Timeline for DFT code for simulating thousands of atoms in a supercell
Current License: CC BY-SA 4.0
16 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 6, 2022 at 0:19 | comment | added | Phil Hasnip | We really need to know how many valence electrons and what kind of simulation you'd like to do. A simple SCF ground state + geometry optimisation is not very challenging even for a conventional plane-wave DFT package until you get well over 10,000 electrons, and even then it's probably memory which hurts you rather than computational time. As a specific example, ~6 years ago I did a 4096-atom (16384 electron) Si supercell; it took CASTEP 30 mins for an SCF ground state, using 576 Ivy Bridge cores. (Si is especially DFT-friendly, of course, so maybe multiply by ~3-4 for tougher things.) | |
| May 17, 2022 at 7:41 | vote | accept | Jack | ||
| May 14, 2022 at 0:57 | answer | added | Karim Elgammal | timeline score: 3 | |
| May 13, 2022 at 9:16 | comment | added | uhoh | in Nature Computational Materials Yu et al. (2019)'s Dodecagonal bilayer graphene quasicrystal and its approximants they validate their approximants using ten million atoms and the "tight-binding propagation method (TBPM)". | |
| May 13, 2022 at 7:09 | answer | added | Ian Bush | timeline score: 9 | |
| May 13, 2022 at 7:01 | comment | added | DLV | FHI-aims is also O(N) and can do this. | |
| May 13, 2022 at 6:44 | comment | added | Ian Bush | @Jack - could you answer the other questions Gregor asks, they are important, and add the details to the question, not as a comment - what you want to calculate can affect you choice (e.g. MD is much harder than a single point energy), as will exactly what kind of chemical system you are studying. | |
| May 13, 2022 at 3:18 | history | became hot network question | |||
| May 13, 2022 at 1:36 | answer | added | Sha | timeline score: 13 | |
| May 12, 2022 at 19:52 | answer | added | Elie H | timeline score: 11 | |
| May 12, 2022 at 19:40 | comment | added | Jack | @Greg, I haven't tried anything except VASP. Currently, I have access to a few hundred nodes in the HPC center. | |
| May 12, 2022 at 19:39 | comment | added | Gregor Michalicek | 1. How much computational resources do you have available? 2. What quantity are you interested in? 3. Is there something special about the system besides the large supercell? 4. How many thousands? 2.000 is very different from 8.000. Depending on the code that could nearly imply a factor 100 in the needed computing resources. | |
| May 12, 2022 at 19:38 | comment | added | Jack | @ElieH, Yes, Thank you very much for your contribution. | |
| May 12, 2022 at 19:37 | comment | added | Elie H | SIESTA !!! Should I make a complete answer to explain its advantages ? | |
| May 12, 2022 at 19:28 | comment | added | Greg | Have you tried cp2k or QE? Do you actually have the computational capacity for such a calculation? | |
| May 12, 2022 at 19:16 | history | asked | Jack | CC BY-SA 4.0 |