[Wien] 'x symmetry' issue

[Stefaan Cottenier]

Dear Laurence and Gavin,

>Two patches may work:
>1) use "init_lapw ....-s sgroup" to skip symmetry

That does not help, because symmetry produces a.o. The LM combinations in case.in2 (which I will not try to construct manually for 102 blocks...)

>2) generate the Symm using a smaller cell than paste those onto the end of the struct file. Maybe also use x patchsymm as well. You may still need to do 1!

x patchsymm : I had forgotten about this one. Tried it now, but it reports 'no position errors found'

The detour via a smaller cell is complicated, because this is a supercell where one impurity breaks a lot of symmetry, and thare are many inequivalent positions with different point groups. It would be cumbersome to try creating all of them via different cells.

> Seems like it could be related to numerical precision.  Taking the test2.struct that you sent, I also see the "2 missing" with WIEN2k 24.1 for Y=0.66666666 for ATOM 74 and 98, but it seems to go away if Y=0.66666667 is used:

I actually had tried that before posting, with ATOM 18, the one that caused the problem. And it still gave the error. But by your comment, I realize that I did not verify whether that error came from the same atom. It may have been from one deeper down the list. If I now replace all 0.666... by a '7' at the end (in about 10 positions), then indeed symmetry passes nicely.

Conclusion: it was indeed a numerical precision issue, one that was not catched by patchsymm, and that appeared in multiple inequivalent positions. One need to correct all of them before 'x symmetry' works.

OK, solved - thanks for your inspiration 🙂.

Stefaan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20260529/0a5e04b4/attachment.htm>