| 6 Aug 2020 |
 Eric Hermes | https://gitlab.com/ase/ase/-/merge_requests/1988 | 20:57:09 |
| Eric Hermes | I am writing a test to prove that this MR fixes the problem. | 20:57:20 |
| Eric Hermes | and also in the process refactoring the existing D3 tests | 20:57:34 |
| Eric Hermes | so they are more pytest-y | 20:57:40 |
 askhl | Significantly funnifying the cell of a physical system is likely to cause trouble with many calculators (convergence etc.) | 20:59:36 |
| Eric Hermes | it's not that significant, is it? | 20:59:55 |
 alin m elena | so if just non-cubic | 21:00:07 |
| askhl | but insignificantly doing it might mean that an error drowns in numerics | 21:00:08 |
| alin m elena | there are some simple systems in gulp at least you can check against somethign else | 21:00:24 |
 chronum | Shouldn't 0.1 to 0.01 bet the sweet spot for that then? | 21:00:35 |
| askhl | It should be a single-atom cell though. IMHO | 21:00:37 |
| askhl | if not HCP | 21:00:43 |
| Eric Hermes | I'd prefer something main-group | 21:01:17 |
| alin m elena | askhl: 2 atoms | 21:01:43 |
| askhl | Can it be graphite? | 21:01:45 |
| Eric Hermes | I thought graphene had a 2 atom cell. and graphite definitely has at least 2 atoms | 21:02:27 |
| alin m elena | cell
4.7602 4.7602 12.9933 90.000000 90.000000 120.0
frac
Al core 0.000000 0.000000 0.352160
O core 0.306240 0.000000 0.250000
space
167
| 21:02:31 |
| alin m elena | this will end up with more due to spacegroup | 21:02:55 |
| Eric Hermes | what about bulk('Al') | 21:03:27 |
| askhl | Al is fcc | 21:03:35 |
| Eric Hermes | it's not HCP (I'm not sure why you want HCP) | 21:03:36 |
| askhl | HCP has an asymmetric cell | 21:03:51 |
| askhl | But we can also do a perturbed FCC | 21:04:02 |
| askhl | I guess Al will be fine | 21:04:06 |
| askhl | +1 to perturbed fcc Al | 21:04:20 |
| Eric Hermes | ok. I'm just doing this in test_dftd3.py for now anyway | 21:04:34 |
| Eric Hermes | but if we find the structure sufficiently useful we can move it somewhere more convenient for other tests. | 21:04:47 |
| askhl | Yes | 21:04:55 |
| askhl | Eric Hermes: do you think the pytest calculator listing in !1975 is all right? | 21:08:03 |
| Eric Hermes | I haven't had an opportunity to look at it in detail | 21:08:31 |
