use pyiron-internal is_skewed

[samwaseda]

Certain fixes in LAMMPS do not work with fix orthogonal etc., which used to work before (essentially before this commit). As I didn't really see how to check whether LAMMPS sees triclinic or orthogonal from pyiron, I decided to implement it inside pyiron. Hopefully this one is acceptable...

[jan-janssen]

But is this the same? For example in the case when we do a monte carlo simulation in combination with minimising each step. Then all cells in pyiron are orthogonal while the cells inside LAMMPS are not.

[jan-janssen]

Certain fixes in LAMMPS do not work with fix orthogonal etc.,

Can you provide some examples what does not work?

[samwaseda]

But is this the same? For example in the case when we do a monte carlo simulation in combination with minimising each step. Then all cells in pyiron are orthogonal while the cells inside LAMMPS are not.

That's not true. If the pyiron cell is orthogonal, then we use change_box all ortho, in which case it's not allowed to become triclinic. I don't see a problem with changing this, but it's unrelated to this PR.

[jan-janssen]

You are right, still I would be interested what fails. As I thought orthogonal boxes are the default for most cases.

[samwaseda]

You are right, still I would be interested what fails. As I thought orthogonal boxes are the default for most cases.

It's actually the callback function. If you are really really interested I can try to think of an easy example, but that's gonna probably still be somewhat cumbersome (both to create one and to read it)...

[samwaseda]

But actually it's maybe anyway time to think about making simple examples for callback functions, so I'm gonna make one anyway.

[pmrv]

This should work, but isn't it more straightforward to check the dot product between cell vectors? Your reasoning is documented though, so I don't mind too much.

[samwaseda]

Uh that's a good point. I don't have a strong opinion about it. The only one thing is that I want to insist on making tolerance a dimensionless value, in which case the computational cost is probably very much the same. If other people prefer dot product I can also change it.

[coveralls]

Pull Request Test Coverage Report for Build 536671936

• 13 of 13 (100.0%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.05%) to 64.762%

---

Totals
Change from base Build 533925843:	0.05%
Covered Lines:	10597
Relevant Lines:	16363

---

💛 - Coveralls

[samwaseda]

I found out that actually it was not the callback function that causes the problem, but apparently pressure=0 is not compatible with the old implementation. So the following lines failed:

job = pr.create_job('Lammps', 'md')
job.structure = pr.create_ase_bulk('Al', cubic=True)
job.calc_md(temperature=300, pressure=0)
job.interactive_open()
job.run()
job.run()
job.interactive_close()

And with this PR this problem was fixed.

With this, I guess it's clear to everyone that this PR is indispensable.

[samwaseda]

And since I promised @jan-janssen , here's a super simple notebook for callback:

callback.ipynb.txt

[jan-janssen]

Looks good to me