Symmetry recognization of primitive cells in CELL_OPT

[Lushan Wang]

Symmetry was not retained during cell-varied geometry optimization of an iodine single crystal (Space Group #64 - Cmce, 8 atoms in unit cell) with primitive cell (4 atoms). No matter we set "KEEP_SPACE_GROUP" and "KEEP_SYMMETRY" to be T or F, with or without set "SYMMETRY" to be "ORTHORHOMBIC" or even "MONOCLINIC_GAMMA_AB Monoclinic (a = b ≠ c, α = β = 90°, γ ≠ 90°)" (as it is in https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/SUBSYS/CELL/CELL_REF.html) in $CELL, the symmetry was not maintained and became monoclinic or triclinic where it can't recognize the structure as a primitive cell of C-centered orthorhombic lattice any more.

My question: In cp2k, must we not use primitive cell for cell-varied structural optimization or am I too naive to set correct parameters to retain symmetry?

I've searched Google and conversations in cp2k group and didn't find relevant contents. Any help would be appreciated. Thanks!

PS: Original Cell information:
      &CELL
            A [angstrom]     3.7632537052    -2.2808521645     0.0000000000
            B [angstrom]     3.7632537052     2.2808521645     0.0000000000
            C [angstrom]     0.0000000000     0.0000000000     9.7061460000
         PERIODIC XYZ
 SYMMETRY MONOCLINIC_GAMMA_AB
      &END CELL
      &COORD
            I     0.0000000000     0.7474991182     1.1868772390
            I     0.0000000000     3.8142052109     8.5192687610
            I    -0.0000000000     1.5333530464     6.0399502390
            I     0.0000000000     3.0283512827     3.6661957610
      &END COORD

Methods:
Functional: PBE + DFT-D3(BJ)
Basis sets: DZVP-MOLOPT-SR-GTH
K-points: 8,8,3
CUTOFF and rel-CUTOFF: 400,50
EPS_SCF: 1E-6
EPS_DEFAULT: 1E-12

[Jürg Hutter]

Hi
I don't have experience with such calculations, but the relevant tests are in
.../tests/Fist/regtest-spgr/
.../tests/QS/regtest-spgr/
Maybe you can see where your example goes beyond what has been tested.
regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Lushan Wang <wanglu...@gmail.com>
Sent: Tuesday, April 23, 2024 7:53 PM
To: cp2k
Subject: [CP2K:20126] Symmetry recognization of primitive cells in CELL_OPT

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/12b9cf36-0665-4e90-b69f-67139d7304aen%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/12b9cf36-0665-4e90-b69f-67139d7304aen%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Lushan Wang]

Dear Prof. Hutter,

Many thanks for your rapid and instructive reply!

The symmetry and space group of all the test cases ("zro" and "quartz") in the two folders you've mentioned (as in version 2023.1) were kept perfectly.

Thus, I used quartz.inp input file for my iodine primitive test case only changing the $CELL and $COORD to iodine.

Here's what I've found. In the situation when "KEEP_SYMMETRY" and "KEEP_SPACE_GROUP" were both set T, the space group of inp file was correctly recognized as #64,Cmce but not maintained after the cell-varied optimization. (Situation 3 in the table, input file attached.)

No.  SYMMETRY             KEEP_ANGLE  inp_file_spgr_recognized  restart_file_spgr*
1    MONOCLINIC_GAMMA_AB  T           2,P-1                     2,P-1
2    NONE                 T           64,Cmce                   64,Cmce
3    NONE                 F           64,Cmce                   13,P2/c

( * restart_file_spgr was tested by vaspkit)

PS:
I went through the related source code to find that my question might relate to the space group identification and maintainance algorithm of the primitive cell in cp2k. I learnt that in "space_groups.F", the "identify_space_group" function (which CG or BFGS calls) links to spglib, while the "spgr_apply_rotations_coord" function is designed for atomic positions asjustment according to specified symmetry operations. But I'm not a seasoned programmer and if I have time I'll go through the details in the future. Again, many thanks for your instructions!

[Jürg Hutter]

Hi

maybe you want to open an issue on
https://github.com/cp2k/cp2k/issues

This way the problem is documented and in view of the programmers
and there is a chance that the author of this part of the code will pick it up.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of Lushan Wang <wanglu...@gmail.com>

Sent: Friday, April 26, 2024 8:30 AM
To: cp2k
Subject: Re: [CP2K:20132] Symmetry recognization of primitive cells in CELL_OPT

A [angstrom] 3.7632537052<tel:(763)%20253-7052> -2.2808521645 0.0000000000
B [angstrom] 3.7632537052<tel:(763)%20253-7052> 2.2808521645 0.0000000000

C [angstrom] 0.0000000000 0.0000000000 9.7061460000
PERIODIC XYZ
SYMMETRY MONOCLINIC_GAMMA_AB
&END CELL
&COORD

I 0.0000000000 0.7474991182<tel:(747)%20499-1182> 1.1868772390
I 0.0000000000 3.8142052109<tel:(814)%20205-2109> 8.5192687610<tel:(519)%20268-7610>
I -0.0000000000 1.5333530464<tel:(533)%20353-0464> 6.0399502390

I 0.0000000000 3.0283512827 3.6661957610
&END COORD

Methods:
Functional: PBE + DFT-D3(BJ)
Basis sets: DZVP-MOLOPT-SR-GTH
K-points: 8,8,3
CUTOFF and rel-CUTOFF: 400,50
EPS_SCF: 1E-6
EPS_DEFAULT: 1E-12

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/12b9cf36-0665-4e90-b69f-67139d7304aen%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/12b9cf36-0665-4e90-b69f-67139d7304aen%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/8c482d84-6dbc-451f-b7f5-d650363f48a4n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/8c482d84-6dbc-451f-b7f5-d650363f48a4n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Lushan Wang]

Dear Prof. Hutter,

Thank you for your suggestions. A new issue has been raised on github: https://github.com/cp2k/cp2k/issues/3384