回复： thermal conductivity with Muller-Plathe

[Carlos_Campana]

Please, do not reply to me but to the list as I am not your private tutor. Nobody can provide you in a short email with all the know-how that you need to develop the expertise required to successfully run simulations to compute transport properties. As always, when not having direct contact with an expert in the field (I am not one of them for sure), your best allies are the original publications where the methods were reported for the first time plus review papers that may cover in detail the potential hurdles/pitfalls to be encountered. A bit of common sense applies as well. Instead of jumping right away to perform the calculation on some complex system of your interest, build some toy-model problem with known solution (like the Ar gas case covered within the Lammps examples) to lower the number of potential mistakes you could make along the way. When available, read the papers listed on the manual page corresponding to the specific Lammps feature you are trying to use. Additionally, checking the Lammps archives is also recommended as you are not the first having requested this kind of assistance. I have seen quite a few users asking about transport properties in the past few months. Imagine what you could find if you do some digging into the archives…

And finally, always post your messages to the list and not an individual user. This is a good “sampling technique” to gather most of the available free advice people here are willing to provide.

Carlos

[LC_Liu]

Carlos made a very helpful comment. I think we can all benefit from that.

As for the non-linearity, there is another picture being more direct. The temperature range may be too large so that thermal conductivity itself is no longer a constant but dependent on temperature. Hence the temperature profile does not hold linear since its slope reflects the current thermal conductivity.

However, this is rarely the case. More possibly it is due to improper numerical settings, such as, too frequent energy exchange, too short simulation domain, too many bins (i.e., too little atoms number to represent )…it is really hard to tell the root cause for your profile with so little information you provided.

So, the best way to learn is to find a literature with detailed simulation parameters to follow. Do what exactly they do. Observe the results. Then, consult experts if you see any ambiguity. This process is indispensable and truly helpful since you can educate yourself by walking through.

Finally, not to you personally, but to all the non-English speaking users. Please do address your name on the bottom of the email, OR, change your name to something readable. Otherwise people do not know how address you in the email.

LC Liu