How to use LJ potential to construct a relatively physical stable single-phase structure?

Hello! I want to use LJ potential to construct a block. And I want this block to be a more physical model. My problem is how to give a proper value for epsilon and sigma to make a stable block if this block’s lattice is FCC or any other form.

I know I need to do derivation of potential function to get force and obtain balance position. But I do not know the specfic method. And I did not find relevant materials about this. It will be quite appreciated if anyone can give me a clue about reference books about potential function. I am totally confused by it.

Any advice will be helpful! Thanks advanced

If you want it to be very very stable, then just increase the epsilon value. But you need to decide what is the ‘correct’ value. And the question is too vague, only you can answer what it means to be ‘more physical’. I don’t think anyone will provide a solid answer to your question.

Thank you for your reply!
A more physical model means this mode does not focus on specfic material. So I have to decide the epsilon and sigma by my own.
By the way, do you know any books or paper about potential function? I really need to supplement this knowledge.
Thank you!

There is exactly one (single component) LJ phase diagram, with whatever solid phases it contains. The only effect of the mass, sigma and epsilon parameters is to change the length, time, and energy/temperature scales of the simulation – which is why LAMMPS gives you the option to work in units lj and worry about physical scales later.

It sounds like you either do not have a supervisor or mentor, or have one who is not directing you to basic information sources and reliable literature reviews. You will struggle greatly and unnecessarily to make good progress from such a situation. No department would expect a scientist to learn electron microscopy or nanorheology purely from an Internet forum, and molecular dynamics is exactly the same.

Thank you for your advice. And I also clearly know my level, but I still need to study MD. Asking question in the Internet is one of the methods I use to attain knowledge.

I agree that I can not be an expert about MD simulation from just studying in the Internet, but I believe that it is definitely possible to get some basic knowledge. So it is very appreciated if you can give me some information like books’ names about potential function and MD,which is more helpful than just judging my level.

Thank you again.

There are many textbooks on MD simulations. Classics are the Frenkel and Smit “Understanding Molecular Simulation” which is more focused on the physics. Two advantages of this one is that it starts with other reading recommendations such as the Allen and Tildesley “Computer simulation of liquids” and other more statistical physics textbooks but also uses LJ models as prototypal example for phase diagrams, high order schemes etc. It is still a very dense book (~600 pages).

My personal go to would still be Hoover’s “Molecular Dynamics” and “Computational Physics Statistical Mechanics” (which you might find on his website) because I like his way of presenting with a physicist point of view, the edition and his figures. But both books might be a bit outdated on certain aspects.

Other personal suggestions would be:

  • Werner Krauth’s “Statistical Physics” which is very good, I think, for good undergrad students as it takes time to introduce the concepts in very clear ways and goes to advanced topics. But it is more physics focused and sticks to Monte-Carlo algorithms.
  • Shang-Keng Ma’s “Statistical Mechanics” which is essentially about the physics (numerical simulation is briefly mentioned) but makes links between thermodynamics and statistical physics on wide range of topics
  • Böttcher and Herrmann “Computational Statistical Physics” which is a compilation and rewriting of lecture notes from the authors and is clearly divided into stochastic methods and a dedicated section about molecular dynamics.
  • Hansen and McDonald “Theory of simple liquids” because it is a go to for simple fluid physics and it has been recently re-edited (2013).

I suggest you to see if your institution/university library gives access to these instead of buying one of your own. It is however important to keep in mind that text reading is only one aspect of the learning and that all these books have different objective and reader profile in mind. Some of them are more technical, some of them more focused one the numerical/physics part etc. This list is also far from exhaustive and might not cover some aspect and needs that are only (or mostly) discussed in the scientific literature, and that some aspects of simulation are only grasped by trial and errors or direct teaching.

But as the question textbook reading often happens to pop here and there with very few recommendations, I though I could give it a shot here.


What you can learn this way strongly depends on the quality of the questions you ask, and that is where you are falling short, time and again. In order to communicate well with people, you need to have a common ground of understanding and that - as you have demonstrated several times now - is missing and therefore you will get the advice to find yourself a proper tutor over and over again unless you change your ways. In fact, at some point, people will get tired of you not following the advice given and then will likely stop responding altogether.

In your situation, I would not be so eager to dismiss well intended advice from people that have significantly more experience than you. MD simulations are as much a craft as they are science and - like with most crafts - you can really learn its fundamentals only then well, if you are doing it with tutoring from some experienced craftsperson. Otherwise you are doomed to repeat many of the errors (often excellent) scientists have done in the 60+ years since the beginning of MD simulations on computers that have then been pointed out and discussed by the MD simulation community over time. Many of the details are not written down. They are more considered “tricks of the trade” and it is assumed that people practicing MD simulations know about them. Thus text books and publications are only taking you so far, but they rarely contain the practical knowledge of how to correctly set up and execute simulations. My preferred comparison is that to an owner’s manual of a car. It will teach you where the switches and pedals are and what functions they have, but it will not teach you how to drive a car.

If you have not found any of those on your own, you have to improve your internet skills significantly too. About any MD simulation software has a bunch of links to related text books. Also, recommendations have been given in this forum repeatedly.


To OP’s case I must say that I do not so often see textbooks recommendations but beside that, these are all very good points.

Having my own library I tend to forget that there is also a dedicated page on LAMMPS website.

They don’t need to be posted often since we have the entire history of the lammps mailing list archives included in this forum, so this is all searchable. Most of the recommended books are rather old, anyway. Plus one does not need to restrict the search to LAMMPS as MD code. The basics of MD simulations are the same for any MD code.

1 Like

Please let me conclude this thread with a few more remarks.

You need to understand that when people here are telling you their opinion this is not because we enjoy to crush people’s confidence and because we want to make your life miserable. Quite the contrary, we want to keep you from making your life miserable by wasting time doing things that are going to be unproductive. The biggest problem in that regard is often “unearned confidence”, i.e. people believe that they can do a lot of things in a specific way because they had been successful in the past with the same approach. In most cases we have already made the mistakes that you are about to make and we want to stop you from going through the same pains that we went through or saw colleagues go through. There is a saying in Germany that translates to “The jug goes to the well until it breaks”, meaning you can be successful doing the same thing until you are not.

Going back to your original request, which is a repeat of a previous post. Basically, the issue here is that you are declaring something as “physical” behavior that isn’t and now you want to “fix” it through a method that cannot work as has been very clearly explained by @srtee. There is not going to be a book or paper that will explain you what you want to have explained, because your premise is already misguided. …and here we get back to the problem of not having a proper tutor. You would not be in this situation, if you had discussed this with a reasonably competent person.

While being persistent and strong willed can be useful in research, it is not such a smart choice when you are repeatedly ignoring the advice of more experienced (and well meaning!!) people.


Much thanks to you akohlmey and Germain!

I will value your advices and I will talk to my tutor. I hope when I ask question here next time, my problem will be meaningful.

I knew I have asked some stupid questions. While these questions actually confused me at that time. Your reply is detailed and warm-hearted, by the way, such reply is rarely seen in Chinese forum. I have learnt a lot here.

Given to my poor English, it is still time-consuming for me to read related materials, which may be one of the cause of those stupid questions.

Much thanks again.