Questions

Dear. professor K. Saaskilahti

I was trying to studying the mean free path of phonon by lammps and had a trouble to do it.

In the mean time, I finally found your research.

Thankfully, you shared your work at https://github.com/ksaaskil/shc-python-tools

I have studied your papers in the github.

Could I ask something for your works?

[1] In your paper, Phys. Rev. B 91, 115426 (2015)

You used the nondimension quantity. but I would like to get the mean free path(MFP) of real unit (e.g. nm or Å )

Is it possible to get the MFP of real uint by your code(you upload)?

[2] Should I run your code only through the python ?

I'm not familiar with python and, in our system, lammps could be run without python?

but , in your description, the cell of amorphous Si looks be constructed by python code(silicon_example)

could I run your work first by only using lammps to get the output and then can I use python code by the output from lammps?

If possible, which output file sould i calcuated from Lammps?

[3] How can I change the target from to amorphous Si to the other material that I want to look for?

What line shoud i change to do it?

[4] getting the MFP from equation image.png

In this equation, q(w) is calculated from the code ? and M(w) is mode nuber at w then it just 3xmodes at the frequency?

Thanks for your kindness to help to study MFP

Could you please give some tips or reference or link to follow up your work ?

Best regards

---

Dear Sue,

thank you for your questions. It has been a long time since I have done research on this topic, so I'm not sure I can answer all your questions. I recommend you contact researchers who have more recently published works using this method.

I recommend you create a new issue instead of using wiki. Like that, both you and I will get notification from all responses.

[1] In your paper, Phys. Rev. B 91, 115426 (2015) You used the nondimension quantity. but I would like to get the mean free path(MFP) of real unit (e.g. nm or Å ) Is it possible to get the MFP of real uint by your code(you upload)?

To get the dimensional quantity, you need to calculate the spectral heat current q(w) using, for example, the code in this repository. Then you scale that with k_b\Delta T to get the dimensionless transmission function. Once you have that as a function of length, you can calculate the MFP using Eq. (16).

[2] Should I run your code only through the python ? I'm not familiar with python and, in our system, lammps could be run without python? but , in your description, the cell of amorphous Si looks be constructed by python code(silicon_example) could I run your work first by only using lammps to get the output and then can I use python code by the output from lammps? If possible, which output file sould i calcuated from Lammps?

It is possible to calculate the heat current decomposition without invoking LAMMPS from Python. The heat current calculation requires (1) interatomic force constants, and (2) atomic velocities from simulation. You can calculate the force constants using any method you like, in this package they're calculated by invoking LAMMPS from Python. Please see the example how to output atomic velocities from your simulation. There are examples in resources folder. Your simulation should output a file that looks like small.simu.vels.dat. The script compactify_vels.cpp can be used to create a compact version of that, with output looking like small.simu.vels.dat.compact.

[3] How can I change the target from to amorphous Si to the other material that I want to look for? What line shoud i change to do it?

You need to change many things. Please take a look at the example and try to understand how it works.

[4] getting the MFP from equation In this equation, q(w) is calculated from the code ? and M(w) is mode nuber at w then it just 3xmodes at the frequency?

I don't quite understand the question. But yes, heat current decomposition q(w) can be calculated using this code. M(w) can be calculated using, for example, Green's function method.

---

Dear. professor K. Saaskilahti

Thanks so much for your kind replies.

Your explanation would be good guides.

As you mentioned, I think that I need to examine your work more detail.

If I have additional questions after, is it okay to ask again in an issue section ?

Thanks, Sue

---

Dear Sue,

yes please, you can ask more questions in the Issues. Btw I'm not a Professor and you can just call me Kimmo :)

Best regards, Kimmo

---