I am trying to use hybrid potential for PCFF and extep potential. When I run the simulation with hybrid potential, I get an error “Bond atom missing on proc 0 at step 2”. While, when I run the same simulation with only PCFF its runs fine. I tried to figure out why this is happening but I failed. Please help me with this.
2.in (4.99 KB)
3.in (1.36 KB)
log.lammps (8.17 KB)
How did you find what potential (pair style) to use between the two subsystems and where did you get the parameters (pair coeffs)?
Your log file shows an extremely high potential energy in the first step, which is not a good sign.
The class 2 potential used are for polymer and the extep are used for BNNT. I am trying to make nanocomposite. For that, I make the model in the material studio and then convert it into lammps data file. I did not minimize the model in the material studio and may be the reason for extremely high potential energy and that is why I want to minimize it in lammps.
But when I run the simulation with PCFF forcefield only, the simulation runs fine. I do this for checking whether the simulation runs with PCFF only or not. For running with PCFF, I assigned arbitrary values to BNNT in the data file. But, for correct simulation, I need to use hybrid potential where I need to use PCFF for polymer and extep for BNNT.
The sample of my data file is:
LAMMPS data file. msi2lmp v3.9.9 / 05 Nov 2018 / CGCMM for 5_5BNNC60.1
9030 atoms
7650 bonds
9450 angles
11700 dihedrals
2700 impropers
12 atom types
6 bond types
8 angle types
11 dihedral types
5 improper types
-0.273660731 57.736339269 xlo xhi
0.253431909 58.263431909 ylo yhi
0.286164898 60.395264898 zlo zhi
Masses
1 12.011150 # cp
2 15.999400 # oh
3 1.007970 # hc
4 1.007970 # ho
5 12.011150 # c
6 12.011150 # cp
7 12.011150 # cp
8 12.011150 # c2
9 1.007970 # hc
10 12.011150 # c2
11 14.0067 # N_R
12 10.811 # B_2
Pair Coeffs # lj/class2/coul/long
1 lj/class2/coul/long 0.0640000000 4.0100000000 # cp
2 lj/class2/coul/long 0.2400000000 3.5350000000 # oh
3 lj/class2/coul/long 0.0200000000 2.9950000000 # hc
4 lj/class2/coul/long 0.0130000000 1.0980000000 # ho
5 lj/class2/coul/long 0.0540000000 4.0100000000 # c
6 lj/class2/coul/long 0.0640000000 4.0100000000 # cp
7 lj/class2/coul/long 0.0640000000 4.0100000000 # cp
8 lj/class2/coul/long 0.0540000000 4.0100000000 # c2
9 lj/class2/coul/long 0.0200000000 2.9950000000 # hc
10 lj/class2/coul/long 0.0540000000 4.0100000000 # c2
11 none
12 none
Bond Coeffs # class2
1 1.4170 470.8361 -627.6179 1327.6345 # cp-cp
2 1.3768 428.8798 -738.2351 1114.9655 # cp-oh
3 1.0982 372.8251 -803.4526 894.3173 # cp-hc
4 0.9650 532.5062 -1282.9050 2004.7658 # oh-ho
…
…
Atoms # full
1 1 1 0.0265 4.308382382 12.73913103 55.11112851 0 0 0 # cp
2 1 1 -0.1268 5.030299881 11.73966792 55.77416932 0 0 0 # cp
3 1 1 -0.1268 5.586002707 10.67740006 55.05542306 0 0 0 # cp
4 1 1 -0.1268 5.424326954 10.60839976 53.66849533 0 0 0 # cp
5 1 1 -0.1268 4.708384298 11.60531328 52.99841799 0 0 0 # cp
6 1 1 -0.1268 4.14887167 12.6681674 53.71575177 0 0 0 # cp
7 1 2 -0.4506 3.756245449 13.78554757 55.85295435 0 0 0 # oh
8 1 3 0.1268 5.152139275 11.78142826 56.84844237 0 0 0 # hc
9 1 3 0.1268 6.13515941 9.905081416 55.5753041 0 0 0 # hc
10 1 3 0.1268 5.850320313 9.781481698 53.11443217 0 0 0 # hc
11 1 3 0.1268 4.580376001 11.55086688 51.92490306 0 0 0 # hc
12 1 3 0.4241 3.233160678 14.44632626 55.28978859 0 0 0 # ho
13 1 3 0.1268 3.592989456 13.43199434 53.18686416 0 0 0 # hc
14 2 1 0.0265 13.50156981 34.38492757 54.32908914 0 0 0 # cp
…
…
Bonds
1 1 1 2
2 1 1 6
3 2 1 7
4 1 2 3
5 3 2 8
6 1 3 4
7 3 3 9
8 1 4 5
9 3 4 10
10 1 5 6
11 3 5 11
12 3 6 13
13 4 7 12
14 1 14 15
15 1 14 19
16 2 14 20
17 1 15 16
18 3 15 21
19 1 16 17
20 3 16 22
21 1 17 18
22 3 17 23
23 1 18 19
24 3 18 24
25 3 19 26
26 4 20 25
27 1 27 28
28 1 27 32
29 2 27 33
30 1 28 29
31 3 28 34
32 1 29 30
33 3 29 35
34 1 30 31
35 3 30 36
36 1 31 32
37 3 31 37
38 3 32 39
39 4 33 38
40 1 40 41
41 1 40 45
42 2 40 46
43 1 41 42
44 3 41 47
45 1 42 43
46 3 42 48
47 1 43 44
48 3 43 49
49 1 44 45
50 3 44 50
51 3 45 52
52 4 46 51
53 1 53 54
54 1 53 58
…
…
Angles
1 1 2 1 6
2 2 2 1 7
3 2 6 1 7
4 1 1 2 3
5 3 1 2 8
6 3 3 2 8
7 1 2 3 4
8 3 2 3 9
9 3 4 3 9
10 1 3 4 5
11 3 3 4 10
12 3 5 4 10
13 1 4 5 6
14 3 4 5 11
15 3 6 5 11
16 1 1 6 5
17 3 1 6 13
18 3 5 6 13
19 4 1 7 12
…
…
You are not answering my question and I suspect that the correct answer to that is the solution to your problem.
Using a hybrid potential with correct syntax is one thing, but having the proper model and parameters that describe the interactions between the two different subsystems are crucial for a correct and meaningful simulation. You currently have lj/cut with parameters 15.0 15.0 which for all which makes no sense at all (for both the pair style and the magnitude of the parameters) considering that you are interfacing a molecular system with point charges with a polar compound.
There are other questionable choices in your input, but those are minor issues compared to this and all the points you are making are irrelevant.
Now, I understand your question.
Yes, its correct that I am not sure what pair style to use for the interaction of the two sub systems and because of this reason I used lj/cut for this. But, now you have pointed out that this is the problem of my system and for correct simulation, I need to have correct pair style and parameters for the interaction of two sub systems. I will try to figure it out and let you know, if I fail to do so.
Thank you for your valuable time and feedback.
Sir, Is it correct to use the potential for the interaction of the two subsystems from the table (attached as fig). This table is taken from the reference " (Mayo) Mayo, Olfason, Goddard III, J Phys Chem, 94, 8897-8909 (1990)".
Sir,
And, also when I run the simulation in lammps serial it’s running fine while when I run the same simulation in lammps mpi it is giving en error “Incorrect args for pair coefficients”. Why this is happening?
2.in (5.52 KB)
log.lammps (11.4 KB)
log.lammps (2.24 KB)
Let me respond to this with a question for you: would it be correct to use pair style lj/class2 instead of lj/class2/coul/long to simulate your polymer?
this is a bug in the “serial” executable and will be fixed in the next LAMMPS release. your input is incorrect and not supposed to run.
the fact that a simulation continues to run does not say anything about it being a meaningful simulation creating meaningful results.
axel.
Sir,
As far as I understand, I think it’s correct to use pair style lj/class 2 instead of lj/class2/coul/long. The only difference between the two is lj/class2 does not consider coulombic interaction in the simulation while lj/class2/coul/long will consider coulombic forces for long interaction and hence, there is a small change in the result.
So, shall I consider the parameters taken for the interaction of the two subsystems from the “Mayo” reference not precisely correct but approximately correct?
Sir,
As far as I understand, I think it’s correct to use pair style lj/class 2 instead of lj/class2/coul/long. The only difference between the two is lj/class2 does not consider coulombic interaction in the simulation while lj/class2/coul/long will consider coulombic forces for long interaction and hence, there is a small change in the result.
that is where you are very wrong. the coulomb contributions are a crucial part of the model. your response confirms my suspicion that it would be a waste of my time to try and explain the more subtle problems with your model and how you plan to set up the interactions.
this ultimately is not a concern of the LAMMPS software and this mailing list, since this is about doing the proper science. the LAMMPS code does not care, it just executes what you ask it to do. it will happily compute utter nonsense for as long as it does not lead to algorithmic instabilities. the responsibility to feed an MD simulation program - and that applies to all of them and any other simulation software for other methods, too - is solely on the shoulders of the person using it. however, the primary responsibility of teaching you how to do correct simulation is not the mailing list, but your adviser/tutor and other colleagues concerned with the validity of your research.
axel.
Sir,
I understand your time is precious. But, my concern is how to implement hybrid potential, where I can implement PCFF potential for my polymer matrix, extep for Boron nitride and Lj for interaction. I also found a paper after getting your mail where the authors implement PCFF for polymer matrix, tersoff for Boron nitride and LJ for interaction.
If I am not able to implement the proper science then please help me because I don’t have any advisor/tutor who is knowledgeable as you are.
Sir, where you are, you might get interacted with only brilliant students or people but there are other countries, other places in the world where the students are poor or average and they are lacking technology. I am not asking you to teach me but at least you can guide me with reference paper or books from where I can learn the solution of my problem and also I can acquire the proper science of molecular simulation.
Sir, it’s not like that, that I posted my question directly on lammps mailing list. Before this, I tried very hard. I also searched on the internet, read so many journals but I failed to get any idea that how to solve my problem. Even, I am trying yet also.
I apologize if my words hurt you in any manner.
Reference Paper/journal- https://www.sciencedirect.com/science/article/abs/pii/S0017931020330416
“Atomic-scale insights into interface thermal resistance between epoxy and boron nitride in nanocomposites”
Sir,
I understand your time is precious. But, my concern is how to implement hybrid potential, where I can implement PCFF potential for my polymer matrix, extep for Boron nitride and Lj for interaction. I also found a paper after getting your mail where the authors implement PCFF for polymer matrix, tersoff for Boron nitride and LJ for interaction.
…and if I had reviewed that paper, I would have suggested rejecting it because of the ill-conceived model. The authors even explicitly admit to it by stating that they disregard and charges on the BN compound without providing proof that their approximation is valid. I do understand why they are doing this, because there is no clean way to combine a tersoff or similar potential for a polar material with a conventional force field, but that doesn’t make it a valid model.
I would also be concerned about the validity of the tersoff potential parameters for a nano particle as its parameterization strategy and potential function does not account for changes in polarization at the surfaces, edges and corners. There is the classical - and more extreme - example of a tersoff potential for SiO2, which will simply not produce stable systems outside of bulk setups. Simulating such systems usually requires some kind of method to handle the change in polarization due to the geometry like it is done with charge equilibration methods. There are manybody force fields that include those.
If I am not able to implement the proper science then please help me because I don’t have any advisor/tutor who is knowledgeable as you are.
That is not a valid argument. First off, I am not in a position where I can accept students to advise and I am not interested to do it but most certainly I have not agreed to do it.
That said, there are people far more knowledgeable in this area of research than me, so by your reasoning you should ask those, not me. Your lack of understanding of basic MD force field physic is quite significant and that you should be able to learn from anybody with some experience in the field. Otherwise that person would not be competent to supervise a project like yours.
Sir, where you are, you might get interacted with only brilliant students or people but there are other countries, other places in the world where the students are poor or average and they are lacking technology. I am not asking you to teach me but at least you can guide me with reference paper or books from where I can learn the solution of my problem and also I can acquire the proper science of molecular simulation.
Sorry, what you are asking is contradicting itself. If I do what you ask me to do, I will act as your adviser with having all the effort but without any of the benefits.
Sir, it’s not like that, that I posted my question directly on lammps mailing list. Before this, I tried very hard. I also searched on the internet, read so many journals but I failed to get any idea that how to solve my problem. Even, I am trying yet also.
How much effort you have invested is irrelevant. The problem is that you are taking on a problem that is far beyond your level of understanding. You need to start with something less demanding and build your skills. This is something that your adviser should be aware of and guide you appropriately. If your adviser does not possess the experience or skills to oversee every detail of your project, then he or she would have to search for a suitably competent collaborator on that project. That is how this is done in science. Unfortunately, it has become quite common that advisers let their students do projects that are beyond their training level and are relying on online forums and other means to make up for their own shortcomings. But similarly, it has become common for students to plead with people in online forums for guiding their research rather than demanding the proper advising and training for their project from the person responsible, i.e. their adviser. If your adviser is not willing to address this issue you should ask for a different project or look for a different adviser. It is most certainly not the responsibility for people like me to resolve this. Like other researchers, I have the liberty to choose what I am working on within the confines of the position I am hired on.
I apologize if my words hurt you in any manner.
I am not offended, but just annoyed. This is a discussion that should not have happened if you had proper advising and training and then would have approached your project in a proper way.
Sir, I already told I don’t have any advisor/supervisor and its not part of any project.
Thank you for your precious time and effort.
Any one else who can help me with this?
No. You did not say that you had no adviser. You had said that your adviser was not as knowledgeable as me, which is not the same. So why are you doing this at all? Do you like to be told off?
Doing this kind of research without any guidance and tutoring is even more misguided and ill-conceived that doing it with an adviser that is lacking the required experience or not putting in the required effort.
Searching for the kind and amount of tutoring that you need on a mailing list like this one is a bad idea.
Thank you axel for your time. I am able to figure out the solution and its works well for me.
