file. Those can only be used if they conform to certain standards and conventions (e.g. for proteins) and then you need a specific tool to process it. And even then, this will almost always not work transparently, but will require some manual interventions, some choices to make and some understanding of the force field in question.
Can you advise any software to generate a LAMMPS file and then visualize and analyze the LAMMPS trajectory for novice LAMMPS users?
there is no software that automatically will do everything correctly and work for all circumstances. this is part of the learning process of learning to do MD.
a list of tools that work with LAMMPS is here https://lammps.sandia.gov/prepost.html
Could you also advise where I can find the description of the LAMMPS system file? I am interested what three columns after the coordinates in the lines of atoms describe?
image flags.
please see https://lammps.sandia.gov/doc/read_data.html
axel.