crystod-mag#
Symmetry-adapted spin bases (cluster multipoles / SAMM): every symmetry-distinct magnetic order allowed on the sites of one element, printed with its irrep, its per-atom spin directions, a ready-to-paste VASP or Quantum ESPRESSO magnetization input, and a VESTA file with spin arrows.
I want to … |
command |
|---|---|
enumerate the magnetic orders at q = 0 |
|
survey every special k point |
|
get a Quantum ESPRESSO input instead |
|
build a magnetic supercell at q ≠ 0 |
|
get the magnetic space group, the responses allowed by symmetry and the altermagnet / AFM class of every basis vector |
|
list the operations of the paramagnetic group each arrangement keeps and breaks |
|
know which magnetic space groups the magnetic orders of a k point give, without a structure file |
|
get every order-parameter direction of a magnetic order at q ≠ 0 (both phases of a complex order parameter) with its magnetic isotropy subgroup and a VESTA file |
|
place those orders on a distorted (low-symmetry) structure and find, for each direction, the domain that is one irrep of that structure, with its magnetic space group and a MAGMOM line in its cell |
|
know which textures of the phonon angular momentum near Γ (s-, p-, d-wave, …) and which nodal planes every arrangement, and the paramagnetic crystal, allow |
|
enumerate the magnetic orders of spins on two or more elements at once, with the copies of one irrep combined across the elements, or superpose listed orders of any irreps |
|
get the actual moments on my crystal for every magnetic isotropy subgroup of one irrep (or of several coupled irreps), each with its group checked on the structure, MAGMOM and VESTA |
|
05-01. Magnetic arrangements of given irreps (--irrep with -c)#
Example directory: example/05-01_mag_irrep (testsuite section 05-01); after pip install: crystod-example 05-01
crystod-mag --parent SG --irrep IR tells which magnetic space group every
order-parameter direction of an irrep gives, without a structure. With a
structure and an element instead of --parent, every row of that table (with
several irreps, every row of the coupled table) is placed on the sites of the
element: the moments of the arrangement in a magnetic cell common to every arm
of every star, its magnetic space group identified again from the structure
and shown next to the table’s, the spin listing, the MAGMOM line, one VESTA
file per row and the magnetic-cell POSCAR. An irrep that the spins of the
element do not carry is refused with the decomposition at its k point.
See also
Theory: Arrangements of the magnetic isotropy subgroups on a crystal — the spin modes that transform with the matrices of the table, why the group of every arrangement is the row’s group by construction, copies, the free amplitudes and the coupled case.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, one irrep |
|
terminal: the decomposition at the k point, the copies of the irrep, the magnetic cell, the table of |
several irreps |
|
the table of every irrep alone (as |
the same, tables only |
add |
the tables with the check column; no blocks, no VESTA files; report |
one direction |
add |
one block: the row of its stratum, its group from the table and from the structure; numbers are amplitudes |
weights of the irreps |
add |
every irrep’s part scaled to largest moment 1, then weighted (finite, nonzero, none below 1e-3 of the largest; negative: the other domain) |
two elements |
|
one arrangement per choice of copies (Al FM, Ni FM, Ni AFM), tagged |
one choice of copies |
add |
one copy number per irrep, as listed under |
conventional cell (I, F, C, A, R lattices) |
add |
every arrangement written in the conventional cell times the smallest multiples commensurate with every arm (the same table and groups); a |
cd example/05-01_mag_irrep
crystod-mag -c POSCAR_Mn3GaN_Pm-3m --element Mn --irrep R4-
* Copies in the spin representation *
R4-: 1 copy: Mn [AFM (q != 0)]
* Magnetic cell *
Magnetic (commensurate) supercell: 2x2x2 (8 primitive cells, 40 atoms), common to every arm of R4-
spin modes: |O(g) Psi - Psi D(g)| <= 2.3e-16 over the elements of G/T_S, Psi^T Psi = c 1 within 2.2e-16
* Order parameter directions and isotropy subgroups *
No. irrep subgroup size index structure
1 R4-(0,0,a) 139.540 I_c4/mmm 2 6 139.540 same
2 R4-(a,a,a) 166.102 R_I-3m 2 8 166.102 same
3 R4-(0,a,a) 74.562 I_bmma 2 12 74.562 same
4 R4-(a,a,b) 12.63 C_c2/m 2 24 12.63 same
5 R4-(0,a,b) 12.64 C_a2/m 2 24 12.64 same
6 R4-(a,b,c) 2.7 P_S-1 2 48 2.7 same
* 2: R4-(a,a,a) -> 166.102 R_I-3m *
order parameter: a = 1
reproduce: --order-parameter "0.57735 0.57735 0.57735"
moments: equal on all 24 Mn sites (|m| = 1.0000)
net moment over the magnetic cell: [0.0, 0.0, 0.0]
magnetic space group: R_I-3m (BNS 166.102, type IV), 24 operations, 12 with time reversal
magnetic point group: -3m1'
UNI symbol: R-3m.1'_c[R-3m]; point group -3m.1'
check (on the written moments): table 166.102, structure 166.102 same; 96 elements of G1'/T_S fix the moments, 96 the order parameter
...
How to read it:
Rows are those of
crystod-mag --parent SG --irrep ..., in the same order, with the same labels, groups, sizes and indices: the arrangement of a row is built from the order parameter the table identifies its group from, so its group is the row’s group by construction. The columnstructureand thecheck (on the written moments):line are the independent check, made on the moments exactly as theMAGMOMline, the listing and the VESTA file carry them (4 decimals, moment tolerance 3e-4 with the largest moment 1): spglib on the magnetic cell with all atoms and moments (in the cell of its translations without time reversal, the larger cell as fallback), and the numbers of elements of G1’/T_S that fix the moments and the order parameter. A disagreement is printed as a NOTE on the row.Reproduce. Every block ends its header with
reproduce: --order-parameter "..."(and--copies ...when an irrep has several copies): those numbers, given back with the same--irrep, rebuild the same written moments, weights included.Letters. A row with several letters is shown with generic values (the ones printed, largest 1): other values with the same pattern of zero and equal components give the same group, and symmetry does not fix them. The
moments:line tells whether the moments of that representative are equal on all sites, take several sizes or vanish on some sites.Direction components refer to the irrep basis of the table (as with
--parent), not to the axes of the POSCAR:GM4+(0,0,a)of Mn3GaN puts the moments along y. The spin listing gives the moments.Copies. An irrep that occurs more than once (GM4+ twice on Mn of Mn3GaN; several elements or Wyckoff orbits) gives one arrangement per choice of copies (
c1,c2, … in the block titles and file names). The copies are split by element, by Wyckoff orbit and, at q = 0, into the ferromagnetic combination and the antiferromagnetic ones, named by multipole rank as the--qpointreport names its bases. Any mixture of copies with the same order parameter has the same group.Several irreps. The arrangements are those of the coupled rows (every irrep condensed); the parts are scaled to largest moment 1 each and weighted. Without
--coefficientsevery row takes the first of the weight sets 1 : 1, 1 : 0.5, 1 : 0.75, 1 : 0.25, 1 : 0.6, 1 : 0.4 (three irreps: analogous sets) under which no site that a part reaches has a vanishing moment, and says why when it is not 1 : 1 (weights: 1 : 1 would cancel the moments on 8 sites; Mn3GaN R4- M3-: rows 1, 5, 6, 13 use 1 : 0.5). Themoments:line separates sites that no irrep reaches (zero on 8 sites (no moment in R4-, M3-)) from moments that given--coefficientscancel (cancelled on 8 sites by the weights).--irrep R4-alone exports the rows of one irrep.Many copies. All choices of copies are exported while there are at most 6 per row; otherwise only the first, with a NOTE, and
--copiesselects another one.--order-parameterruns save their report as..._opand tag their blocks and files with the copies.Conventional cell. The magnetic cell is a diagonal supercell of the primitive cell, which is awkward to look at for a centred lattice.
--conventionalwrites every arrangement in the conventional cell instead: the cell that--qpoint --conventionaluses for the same POSCAR (the conventional matrix of the centring letter applied to the primitive cell as given, so the input should be the standard primitive cell ofphonopy --symmetry, a PPOSCAR), times the smallest multiples along its axes for which k . T is an integer for every arm of every star. The spin modes, weights, rows and the counts over G1’/T_S stay on the diagonal magnetic cell (named in theMagnetic (commensurate) supercell:line); the moments are re-tiled onto the conventional cell, which is exact because the arrangement is periodic under every such translation. The spin listing, theMAGMOMline, the site counts of themoments:line, the net moment, the files and the columnstructure(spglib on the conventional cell as written) refer to the conventional cell, and thereproduce:line ends with--conventional. On a P lattice only the names and theDisplay cell:line change. An input primitive cell off the standard primitive setting gives a cell that is not the conventional one, and a NOTE under theDisplay cell:line says so (the arrangements hold for the cell as written).
Errors
An irrep that the spins of the element(s) do not carry, an unknown label, an
irrep given twice, --qpoint (the label carries its k point) and the options
of the --qpoint report (--superpose, --show-operations, --subgroup-poscar, --phonon-angular-momentum) stop
with an ERROR: line and no files; every invalid irrep is reported at once:
$ crystod-mag -c POSCAR_Mn3GaN_Pm-3m --element Mn --irrep R4+
ERROR: R4+ is not contained in the spin representation of Mn at R [0.5, 0.5, 0.5]; the decomposition there is 1.0 [R1-(1)] + 1.0 [R3-(2)] + 1.0 [R4-(3)] + 1.0 [R5-(3)]
--coefficients with a zero, nan or inf value, the wrong number of values
or a weight below 1e-3 of the largest (it would vanish in the written
moments), --copies beyond the copies of an irrep, and a -c file that is
not a POSCAR are refused the same way. On a structure off its ideal
positions whose irreps the survey cannot label at the given --tolerance,
the decomposition is printed with irrep_N names and a NOTE, the copies
come from the irrep matrices alone, and every row is still checked on the
written moments; symmetrizing the structure first is better.
The magnetic cell is the diagonal supercell commensurate with every arm of
every star; it is refused above 64 primitive cells or 12 cells along one
axis. With --conventional the conventional cell is refused above 12
conventional cells along one axis or 256 primitive cells; the ERROR names it
and suggests running without --conventional.
Python API
from phonopy.interface.vasp import read_vasp
from crystod import mag
result = mag.magnetic_irrep_arrangements(read_vasp("POSCAR_Mn3GaN_Pm-3m"), ["Mn"], ["R4-", "M3-"])
[(a.number, a.row.label, a.row.bns_number, a.check) for a in result.arrangements][:2]
# [(1, 'R4-(0,0,a) M3-(0;0;d)', '129.421', 'same'), (2, 'R4-(0,a,a) M3-(d;0;0)', '62.453', 'same')]
a = result.arrangements[0] # mag.MagneticArrangement
a.letters, a.moment_text, a.magmoms.shape
one = mag.magnetic_irrep_arrangements(read_vasp("POSCAR_Mn3GaN_Pm-3m"), ["Mn"], ["R4-", "M3-"],
order_parameter=["0", "0", "a", "0", "0", "b"])
one.stratum # 1
MagneticIrrepAnalysis has irreps, kpoints, decompositions, copies
(IrrepCopy: irrep, number, element, orbit, kind, rank_name,
intertwiner), singles, rows, arrangements, the magnetic cell
(cell_matrix, supercell_lattice, supercell_symbols,
supercell_positions: the written cell; supercell_sizes: the diagonal
cell of the spin modes; conventional_centring, conventional_multiples
with conventional=True) and residuals; MagneticArrangement has row,
copies, tag, order_parameter, letters, field, magmoms,
net_moment, group, responses, collinear, check, note,
weight_reason, reproduce (the --order-parameter numbers of the block)
and vesta_name; magmoms are the written (rounded) moments.
copies=[i, j] selects one choice of copies, conventional=True writes
the arrangements in the conventional cell (magmoms then follow its atoms;
field stays on the diagonal cell). The function writes no files.
05-02. Symmetry-adapted spin bases#
Example directory: example/05-02_spin_basis (testsuite section 05-02); after pip install: crystod-example 05-02
Note
The example/... paths on this page are those of a clone of the repository,
used from its root. After pip install CrystOD, run crystod-example CC-NN (the
number of the example directory) in your working directory first: it copies
that directory and the shared test_POSCARs/ into ./example/, and the
commands then run from there unchanged (see crystod-example).
Which magnetic orders can the moments on the sites of one element take in
this crystal? crystod-mag enumerates every symmetry-distinct ferromagnetic
and antiferromagnetic arrangement at a q point, sorted by irrep, and prints
each with its per-atom spin directions, a ready-to-paste VASP or Quantum
ESPRESSO magnetization input, a POSCAR of the magnetic cell and a VESTA file
with spin arrows. The enumeration follows the cluster-multipole /
symmetry-adapted multipole moment (SAMM) framework of M.-T. Suzuki et al.
[PRB 95, 094406 (2017); PRB 99, 174407 (2019)].
See also
Theory: Spin bases as axial-vector SALCs — why the spins are treated as axial vectors, how the FM dipole is split from the higher multipoles and the multipole ranks are assigned, how complex and two-dimensional irreps get real partners, and which combined directions symmetry singles out.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, q point |
|
terminal: the spin irrep decomposition, then every basis vector with its spins, net moment, magnetic space group and |
the same |
add |
the same with the Quantum ESPRESSO |
the same at q ≠ 0 |
|
the same in the commensurate magnetic supercell, with the order-parameter directions of every basis and their magnetic isotropy subgroups ( |
primitive POSCAR, magnetic element |
|
terminal: the spin irrep decomposition at every special k point; file: the report |
crystod-mag -c example/test_POSCARs/221_PPOSCAR_AlNi3 --element Ni --qpoint 0 0 0
crystod-mag -c example/test_POSCARs/221_PPOSCAR_AlNi3 --element Ni --qpoint 0 0 0 --format qe
# survey mode: spin-multipole irreps at ALL special k points
crystod-mag -c example/test_POSCARs/221_PPOSCAR_AlNi3 --element Ni
The output opens with the magnetic sites and the complete irrep enumeration — for the Ni 3c cluster of AlNi3 (the Mn3Ir geometry), 9 spin degrees of freedom decompose into exactly three symmetry-adapted families:
* Structure *
Space group: Pm-3m (#221)
Magnetic sites: Ni x 3
Ni1: [0.5, 0.5, 0.0]
Ni2: [0.5, 0.0, 0.5]
Ni3: [0.0, 0.5, 0.5]
* Selected Q point *
Selected q-point: GAMMA = [0.0, 0.0, 0.0]
* Spin (axial-vector) representation *
Spin (axial-vector) representation on Ni sites: 9 dimensions
Decomposition: 2 x GM4+(3) + GM5+(3)
* Symmetry-adapted spin bases *
GM4+(3): dim 3 [FM, dipole]
GM4+(3): dim 3 [AFM, octupole]
GM5+(3): dim 3 [AFM, octupole]
Every basis vector is then printed with its per-atom spin directions, net moment, its magnetic space group and class (section 05-03), and the magnetization input. The [111] component of the GM4+ cluster octupole is the experimentally realized 120-degree structure of Mn3Ir — note the exactly vanishing net moment:
* GM4+(3) [AFM, octupole] *
...
component 111:
Ni1: S = [ 0.4082, 0.4082, 0.8165]
Ni2: S = [ 0.4082, -0.8165, -0.4082]
Ni3: S = [-0.8165, 0.4082, -0.4082]
net moment: [0.0, 0.0, -0.0]
magnetic space group: R-3m' (BNS 166.101, type III), 12 operations, 6 with time reversal
magnetic point group: -3m'
UNI symbol: R-3m'; point group -3m'
allowed by symmetry: net magnetization along [11-1]; electric polarization none; linear magnetoelectric effect none; piezomagnetism 4 components
collinear order: noncollinear (not classified)
spin directions (all atoms, POSCAR order):
Al1: [0, 0, 0]
Ni2: [0.4082, 0.4082, 0.8165]
Ni3: [0.4082, -0.8165, -0.4082]
Ni4: [-0.8165, 0.4082, -0.4082]
MAGMOM = 0 0 0 0.4082 0.4082 0.8165 0.4082 -0.8165 -0.4082 -0.8165 0.4082 -0.4082
...
* Output files *
...
Spin basis written to: POSCAR_AlNi3_spin_GM4+_octupole_111.vesta
...
Supercell POSCAR written to: S221_PPOSCAR_AlNi3_GM (atom order of the MAGMOM lines)
Report written to: samm_AlNi3_Ni_GM (this output)
How to read the output.
* Spin (axial-vector) representation *: the spin degrees of freedom of the magnetic sites (3 per site, 9 for the three Ni of AlNi3) and their irreps at q;2 x GM4+(3)is the three-dimensional irrep GM4+ occurring twice.* Symmetry-adapted spin bases *: one line per copy of an irrep. FM and AFM refer to the net moment; every AFM basis satisfiessum_i S_i = 0exactly. The copies of an irrep are named after the ranks at which it occurs among the magnetic multipoles, lowest first: dipole (rank 1, the FM copy), quadrupole (rank 2), octupole (rank 3), hexadecapole (rank 4), …; an AFM basis is not always an octupole or higher (the in-plane orders of Cr2O3 areGM3-(2) [AFM, quadrupole], section 05-03).For the Mn3Ir-type Ni 3c cluster of AlNi3,
9 dims = 2 x GM4+(3) + GM5+(3): the GM4+ (T1g) cluster dipole (FM), the GM4+ (T1g) cluster octupole (AFM: the experimentally realized 120-degree structure of Mn3Ir, which shares the irrep with the dipole and hence allows the anomalous Hall effect), and the GM5+ (T2g) cluster octupole (AFM).component 111:one exported arrangement of the basis, tagged by its direction (the tags are listed in section 05-03);Ni1: S = [...]the spin of each magnetic site,net momenttheir sum.magnetic space group…collinear order: the symmetry of the arrangement, the responses it allows and its class (section 05-03).spin directions (all atoms, POSCAR order)andMAGMOM =: printed by default for every basis vector, the spins of all atoms of the magnetic cell, in the atom order of the magnetic-cell POSCAR written by the run.Survey mode (no
--qpoint): the spin (axial-vector) irrep decomposition is listed for every special k point of the space group — the magnetic counterpart of thecrystodSALC survey (e.g. for AlNi3: GM:2.0 [GM4+(3)] + 1.0 [GM5+(3)], R:R2+ + R3+ + R4+ + R5+, …). At a time-reversal-invariant point a complex irrep and its conjugate are counted once under their pair label, as with--qpoint(2.0 [GM3+GM4+(2)]for the Ti sites of P4/m Ti2Ga3). The main command prints the same terms in the format of its orbital SALC, with the little group of every k point:crystod -c POSCAR --element Ni --axial(see Spin configurations).
Quantum ESPRESSO input (--format qe)
With --format qe, the same 120-degree octupole becomes (Ni split into three
types because the three spins point in three different directions):
Quantum ESPRESSO noncollinear magnetization (&SYSTEM):
noncolin = .true.
! type 1 (Al): Al1 -- non-magnetic
! type 2 (Ni1): Ni2 S = [0.4082, 0.4082, 0.8165]
starting_magnetization(2) = 1.0000
angle1(2) = 35.2611
angle2(2) = 45.0000
! type 3 (Ni2): Ni3 S = [0.4082, -0.8165, -0.4082]
starting_magnetization(3) = 1.0000
angle1(3) = 114.0928
angle2(3) = -63.4378
! type 4 (Ni3): Ni4 S = [-0.8165, 0.4082, -0.4082]
starting_magnetization(4) = 1.0000
angle1(4) = 114.0928
angle2(4) = 153.4378
! angle1 = polar angle from the z axis (deg); angle2 = azimuth from the x axis in the xy plane (deg).
! Split Ni into the types above in ATOMIC_SPECIES / ATOMIC_POSITIONS to realize this spin arrangement.
The Quantum ESPRESSO block:
noncolin = .true.with per-typestarting_magnetization(i)/angle1(i)(polar angle from z) /angle2(i)(azimuth from x) for the&SYSTEMnamelist, where the magnetic element is split into one atom type per distinct spin direction (the atom membership of each type is printed as comments forATOMIC_SPECIES/ATOMIC_POSITIONS).
Combined directions of two- and three-dimensional bases
Combined directions of two- and three-dimensional bases. The components
of a two- or three-dimensional basis are the (a,0) / (a,0,0) directions of
its order parameter. The (a,a) / (a,a,0) direction, the equal-weight
combination of two components, is exported as well wherever symmetry singles
it out (it then has an isotropy subgroup of its own). A three-dimensional
x/y/z basis also gets 111 (x + y + z), the (a,a,a) direction; another
three-dimensional basis gets its first singled-out pair and the sum of the
three when singled out. A combination is named after the sum of the
component directions (the spin direction of a dipole), else in the
order-parameter notation over the components in their order: aa (two
components along the same axis), aa0 and aaa (comp<N> components):
crystal |
basis |
components |
combination |
|---|---|---|---|
AlNi3 (Pm-3m), Ni, Γ |
GM4+ dipole |
|
|
AlNi3, Ni, M |
M5+ / M5- |
|
|
CeRu2Si2 (I4/mmm), Ce, Γ |
GM5+ dipole |
|
|
TiO2 (P4_2/mnm), Ti, Γ |
GM5+ dipole |
|
|
AlNi3, Ni, R |
R4+ / R5+ |
|
|
Y2Ir2O7 (Fd-3m), Ir, Γ |
GM4+ octupole |
|
|
Pm-3, m-3, 23 (e.g. 200_PPOSCAR_NaCl7) |
GM4+ |
|
|
hexagonal and trigonal (Cr2O3, LuFeO3) |
E irreps |
|
none |
YMnO3 (P6_3/mmc), Mn, K |
K3 / K5 |
|
none / |
Only singled-out arrangements are added: where the components of a
comp<N> basis are not the (a,0,0)-type partners, fewer combinations come
out (none for the Se GM4+ of Rb3AsSe16, Fd-3), never a general direction.
At q ≠ 0 the combinations are taken over the rows w of a copy only, and every
one-parameter order-parameter direction that neither a component nor a
combination lies on is exported as dir<k> (section
05-06).
Output files
Output files.
POSCAR_<formula>_spin_<irrep>_<dipole|quadrupole|octupole|...>_<direction>.vesta, one per component (e.g.POSCAR_AlNi3_spin_GM4+_octupole_111.vestafor the 120-degree Mn3Ir-type state), with spin arrows on the magnetic atoms scaled so that the largest spin gets--amplitudeAngstroms (default 1.5). At q ≠ 0 the multipole part is the q name, and the components include the partners i w and thedir<k>directions (POSCAR_YMnO3_spin_K3_K_comp2.vesta,POSCAR_YMnO3_spin_K5_K_dir1.vesta).samm_<formula>_<element>_<q>(samm = symmetry-adapted magnetic moments): the whole report of a--qpointrun, saved exactly as printed in the current directory, next to the VESTA files, so that> fileis not needed:samm_AlNi3_Ni_GMabove,samm_AlNi3_Ni_Rfor--qpoint Ror--qpoint 0.5 0.5 0.5. The formula is the reduced formula of the primitive cell (as in the VESTA names), the element is--element, and the q point is named as in the lineSelected q-point:, withGMfor GAMMA;--conventionaladds_conv, as it does to the VESTA names. A point off the special points is named by its line or plane (DTfor--qpoint 0 0 1/4), so two points of one line share the name, as their VESTA files do. Warnings on the standard error stream are not in the file.S<POSCAR name>_<q>: the magnetic cell of the MAGMOM lines as a VASP 5 POSCAR with direct coordinates, in the current directory. The name isSbefore the file name of-cwithout its directory, and the q part of the report name (GMfor GAMMA, a character other than a letter, a digit,+or-replaced by_,_convwith--conventional):-c POSCAR_VO2_P42mnm --qpoint A(or--qpoint 0.5 0.5 0.5) writesSPOSCAR_VO2_P42mnm_A,-c POSCAR --qpoint GMwritesSPOSCAR_GM, and-c ../139_PPOSCAR_La3Ni2O7 --qpoint 0 0 0.5 --conventionalwritesS139_PPOSCAR_La3Ni2O7_X_conv. Its cell and atom order are those of thespin directionslistings, the MAGMOM lines and the VESTA files, so a MAGMOM line goes into the INCAR of a noncollinear run (LNONCOLLINEAR = .TRUE.) on this POSCAR as it is (it has 3 numbers per atom of the file). The species keep the blocks of the input (V Owith16 32for the 2x2x2 cell of VO2 at A), so the POTCAR of the input cell fits. At q = 0 the cell is the input cell itself (the conventional cell with--conventional).The line under
* Output files *reads(atom order of the MAGMOM lines), or(atom order of the spin listings)with--format qe, whose input has no MAGMOM line (the Quantum ESPRESSO types then group atoms by direction, not by position in the file).One report and one magnetic-cell POSCAR are written per run, and a file of the same name is overwritten; a run that stops with an
ERRORwrites neither.The survey mode (no
--qpoint) and--classify-onlywrite no VESTA file and no magnetic-cell POSCAR, but save their report, exactly as printed, assamm_<formula>_<element>_survey(samm_AlNi3_Ni_survey) andsamm_<formula>_<element>_<q>_classify(samm_TiO2_Ti_GM_classify; the q part as above,_convincluded), announced by* Output files */Report written to: ... (this output)like the report of a--qpointrun.--no-files(only in these two modes) prints without saving; so does the environment variableCRYSTOD_NO_REPORT=1.--classify-onlywith--subgroup-poscarsaves no file.
Magnetic supercell at q ≠ 0
Magnetic supercell at q ≠ 0.
For q != 0 (e.g.
--qpoint R) the commensurate magnetic supercell is built automatically, with the Bloch phase applied to the spins and the MAGMOM/VESTA output referring to the supercell.The atoms of the supercell are listed atom by atom, as in the
SPOSCARof phonopy: each atom of the primitive cell is followed by its images in the other cells of the magnetic cell (for AlNi3 at R, the 2x2x2 cell listsAl1toAl8and thenNi9toNi32), so the species blocks of the input stay grouped and the POTCAR of the input cell matches theSPOSCAR_*file. The same holds for the conventional cell of--conventional.The supercell takes the denominator of each coordinate of q along the primitive axes, up to 12; a q that does not fit such a cell (
0 0 0.15, that is 3/20) is refused with an ERROR that names the smallest commensurate supercell (1x1x20), instead of a cell that would cut the modulation.
Options
Options.
option |
meaning |
|---|---|
|
the magnetic element whose sites carry the spins |
|
the q point, as a name ( |
|
|
|
the arrow length of the largest spin in the VESTA files, in Angstrom (default 1.5) |
|
exports the spin structures in the conventional cell instead of the primitive cell (VESTA files get a |
|
the symmetry tolerance (default 1e-5, a distance in Angstrom), used by spglib and for matching the sites onto each other, both in the spin representation and in the collinear classification. The Bloch phases are built from the site positions symmetrized by the space group, so a structure slightly off its ideal positions decomposes and classifies like the ideal one (an AlNi3 cell with the Ni coordinates off by about 1e-4 gives |
|
the magnetic-group table alone; the operations kept and broken (section 05-03) |
|
survey and |
|
at q ≠ 0: the bases placed on a distorted structure whose cell is a magnetic cell of q, every direction aligned with its structural domain (section 05-07) |
Requirements
Requirements. The input must be a primitive cell. It is used as given
(the spin listings and MAGMOM follow its atom order), and a conventional cell
or a supercell is refused with the same one-line ERROR in the survey and the
--qpoint modes (the PPOSCAR of phonopy --symmetry is a primitive cell).
For the two-atom cubic cell of bcc Fe:
ERROR: the input cell is not primitive (2 atoms, the primitive cell has 1); crystod-mag uses the cell as given (the spin listings and MAGMOM follow its atom order) and needs a primitive cell, such as the PPOSCAR of phonopy --symmetry (--conventional then exports the spin structures in the conventional cell).
05-03. Magnetic space groups and collinear classes (--classify-only)#
Example directory: example/05-03_magnetic_group (testsuite section 05-03); after pip install: crystod-example 05-03
Which of the magnetic orders that the symmetry of the crystal admits is an
altermagnet, and which one allows a magnetoelectric effect? Every spin basis
vector that crystod-mag --qpoint exports is also an arrangement of moments
with its own symmetry: for each exported component the command gives the magnetic space
group (MSG) of that arrangement, its magnetic point group (MPG), the
macroscopic responses the MPG allows by symmetry (a net magnetization, an
electric polarization, the linear magnetoelectric effect, piezomagnetism) and, for a collinear
arrangement, its class: ferromagnet, ferrimagnet, conventional
antiferromagnet, or altermagnet with its d-, g- or i-wave type.
See also
Theory: Magnetic space groups and collinear classes of spin arrangements — the BNS types and symbols, the magnetic point group and the UNI symbol, the response tensors, the rules of the collinear classification and of the altermagnet wave type, the real spin field at q ≠ 0, and the order-parameter reading of the kept and broken operations.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, q point |
|
terminal: the list of bases and the table |
the same |
without |
the listing of section 05-02, each component with five more lines: MSG, MPG, UNI symbol, allowed responses, collinear order |
the same |
add |
the operations of the paramagnetic group that each component keeps and breaks (below) |
--classify-only prints the table alone:
crystod-mag -c example/test_POSCARs/136_PPOSCAR_TiO2 --element Ti --qpoint GM --classify-only
...
* Symmetry-adapted spin bases *
GM3+(1): dim 1 [FM, dipole]
GM5+(2): dim 2 [FM, dipole]
GM5+(2): dim 2 [AFM, octupole]
GM2+(1): dim 1 [AFM, octupole]
* Magnetic groups and collinear classes *
basis component MSG (BNS) type MPG net magnetization class MSG (UNI)
GM3+(1) [FM, dipole] z P4_2/mn'm' (136.501) III 4/mm'm' along z ferromagnet P4_2/mn'm'
GM5+(2) [FM, dipole] 1m10 Cmm'm' (65.486) III mm'm' along [1-10] ferromagnet Cmm'm'
GM5+(2) [FM, dipole] 110 Cmm'm' (65.486) III mm'm' along [110] ferromagnet Cmm'm'
GM5+(2) [FM, dipole] x Pnn'm' (58.398) III mm'm' along x ferromagnet Pnn'm'
GM5+(2) [AFM, octupole] 1m10 Cmm'm' (65.486) III mm'm' along [1-10] altermagnet (d-wave) Cmm'm'
GM5+(2) [AFM, octupole] 110 Cmm'm' (65.486) III mm'm' along [110] altermagnet (d-wave) Cmm'm'
GM5+(2) [AFM, octupole] x Pnn'm' (58.398) III mm'm' along y altermagnet (d-wave) Pnn'm'
GM2+(1) [AFM, octupole] z P4_2'/mnm' (136.499) III 4'/mmm' none altermagnet (d-wave) P4_2'/mnm'
MSG: magnetic space group from the moment directions (BNS number, type I-IV);
MPG: magnetic point group; net magnetization: the direction allowed by symmetry;
class: from the signs of the moments along their common axis only (nonrelativistic).
MSG and MPG symbols are in the standard BNS setting; directions are in the Cartesian axes of the input cell.
MSG (UNI): unified symbol (Campbell et al., Acta Cryst. A78, 99 (2022)), the BNS symbol with .1 (type I) or .1' (II); type IV: .1'_t[F], t the anti-translation, F the family space group.
...
* Output files *
Report written to: samm_TiO2_Ti_GM_classify (this output)
...
Reference (collinear classes): L. Smejkal, J. Sinova and T. Jungwirth, Phys. Rev. X 12, 031042 (2022).
How to read the output.
component: the tag of the listing and of the VESTA file name of the component:x,y,z,110,111,1m10,aa, …, with_2,_3when the same tag occurs again in the same basis (x_2is the secondxcomponent and the file..._x_2.vesta), andcomp<N>for a component without a single direction: the N-th basis vector of its space, counted from 1 like the mode numbers of the other commands (POSCAR_AlNi3_spin_R4+_R_comp1.vestais the first of the three R4+ components). At a q that is not time-reversal invariant a d-dimensional basis has 2d components, the rows w (comp1..comp<d>) and their partners i w (comp<d+1>..comp<2d>, or the axis tag of a collinear row);dir<k>is a one-parameter order-parameter direction exported on its own (section 05-06).MSG (BNS),magnetic space group: the group of the arrangement as spglib finds it, with its Belov-Neronova-Smirnova (BNS) symbol and number (P4_2'/mnm', 136.499) and its type: I, no operation with time reversal; II, the grey group of a non-magnetic crystal; III, half of the operations combined with time reversal (primed); IV, a lattice translation combined with time reversal (the subscript of the lattice letter,P_I,C_c,R_I). The operations are counted in the magnetic cell.MPG,magnetic point group: derived from the BNS symbol and checked against the operations (a symbol that disagrees is printed with?).MSG (UNI),UNI symbol: the unified symbol of Campbell et al., Acta Cryst. A78, 99 (2022): the BNS symbol with.1(type I) or.1'(type II), unchanged for type III (P4_2'/mnm'); a type IV group reads.1'_t[F], t the anti-translation and F the family space group (P_I4/mncisP4/mnc.1'_I[I4/mmm]).net magnetization,allowed by symmetry: the direction of a net magnetization the MPG allows, in the Cartesian axes of the input cell (along z,in the xy plane,any direction,none), and the numbers of independent components of the linear magnetoelectric tensor and of the piezomagnetic tensor. Symmetry decides what is allowed, not how large it is: an allowed response can be small. An AFM basis vector that shares its irrep with the FM dipole (the GM4+ octupoles of AlNi3) allows a net magnetization by symmetry although its own moments add up to zero.electric polarization, after the net magnetization: the direction of a spontaneous electric polarization the MPG allows, in the same words. The polarization is a time-even polar vector, so time reversal does not constrain it: it is allowed when the point group of the spatial parts of all operations is polar, and an MPG with-1'(Cr2O3, -3’m’) forbids it as-1does. The direction is followed by(induced by the spin order)when the crystal without spins (the same structure, same tolerance) allows no polarization, and by(the crystal without spins: along z)when that crystal allows fewer components than the spin arrangement; nothing follows it when both allow the same components.class,collinear order, from the signs of the moments along their common axis only (nonrelativistic):ferromagnetandferrimagnet(nonzero net moment),compensated, sublattices not related by symmetry,antiferromagnet (T tau)(a pure translation exchanges the sublattices) andantiferromagnet (P T)(the inversion does; the bands stay spin degenerate in the nonrelativistic limit),altermagnet (d-wave)/(g-wave)/(i-wave)(only rotations or mirrors exchange the sublattices), andnoncollinear (not classified). For an altermagnet the listing adds the lowest-order spin splitting allowed by symmetry (~ k_x k_yfor rutile) and the spin-degenerate planes, in the Cartesian axes of the input cell.
More on reading the output
Two frames meet in one row: the MSG and MPG symbols are those of the standard BNS setting, whatever the orientation of the input cell, while the net magnetization, the spin-splitting polynomials and the degenerate planes are given in the Cartesian axes of the input cell. A row
Pnn'm' ... mm'm' ... along yis therefore consistent: the standard setting of Pnn’m’ allows a moment along its own x axis, which is y of the input cell.magnetic space group: not identified by spglib (N operations, M with time reversal): the MSG, type and MPG cells of the table hold-, and the responses and the class are still computed from the operations.At q = 0 every AFM basis vector has a zero net moment, and the closing sentence
All AFM bases satisfy sum_i S_i = 0says so. A note under the table names every merged pair of a complex irrep and its conjugate (GM3+GM4+(2), the label thatcrystod-group --parentprints for the same pair); each copy of such a pair has two real partners.
The listing without --classify-only
Without --classify-only every component of the listing gets five more
lines after its net moment, here for the antiparallel pair along c (the
RuO2 / MnF2 order):
* GM2+(1) [AFM, octupole] *
component z:
Ti1: S = [ 0.0000, 0.0000, 1.0000]
Ti2: S = [ 0.0000, 0.0000, -1.0000]
net moment: [0.0, 0.0, 0.0]
magnetic space group: P4_2'/mnm' (BNS 136.499, type III), 16 operations, 8 with time reversal
magnetic point group: 4'/mmm'
UNI symbol: P4_2'/mnm'; point group 4'/mmm'
allowed by symmetry: net magnetization none; electric polarization none; linear magnetoelectric effect none; piezomagnetism 2 components
collinear order: altermagnet (d-wave; spin splitting allowed by symmetry ~ k_x k_y; spin-degenerate planes perpendicular to [100], [010])
spin directions (all atoms, POSCAR order):
...
Moment directions versus sign pattern (MnO2)
Moment directions versus sign pattern (MnO2). The MSG uses the moment directions, the class only their sign pattern, and the two can differ: in MnO2 (Pnnm) the antiparallel pair along z has the type I group Pnnm and along x the type III group Pnn’m’ (standard setting; here it allows a net magnetization along y of the input cell), while both are d-wave altermagnets:
crystod-mag -c example/test_POSCARs/58_PPOSCAR_MnO2 --element Mn --qpoint GM --classify-only
...
* Magnetic groups and collinear classes *
basis component MSG (BNS) type MPG net magnetization class MSG (UNI)
GM1+(1) [AFM, octupole] z Pnnm (58.393) I mmm none altermagnet (d-wave) Pnnm.1
GM2+(1) [FM, dipole] z Pn'n'm (58.397) III m'm'm along z ferromagnet Pn'n'm
GM4+(1) [FM, dipole] y Pnn'm' (58.398) III mm'm' along y ferromagnet Pnn'm'
GM4+(1) [AFM, octupole] x Pnn'm' (58.398) III mm'm' along y altermagnet (d-wave) Pnn'm'
GM3+(1) [FM, dipole] x Pnn'm' (58.398) III mm'm' along x ferromagnet Pnn'm'
GM3+(1) [AFM, octupole] y Pnn'm' (58.398) III mm'm' along x altermagnet (d-wave) Pnn'm'
...
Magnetoelectric Cr2O3 and hematite
Magnetoelectric Cr2O3 and hematite. On the four Cr sites of R-3c
(example/05-03_magnetic_group/Cr2O3/PPOSCAR_Cr2O3_R-3c, rhombohedral primitive cell
with the hexagonal c axis along z), the order + - + - along c (GM1- z) is the
Cr2O3 order: R-3’c’ (167.106), MPG -3’m’, a conventional antiferromagnet
(P T) with the linear magnetoelectric effect allowed by symmetry,
alpha_11 = alpha_22 and alpha_33. The order + - - + (GM1+ z, the
hematite order below the Morin transition) is the type I group R-3c
(167.103), a g-wave altermagnet with piezomagnetism allowed. With the
Python API, format_magnetic_responses prints the tensor forms:
from phonopy.interface.vasp import read_vasp
from crystod import mag
result = mag.analyze_spin_basis(read_vasp("Cr2O3/PPOSCAR_Cr2O3_R-3c"), "Cr", "GM")
cr2o3 = next(c for c in result.components if c.label.startswith("GM1-"))
print(cr2o3.group.bns_symbol, cr2o3.group.magnetic_point_group)
print("\n".join(mag.format_magnetic_responses(cr2o3.responses)))
R-3'c' -3'm'
net magnetization: forbidden by symmetry
electric polarization: forbidden by symmetry
linear magnetoelectric effect: allowed by symmetry, 2 independent components
alpha 1 2 3
1 alpha11 0 0
2 0 alpha11 0
3 0 0 alpha33
piezomagnetism: forbidden by symmetry
The orders with the moments in the plane belong to the two-dimensional irreps GM3+ and GM3-. Their partners are taken along the two-fold axis of R-3c that lies along a (x of this cell) and perpendicular to it (y), and every arrangement keeps a monoclinic group with that two-fold axis:
crystod-mag -c example/05-03_magnetic_group/Cr2O3/PPOSCAR_Cr2O3_R-3c --element Cr --qpoint GM --classify-only
...
GM3+(2) [FM, dipole] x C2/c (15.85) I 2/m along x ferromagnet C2/c.1
GM3+(2) [FM, dipole] y C2'/c' (15.89) III 2'/m' in the yz plane ferromagnet C2'/c'
GM3+(2) [AFM, octupole] y C2/c (15.85) I 2/m along x altermagnet (g-wave) C2/c.1
GM3+(2) [AFM, octupole] x C2'/c' (15.89) III 2'/m' in the yz plane altermagnet (g-wave) C2'/c'
GM3-(2) [AFM, quadrupole] x C2'/c (15.87) III 2'/m none antiferromagnet (P T) C2'/c
GM3-(2) [AFM, quadrupole] y C2/c' (15.88) III 2/m' none antiferromagnet (P T) C2/c'
GM3-(2) [AFM, quadrupole] y_2 C2'/c (15.87) III 2'/m none antiferromagnet (P T) C2'/c
GM3-(2) [AFM, quadrupole] x_2 C2/c' (15.88) III 2/m' none antiferromagnet (P T) C2/c'
...
The GM3+ octupole is the hematite order + - - + with the moments in the plane: C2/c with the moments perpendicular to the two-fold axis, C2’/c’ with the moments along it, where a net magnetization perpendicular to the moments (in the yz plane) is allowed by symmetry. The first GM3- copy is the order
of Cr2O3 with the moments in the plane, the second the order + + - -. Each AFM copy lists its partners in the order of the FM partners they transform like, not by their own moment directions: the first octupole row (moments along y) has the group of the FM dipole along x.
Complex conjugate pairs
Complex conjugate pairs. In P4/m the irreps GM3+ and GM4+ (and GM3- and GM4-) are one-dimensional and complex; each pair forms one real two-dimensional space, listed under its pair label. For the Ti 4k sites of Ti2Ga3 the in-plane FM dipole and the in-plane octupole both belong to GM3+GM4+:
crystod-mag -c example/test_POSCARs/83_PPOSCAR_Ti2Ga3 --element Ti --qpoint GM --classify-only
...
Decomposition: GM1+(1) + 2 x GM1-(1) + 2 x GM3+GM4+(2) + GM3-GM4-(2) + GM2+(1) + 2 x GM2-(1)
* Symmetry-adapted spin bases *
GM1+(1): dim 1 [FM, dipole]
GM1-(1): dim 1 [AFM, quadrupole]
GM1-(1): dim 1 [AFM, hexadecapole]
GM3+GM4+(2): dim 2 [FM, dipole]
GM3+GM4+(2): dim 2 [AFM, octupole]
GM3-GM4-(2): dim 2 [AFM, quadrupole]
GM2+(1): dim 1 [AFM, octupole]
GM2-(1): dim 1 [AFM, quadrupole]
GM2-(1): dim 1 [AFM, quadrupole]
* Magnetic groups and collinear classes *
basis component MSG (BNS) type MPG net magnetization class MSG (UNI)
GM1+(1) [FM, dipole] z P4/m (83.43) I 4/m along z ferromagnet P4/m.1
GM1-(1) [AFM, quadrupole] comp1 P4/m' (83.46) III 4/m' none noncollinear P4/m'
GM1-(1) [AFM, hexadecapole] comp1 P4/m' (83.46) III 4/m' none noncollinear P4/m'
GM3+GM4+(2) [FM, dipole] x P2'/m' (10.46) III 2'/m' in the xy plane ferromagnet P2'/m'
GM3+GM4+(2) [FM, dipole] y P2'/m' (10.46) III 2'/m' in the xy plane ferromagnet P2'/m'
GM3+GM4+(2) [AFM, octupole] x P2'/m' (10.46) III 2'/m' in the xy plane altermagnet (d-wave) P2'/m'
GM3+GM4+(2) [AFM, octupole] y P2'/m' (10.46) III 2'/m' in the xy plane altermagnet (d-wave) P2'/m'
GM3-GM4-(2) [AFM, quadrupole] z P2'/m (10.44) III 2'/m none antiferromagnet (P T) P2'/m
GM3-GM4-(2) [AFM, quadrupole] z_2 P2'/m (10.44) III 2'/m none antiferromagnet (P T) P2'/m
GM2+(1) [AFM, octupole] z P4'/m (83.45) III 4'/m none altermagnet (d-wave) P4'/m
GM2-(1) [AFM, quadrupole] comp1 P4'/m' (83.47) III 4'/m' none noncollinear P4'/m'
GM2-(1) [AFM, quadrupole] comp1_2 P4'/m' (83.47) III 4'/m' none noncollinear P4'/m'
...
component: the tag of the listing and of the VESTA file name (comp<N>: the N-th basis vector of its space).
GM3+GM4+(2): a complex irrep and its conjugate, merged into one real space (the physically irreducible representation).
GM3-GM4-(2): a complex irrep and its conjugate, merged into one real space (the physically irreducible representation).
The in-plane octupole shares the pair with the dipole, so it allows a net
magnetization in the plane by symmetry although its own moments add up to
zero, as the GM4+ octupoles of AlNi3 do. In Python,
SpinBasisAnalysis.pair_labels lists the merged pairs.
Classification at q != 0
q != 0. The classification refers to the real spin field of the magnetic
cell. At a q that is not time-reversal invariant that field depends on the
phase of the complex order parameter, and the components of a basis are its
real and imaginary parts, w and i w, which together span the real
order-parameter space; every one-parameter direction in it, with its
magnetic isotropy subgroup, is listed under * Order-parameter directions *
(section 05-06). For the NiO-type
order on the Na sites of rock salt at L (moments along [111]):
crystod-mag -c example/test_POSCARs/225_PPOSCAR_NaCl --element Na --qpoint L --classify-only
...
* Magnetic groups and collinear classes *
basis component MSG (BNS) type MPG net magnetization class MSG (UNI)
L2+(1) [AFM (q != 0)] 111 R_I-3c (167.108) IV -3m1' none antiferromagnet (T tau) R-3c.1'_c[R-3m]
...
At a time-reversal-invariant q != 0 a complex irrep and its conjugate are
listed under their pair label, as at q = 0, also when they are
two-dimensional. For Zn at A of ZnO (P6_3mc) the decomposition is
A2A3(2) + A5A6(4), the list of bases reads A5A6(4): dim 4 [AFM (q != 0)],
and the table has its four real partners as comp1 … comp4: the real
parts of the two complex partners followed by their imaginary parts.
Limits
Limits.
Only collinear arrangements are classified; a noncollinear component gets
noncollinear (not classified)(its MSG and responses are still given).The table describes the components as
crystod-magexports them: for an irrep of dimension two or three each exported component is one arrangement, and another partner direction can have another MSG.At q != 0 the components span the real order-parameter space of their basis (at a q that is not time-reversal invariant, the rows w and their partners i w), and every one-parameter direction of it is exported, so no magnetic isotropy subgroup depends on a phase convention; the components of a direction with two or more free parameters are not singled out (
freein* Order-parameter directions *).The MPG symbol is derived from the BNS symbol and checked; the operations (
MagneticGroupInfo.rotations,.translations,.time_reversals) are the authority.--classify-onlyneeds--qpoint; the survey mode (no--qpoint) does not classify.At q = 0 the partners of a two- or three-dimensional irrep lie along directions that symmetry fixes; which of several equivalent choices is printed is a convention (in a trigonal or hexagonal crystal the a axis of the input cell and the direction perpendicular to it, whichever of the equivalent in-plane axes spglib calls a), and a rotated input cell gives the same groups and classes with the directions rotated. The partners of a merged complex pair are not fixed by symmetry and follow a convention (Real partners of merged complex pairs and repeated irreps).
Python API
Python API.
from phonopy.interface.vasp import read_vasp
from crystod import mag
result = mag.analyze_spin_basis(read_vasp("136_PPOSCAR_TiO2"), "Ti", "GM")
for row in mag.magnetic_summary_rows(result):
print(row)
component = result.components[-1] # GM2+(1) [AFM, octupole], z
component.group.bns_number, component.collinear.wave, component.collinear.spin_splitting
# ('136.499', 'd-wave', ('k_x k_y',))
mag.magnetic_space_group(lattice, positions, numbers, magmoms),
mag.macroscopic_responses(group) and mag.classify_collinear(...) work on
any cell and spin field; their docstrings give the arguments.
analyze_spin_basis takes --tolerance as symprec and raises ValueError,
with the text of the ERROR lines of the command, for a cell that is not
primitive and for a q that does not fit a magnetic cell of at most 12 cells
along each axis. At q ≠ 0, result.directions lists the order-parameter
directions (mag.OrderParameterDirection) and every component carries
direction_label and direction_subgroup (section
05-06).
Operations kept and broken (--show-operations)#
Which operations of the paramagnetic group does a magnetic order keep, and
which does it break? With --show-operations every component also lists the
operations of the paramagnetic group G + theta G (I4/mmm1'), one per coset
of the lattice translations, split into those the arrangement keeps and
those it breaks.
See also
Theory: Operations kept and broken as an order-parameter condensation — why the kept operations are those on which the time-odd order parameter equals +1, and how they are matched to the operations of the crystal.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, q point |
|
the listing of section 05-02, each component with its kept and broken operations |
the same |
add |
the table, then the lists as a block |
For the Cu site of La2CuO4 (I4/mmm, the site of Ce in CeRu2Si2) with the moments along c:
crystod-mag -c example/05-02_spin_basis/La2CuO4_I4mmm/139_PPOSCAR_La2CuO4 --element Cu --qpoint GM --show-operations
* GM3+(1) [FM, dipole] *
component z:
...
magnetic space group: I4/mm'm' (BNS 139.537, type III), 16 operations, 8 with time reversal
...
operations of the paramagnetic group I4/mmm1' (16 {R|t} and 16 {R|t}' = theta{R|t}, one per coset; t in the conventional cell):
kept (16): {E|0} {C4z|0} {C2z|0} {C4z^3|0} {I|0} {IC4z|0} {IC2z|0} {IC4z^3|0}
{C2x|0}' {C2y|0}' {C2[110]|0}' {C2[1-10]|0}' {IC2x|0}' {IC2y|0}' {IC2[110]|0}' {IC2[1-10]|0}'
broken (16): {C2x|0} {C2y|0} {C2[110]|0} {C2[1-10]|0} {IC2x|0} {IC2y|0} {IC2[110]|0} {IC2[1-10]|0}
{E|0}' {C4z|0}' {C2z|0}' {C4z^3|0}' {I|0}' {IC4z|0}' {IC2z|0}' {IC4z^3|0}'
How to read the output.
{R|t}'is theta{R|t}, the operation followed by time reversal (the prime of the BNS symbols).R is named as in Bradley and Cracknell (E, I,
C4z,C4z^3,C2[1-10],IC2x= the mirror perpendicular to x; x, y, z along the axes of an orthogonal conventional cell, else[uvw]in the conventional cell), and t is given in the conventional cell.kept: the operations of the magnetic space group of the arrangement;broken: the other operations of G + theta G.Where q != 0 an operation can be kept both ways,
{R|t}with a translation of the magnetic cell and{R|t}'with an anti-translation (type IV); a note line says so.
--show-operations needs --qpoint. In Python,
crystod.spin_basis.parent_operations(analysis, component) gives the lists
as ParentOperation (symbol, kept, kept_primed).
05-04. Magnetic isotropy subgroups (--parent)#
Testsuite section 05-04; no structure file and no example directory are needed.
Which magnetic space groups can a magnetic order of a given k point give in a
space group, without any structure file? This is the magnetic counterpart of
crystod-group --parent SG --kpoint K:
every order-parameter direction of every irrep of the k point is listed with
its magnetic space group (BNS number and symbol), the cell size and the index.
See also
Theory: Magnetic isotropy subgroups —
the definition of the magnetic isotropy subgroup with time reversal acting as
−1 on the order parameter, why the table has directions that the structural
one lacks, how the groups are identified, and the validation against the
magnetic isotropy subgroups tabulated in ISOTROPY’s data_magnetic.txt.
Input |
Command |
Output |
|---|---|---|
space group, k point |
|
terminal: every irrep of the k point, each order-parameter direction with its magnetic space group, size and index |
space group, one irrep |
|
the same for that irrep |
space group, several irreps |
|
each irrep alone, then the coupled order parameters (below) |
crystod-mag --parent Pm-3m --qpoint GM
* Supergroup *
Pm-3m (No. 221)
* Kpoint *
GM
k = (0, 0, 0) in the primitive basis, star of 1 arm(s)
* Order parameter directions and isotropy subgroups *
irrep subgroup size index
GM1+(a) 221.92 Pm-3m 1 2
GM2+(a) 221.95 Pm-3m' 1 2
...
GM4+(0,0,a) 123.345 P4/mm'm' 1 6
GM4+(a,a,a) 166.101 R-3m' 1 8
GM4+(0,a,a) 65.486 Cmm'm' 1 12
GM4+(0,a,b) 10.46 P2'/m' 1 24
GM4+(a,a,b) 12.62 C2'/m' 1 24
GM4+(a,b,c) 2.4 P-1 1 48
GM5+(0,0,a) 123.342 P4'/mm'm 1 6
GM5+(a,a,a) 166.97 R-3m 1 8
GM5+(0,a,a) 65.486 Cmm'm' 1 12
GM5+(0,a,b) 10.46 P2'/m' 1 24
GM5+(a,a,b) 12.58 C2/m 1 24
GM5+(a,b,c) 2.4 P-1 1 48
...
How to read the output.
irrep: the irrep with an order-parameter direction (GM4+(0,a,a)). The irrep is ISOTROPY’s mGM4+; it is printed without them, as everywhere incrystod-mag. The order-parameter components refer to the same real irrep basis ascrystod-group --parent, which can differ from ISOTROPY’s by an orthogonal change.subgroup: the BNS number and symbol of the magnetic space group. GM4+ is the ferromagnet: moments along a cube axis give P4/mm’m’, along [111] R-3m’, along [110] Cmm’m’.The table has directions that the structural table does not have: moments in a (100) plane (P2’/m’) or a (110) plane (C2’/m’) are the general direction P-1 of
crystod-group --parent Pm-3m --irrep GM4+.size: the cell of the translations without time reversal, in parent primitive cells.index: [G1’ : H], the number of magnetic domains (6 for P4/mm’m’: ±x, ±y, ±z).
A single irrep off Γ (--irrep R4+)
At a k point off Γ the subgroups are of type IV wherever a translation reverses the order parameter:
crystod-mag --parent Pm-3m --irrep R4+ # mR4+ is accepted too
irrep subgroup size index
R4+(0,0,a) 140.550 I_c4/mcm 2 6
R4+(a,a,a) 167.108 R_I-3c 2 8
R4+(0,a,a) 74.562 I_bmma 2 12
R4+(0,a,b) 12.64 C_a2/m 2 24
R4+(a,a,b) 15.90 C_c2/c 2 24
R4+(a,b,c) 2.7 P_S-1 2 48
Options
Options.
--qpointtakes a special-point name or three primitive coordinates.--irreptakes one irrep, a pair label of a complex irrep and its conjugate (GM3+GM4+), or ISOTROPY’s spelling withm(mR4+).--parenttakes no structure file and no--element,--format,--amplitude,--conventional,--tolerance,--classify-onlyor--show-operations.
Python API
Python API.
from crystod import mag
for row in mag.magnetic_isotropy_subgroups("Pm-3m", "GM4+"):
print(row.label, row.bns_number, row.bns_symbol, row.uni_symbol, row.size, row.index)
# GM4+(0,0,a) 123.345 P4/mm'm' P4/mm'm' 1 6 ...
tables = mag.magnetic_isotropy_subgroups_at_kpoint("Pm-3m", "R") # {irrep: [rows]}
A row (MagneticIsotropySubgroup) also carries the type, the OG number, the
magnetic point group, the unitary subgroup (unitary_number,
unitary_symbol), n_free and n_operations; bad input raises
ValueError with the text of the command’s ERROR line.
Several irreps (coupled order parameters)#
Testsuite section 05-10; example directory example/05-10_mag_coupled (README
transcript only); after pip install: crystod-example 05-10.
A magnetic order often condenses more than one irrep at once. With several
irreps after --irrep, every irrep is listed alone (as with one irrep), and
then the coupled order parameters: the magnetic isotropy subgroups of the
directions in which every irrep given has a nonzero amplitude. This is the
magnetic counterpart of
crystod-group --parent SG --irrep A B.
See also
Theory: Coupled magnetic order parameters — the stabilizer in G1’ of a vector of the direct sum, why the enumeration is complete, the secondary order parameters, and how the table was validated.
Input |
Command |
Output |
|---|---|---|
space group, two or more irreps |
|
terminal: the table of every irrep alone, then the coupled directions with their magnetic space group, size and index |
one coupled direction |
|
terminal: the magnetic isotropy subgroup of that direction (BNS, UNI symbol, unitary subgroup, size, index, sublattice basis) |
crystod-mag --parent Pm-3m --irrep R4+ M3+
* Supergroup *
Pm-3m (No. 221)
* Coupled irreps *
R4+: order parameter dimension 3
M3+: order parameter dimension 3 (star of 3 arm(s) x small dim 1)
coupled order parameter dimension 6 (3 + 3)
* Order parameter directions and isotropy subgroups (R4+ alone) *
irrep subgroup size index
R4+(0,0,a) 140.550 I_c4/mcm 2 6
...
R4+(a,b,c) 2.7 P_S-1 2 48
* Order parameter directions and isotropy subgroups (M3+ alone) *
irrep subgroup size index
M3+(0;0;d) 127.397 P_C4/mbm 2 6
M3+(d;d;d) 229.143 Im-3m' 4 8
...
* Order parameter directions and isotropy subgroups (coupled) *
irrep subgroup size index
R4+(0,0,a) M3+(0;0;d) 127.397 P_C4/mbm 4 12
R4+(0,a,a) M3+(d;0;0) 62.454 P_Bnma 4 24
R4+(0,0,a) M3+(0;d;0) 63.467 C_amcm 4 24
R4+(0,0,a) M3+(d;d;0) 137.516 P_c4_2/nmc 8 24
R4+(0,a,a) M3+(0;d;d) 63.466 C_cmcm 8 48
R4+(a,a,a) M3+(d;d;d) 166.101 R-3m' 8 64
...
R4+(0,0,a) M3+(d;d;e) 129.415 P4'/nmm' 8 48
...
R4+(a,b,c) M3+(d;e;f) 2.4 P-1 8 384
Every irrep of the coupled table condenses with a nonzero
amplitude; every irrep keeps its own independent free
parameters (R4+: a, b, c; M3+: d, e, f).
...
How to read the output.
The blocks
(R4+ alone)and(M3+ alone)are the one-irrep tables, with the letters of the second irrep shifted (M3+(0;0;d)), so that every irrep keeps its own free parameters.Every row of the coupled table condenses all irreps given; a direction in which one irrep is zero belongs to that irrep’s own table. The 22 rows are every stratum of the coupled order parameter, one per class of conjugate subgroups in Pm-3m1’.
R4+(0,a,a) M3+(d;0;0)is the a⁻a⁻c⁺ pattern of the Pnma perovskites; as two magnetic irreps it gives the type IV group P_Bnma, whose unitary part is Pnma, in a cell of 4 parent cells with 24 domains.sizeandindexare those of the one-irrep table: the cell of the translations without time reversal, and [Pm-3m1’ : H].
crystod-mag --parent Pm-3m --irrep R4+ M3+ --order-parameter "0 a a d 0 0"
* Supergroup *
Pm-3m (No. 221)
* Magnetic isotropy subgroup *
R4+(0,a,a) M3+(d;0;0) -> 62.454 P_Bnma (BNS), type IV
UNI symbol: Pnma.1'_B[Bbmm]; magnetic point group mmm1'
unitary subgroup: Pnma (No. 62)
cell size 4, index 24
sublattice basis (parent primitive units): (2,0,0), (1,1,0), (0,0,2)
--order-parameter takes the components irrep after irrep, in the order of
--irrep, with the spellings of crystod-group --order-parameter (a,
-a, 2a, a+b, 0; separate words or one string with ; or ,). A part
may be zero. Irrational factors printed as rounded decimals (0.529b) are
refused, because the rounded direction has another subgroup. With one irrep,
--order-parameter gives the subgroup of that direction alone.
Reading notes: secondary order parameters, sizes and indices
Every row condenses all irreps given. The coupled table never repeats a one-irrep direction with the other irreps at zero.
Rows with the group of a one-irrep row. Some coupled rows have the same magnetic space group, size and index as a row of one irrep alone. These are the directions where the subgroup of one order parameter already allows the other irrep: the second irrep then appears as a secondary order parameter induced by the first, and the row is that one-irrep subgroup with the induced amplitude written out. In Pm-3m
--irrep GM4+ GM5+, 8 of the 12 one-irrep directions do so: the ferromagnet along [011], Cmm’m’, is the coupled rowGM4+(0,a,-a) GM5+(0,d,d)65.486 Cmm’m’ with the index 12 ofGM4+(0,a,a)alone, whereas the ferromagnet along z (P4/mm’m’) induces no GM5+ and has no coupled row of its own. In R4+ M3+ no direction does.Sizes. The cell of the coupled subgroup contains the cells of both irreps: R4+ (size 2) and M3+ (size 2 or 4) give 4 or 8.
Indices. The index is the number of domains of the coupled order. The coupled subgroup lies in the subgroup of each irrep alone, so its index is a multiple of theirs (P_C4/mbm: 12, twice the 6 of
R4+(0,0,a)and ofM3+(0;0;d)alone).Labels. The irreps are printed without the
mof ISOTROPY (mR4+), as everywhere incrystod-mag;mR4+is accepted as input.
Python API: coupled tables and time-even irreps
from crystod import mag
result = mag.coupled_magnetic_isotropy_subgroups("Pm-3m", ["R4+", "M3+"])
for row in result.coupled:
print(row.label, row.bns_number, row.bns_symbol, row.uni_symbol, row.size, row.index)
# R4+(0,0,a) M3+(0;0;d) 127.397 P_C4/mbm P4/mbm.1'_C[rP4/mmm] 4 12 ...
result.singles["R4+"] # the table of R4+ alone (letters a, b, c)
record = mag.coupled_magnetic_order_parameter("Pm-3m", ["R4+", "M3+"], "0 a a d 0 0".split())
record.row.bns_symbol, record.basis # P_Bnma, the sublattice basis
A distortion and a magnetic order can be combined from Python (the
command line does not take time-even irreps yet): parities gives every
irrep its time-reversal parity, "odd" (magnetic, the default) or "even"
(structural, time reversal acts as +D on it). The time-odd irreps are then
printed with ISOTROPY’s m:
result = mag.coupled_magnetic_isotropy_subgroups(
"Pm-3m", ["R4+", "GM4+"], parities=["even", "odd"])
# R4+ alone: grey groups (140.542 I4/mcm1', ...)
# R4+(0,0,a) mGM4+(0,0,d) 140.547 I4/mc'm' 2 12 -- an a0a0c- tilt with a ferromagnet along z
The rows are CoupledMagneticIsotropySubgroup records (irreps,
direction, components, BNS, type, UNI symbol, magnetic point group,
unitary subgroup, size, index, n_free, n_operations); bad input raises
ValueError.
05-06. Order-parameter directions of the spin bases#
Example directory: example/05-06_mag_directions (testsuite section 05-06); after pip install: crystod-example 05-06
Which magnetic orders does one irrep allow at a q point off Γ, and which
magnetic space group does each give? At a q that is not time-reversal
invariant (K of a hexagonal crystal, whose star has the two arms K and -K) the
order parameter of a d-dimensional copy is complex: the spin field
Re[η w exp(2πi q·T)] depends on the phase of η as well as on its size, and the
phase is a physical degree of freedom. crystod-mag --qpoint therefore exports
2d components per basis, the rows w (comp1 .. comp<d>, as before) and
their partners i w (comp<d+1> .. comp<2d>, the phase turned by 90 degrees;
the partner of a collinear row keeps its axis tag), and lists every
order-parameter direction of every basis in the magnetic cell, with
ISOTROPY’s label where the q point has a table.
See also
Theory: Order-parameter directions of the spin bases —
the group that acts on the real order-parameter space of a basis in the
magnetic cell, how the strata are found and matched to the rows of
crystod-mag --parent, and how the copies of an irrep that occurs more than
once are chosen.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, q ≠ 0 |
|
terminal: |
the same, table only |
add |
the direction block and the summary table, with the |
cd example/05-06_mag_directions
crystod-mag -c 194_PPOSCAR_YMnO3 --element Mn --qpoint K
* Symmetry-adapted spin bases *
K2(1): dim 2 [AFM (q != 0)]
K3(1): dim 2 [AFM (q != 0)]
K5(2): dim 4 [AFM (q != 0)]
K6(2): dim 4 [AFM (q != 0)]
* Magnetic cell *
Magnetic (commensurate) supercell: 3x3x1
q is not time-reversal invariant: the order parameter of a d-dimensional basis is complex, ...
* Order-parameter directions *
irrep direction in comp basis subgroup (BNS) size index component
K2(0.282a;a) (1,0) 193.258 P6_3'/m'c'm 3 6 comp1
K2(a;0.974a) (0,1) 193.261 P6_3/m'c'm' 3 6 comp2
K2(a;b) - 188.217 P-6'c'2 3 12 free (2 parameters)
K3(a;0.974a) (1,0) 193.255 P6_3/m'cm 3 6 comp1
K3(0.282a;a) (0,1) 193.259 P6_3'/m'cm' 3 6 comp2
K3(a;b) - 188.218 P-6'c2' 3 12 free (2 parameters)
K5(0,0;0.628a,a) (0,0,1,0) 189.221 P-62m 3 12 z_3
K5(0,0;a,0.529a) (1,0,0,0) 189.225 P-62'm' 3 12 z
K5(0.628a,a;0.628a,a) (0,0,0.707,-0.707) 63.457 Cmcm 3 18 dir1
K5(a,0.529a;a,0.529a) (0.707,-0.707,0,0) 63.459 Cm'cm 3 18 dir2
K5(0.529a,-a;-0.529a,a) (0,0,0.707,0.707) 63.460 Cmc'm 3 18 dir3
K5(a,-0.628a;-a,0.628a) (0.707,0.707,0,0) 63.462 Cm'c'm 3 18 aa
K5(0,0;a,b) - 174.133 P-6 3 24 free (2 parameters)
...
K5(a,b;c,d) - 6.18 Pm 3 72 free (4 parameters)
...
irrep direction: ISOTROPY's label, as crystod-mag --parent 194 --qpoint K lists it
in comp basis: spin field = sum_k c_k (field of the k-th basis component, as listed), unit length
size: cell of the translations without time reversal, in primitive cells; index: number of domains
dir<k>: a one-parameter direction that no other component lies on, exported as a component of its own
...
* K3(1) [AFM (q != 0)] *
component comp1:
Mn1: S = [ 1.0000, 0.0000, 0.0000]
Mn2: S = [-1.0000, 0.0000, -0.0000]
net moment over the magnetic supercell: [0.0, 0.0, 0.0]
order-parameter direction: K3(a;0.974a), magnetic isotropy subgroup 193.255 P6_3/m'cm
magnetic space group: P6_3/m'cm (BNS 193.255, type III), 72 operations, 36 with time reversal
...
component comp2:
Mn1: S = [ 0.0000, 1.0000, 0.0000]
Mn2: S = [-0.0000, 1.0000, -0.0000]
net moment over the magnetic supercell: [-0.0, -0.0, 0.0]
order-parameter direction: K3(0.282a;a), magnetic isotropy subgroup 193.259 P6_3'/m'cm'
magnetic space group: P6_3'/m'cm' (BNS 193.259, type III), 72 operations, 36 with time reversal
...
How to read the output.
comp1of K3 is the arrangement of before, the direction K3(a;0.974a) ofcrystod-mag --parent 194 --qpoint Kwith the group P6_3/m’cm; its partnercomp2, the same pattern with the phase turned by 90 degrees, is K3(0.282a;a), P6_3’/m’cm’, which the export lacked before. Any other phase is the generic direction K3(a;b), P-6’c2’ (index 12).irrep direction: ISOTROPY’s label of the stratum, as--parentprints it (matched by the BNS number of the spin field, the index, the free parameters and the size). At a q without a table (not a special point) the column holds the irrep and the coefficients,SM4 (-0.601,-0.799).in comp basis: the coefficients c_k of a one-parameter direction in the basis components of its basis, in the order of the listing: the spin field is sum_k c_k times the field of component k as listed (largest moment 1), so aMAGMOMline of any direction is the same combination of theMAGMOMlines of the components.subgroup (BNS): the magnetic isotropy subgroup, the magnetic space group of the spin field (of a generic field of the stratum for two or more free parameters).size: the cell of the translations without time reversal, in primitive cells;index: [G1’ : H], the number of domains in the whole paramagnetic group.component: the exported component that lies on the direction (comp1,z_3, the combinationaa, ordir<k>),free (n parameters)for a direction with two or more free parameters, andnot in this magnetic cellfor a row of the table whose arms are not all in the cell (the multi-arm directions of a star such as X of a cubic crystal in its 1x2x1 cell).dir<k>: a one-parameter direction that neither a component nor a combination lies on, exported as a component of its own (K5 and K6: the four Cmcm-type directions, one of them already the combinationaa). Its VESTA file isPOSCAR_YMnO3_spin_K5_K_dir1.vesta, and it has its ownMAGMOMline, magnetic space group and class.Every component block names the direction it lies on in the line
order-parameter direction:(a component in a stratum of two or more free parameters says so:comp1of SM4 at--qpoint 1/3 0 0lies in the two-parameter stratum Am’m’2).
The phase of the K3 order parameter is the angle between the two patterns of
the ferroelectric P6_3cm phase: mapped onto the six Mn of the P6_3cm cell
(185_PPOSCAR_YMnO3, a √3×√3 cell of the parent), comp1 is
0.25 GM1 + 0.75 GM4 and comp2 is 0.75 GM1 + 0.25 GM4 (squared overlaps with
the GM1 64pole and GM4 octupole bases of
crystod-mag -c 185_PPOSCAR_YMnO3 --element Mn --qpoint GM), 60 and 30
degrees from GM1.
Time-reversal-invariant q and merged pairs
At a time-reversal-invariant q (R, X, M of a cubic crystal) the bases are real
and have their d components as before; the block lists their directions as
well, matched to the components and combinations: for AlNi3 at R, R4+(0,0,a)
is comp1 (I_c4/mcm), R4+(0,a,a) aa0 (I_bmma) and R4+(a,a,a) aaa
(R_I-3c), so no dir<k> is needed. A merged complex pair (A5A6(4) of ZnO
at A) has only directions with two or more free parameters, all free. A
mirror-image representative is matched to its table row with a note
(Z3: 96.148 P_c4_32_12 is the enantiomorphic partner of the table's 92.116
for TiO2 at Z).
Repeated irreps and other q points
At a q that is not time-reversal invariant, the copies of an irrep that
occurs more than once are chosen so that the operations taking q to -q keep
each copy (spgrep’s copies are mixed by them): for Mn of YMnO3 at
q = (1/3,0,0) (--qpoint 1/3 0 0, the line SM, no table) each of SM2, SM3
and SM4 occurs twice, and each copy has its two one-parameter directions
(Cm’cm, Cm’c’m, …), exported as dir1, dir2 and named by their
coefficients. Where no such choice exists (an irrep paired with another one),
the directions of the joint space are listed under both labels
(SM4(1) + SM4(1)); a basis whose fields do not span a space the operations
keep gets a NOTE in the block instead.
Python API
from phonopy.interface.vasp import read_vasp
from crystod import mag
result = mag.analyze_spin_basis(read_vasp("194_PPOSCAR_YMnO3"), "Mn", "K")
for d in result.directions:
print(d.label, d.direction_label, d.subgroup, d.size, d.index, d.n_free, d.component)
# K2(1) K2(0.282a;a) 193.258 P6_3'/m'c'm 3 6 1 comp1 ...
k3 = [c for c in result.components if c.label.startswith("K3")]
[(c.tag, c.direction_label, c.direction_subgroup) for c in k3]
# [('comp1', 'K3(a;0.974a)', "193.255 P6_3/m'cm"), ('comp2', 'K3(0.282a;a)', "193.259 P6_3'/m'cm'")]
print("\n".join(mag.format_direction_table(result)))
OrderParameterDirection carries entry_index, label, direction_label,
bns_number, bns_symbol (subgroup joins them), size, index,
n_free, component, coefficients (in the basis components, unit length)
and realized (False for a table row the magnetic cell does not realize);
SpinBasisComponent.direction is the direction a component lies in.
05-07. Spin bases aligned with a subgroup structure (--subgroup-poscar)#
Example directory: example/05-06_mag_directions (testsuite section 05-07); after pip install: crystod-example 05-06
Which of the magnetic orders of a q point off Γ does a distorted phase of the
same compound pick, and which magnetic space group does it then have? When the
structural distortion and the magnetic order share their irrep, the answer
depends on the angle between the two order parameters: YMnO3 (P6_3/mmc) turns
ferroelectric P6_3cm through the structural K3 distortion, and the K3 spin
order of Mn gives P6_3cm on that structure when it is parallel to the
structural domain (GM1 of P6_3cm), P6_3’cm’ at 90 degrees to it (GM4), and
P3c1 at any other angle. --subgroup-poscar CHILD places the bases at q ≠ 0
on the distorted structure and, for every order-parameter direction, finds the
domain whose spin field is one irrep of CHILD at GM.
Input |
Command |
Output |
|---|---|---|
primitive POSCARs of the parent and of a distorted phase whose cell is a magnetic cell of q, magnetic element, q ≠ 0 |
|
terminal: everything of |
the same, tables only |
add |
the three blocks, no listing and no files |
cd example/05-06_mag_directions
crystod-mag -c 194_PPOSCAR_YMnO3 --element Mn --qpoint K --subgroup-poscar 185_PPOSCAR_YMnO3
* Subgroup structure *
Subgroup structure: 185_PPOSCAR_YMnO3, P6_3cm (No. 185)
Magnetic sites: Mn x 6 (3 primitive cells of P6_3/mmc)
q = K (-1/3, 2/3, 0) folds onto GM of P6_3cm; the fields of each basis are made of:
K2(1) -> GM2 [A2] + GM3 [B2]
K3(1) -> GM1 [A1] + GM4 [B1]
K5(2) -> GM5 [E2] + GM6 [E1]
K6(2) -> GM5 [E2] + GM6 [E1]
* Transformation matrix (parent -> child) *
P6_3/mmc (No. 194) -> P6_3cm (No. 185), the cells of the input files
[ -2 1 0 ] [ 0 ]
[ -1 -1 0 ] [ 0 ]
[ 0 0 1 ] [ 0.2483 ]
...
(spins through the fractional components, S_parent = P S_child)
* Spin bases on the subgroup structure *
irrep direction field child irrep MSG on the child (BNS) angle in comp basis
K2(0.282a;a) comp1 0.75 GM2 + 0.25 GM3 158.59 P3c'1 - (1,0)
K2(0.282a;a) comp1_GM3 GM3 185.199 P6_3'c'm 60 (0.5,0.866)
K2(a;0.974a) comp2 0.25 GM2 + 0.75 GM3 158.59 P3c'1 - (0,1)
K2(a;0.974a) comp2_GM2 GM2 185.201 P6_3c'm' 60 (-0.866,0.5)
K2(a;b) generic GM2 + GM3 158.59 P3c'1 - -
K3(a;0.974a) comp1 0.25 GM1 + 0.75 GM4 158.57 P3c1 - (1,0)
K3(a;0.974a) comp1_GM1 GM1 185.197 P6_3cm 60 (0.5,0.866)
K3(0.282a;a) comp2 0.75 GM1 + 0.25 GM4 158.57 P3c1 - (0,1)
K3(0.282a;a) comp2_GM4 GM4 185.200 P6_3'cm' 60 (-0.866,0.5)
K3(a;b) generic GM1 + GM4 158.57 P3c1 - -
K5(0,0;0.628a,a) z_3 0.5 GM5 + 0.5 GM6 8.32 Cm - (0,0,1,0)
K5(0,0;0.628a,a) z_4 0.5 GM5 + 0.5 GM6 8.32 Cm - (0,0,0,1)
K5(0,0;0.628a,a) - no single irrep - - -
...
K5(0.628a,a;0.628a,a) dir1 0.25 GM5 + 0.75 GM6 9.37 Cc - (0,0,0.707,-0.707)
K5(0.628a,a;0.628a,a) dir1_GM5 GM5 36.172 Cmc2_1 60 (-0.632,0.632,0.316,-0.316)
...
K5(a,b;c,d) generic GM5 + GM6 1.1 P1 - -
...
* K3(1) [AFM (q != 0)] on 185_PPOSCAR_YMnO3 *
component comp1_GM1 (K3(a;0.974a), GM1 [A1] of P6_3cm; 60 degrees from comp1):
Mn1: S = [-0.8660, 0.5000, 0.0000]
...
magnetic space group: P6_3cm (BNS 185.197, type I), 12 operations, 0 with time reversal
...
MAGMOM = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -0.8660 0.5000 0.0000 ...
component comp2_GM4 (K3(0.282a;a), GM4 [B1] of P6_3cm; 60 degrees from comp2):
...
magnetic space group: P6_3'cm' (BNS 185.200, type III), 12 operations, 6 with time reversal
...
How to read the output.
* Subgroup structure *: CHILD, its space group, and the irreps of CHILD at GM that the fields of every basis are made of: the folding of q onto GM, the rows ofcrystod-group --correlate --supergroup-poscar -c --subgroup-poscar CHILDfor q (2 x K3 -> GM1 + GM4).* Transformation matrix (parent -> child) *: the cell relation ofcrystod-group --correlate, x_parent = P x_child + p between the fractional coordinates of the two input files; each magnetic atom of CHILD takes the spin of the site and cell of the magnetic cell it lands on, carried through its fractional components (S_parent = P S_child, so a lattice that is strained against the parent keeps the symmetry relation).* Spin bases on the subgroup structure *, one row per field, grouped by the directions of* Order-parameter directions *:a component as exported (
comp1,z_3,aa,dir1): the irreps of CHILD it is made of, with the squared norms of their parts on CHILD (0.25 GM1 + 0.75 GM4), and its magnetic space group on CHILD (spglib on the cell of CHILD with these moments);<component>_<irrep>: the domain of the direction (an image under the operations of the grey group G1’, so the same magnetic isotropy subgroup type) whose field is that one irrep of CHILD, named after the nearest exported component on the direction;angleis the angle between the two in the real order-parameter space (the phase of the order parameter at a q that is not time-reversal invariant) andin comp basisits coefficients as in* Order-parameter directions *. Domains of one direction that give different groups on CHILD get one row each;-withno single irrep: no domain of the direction is one irrep of CHILD (the P-62m-type directions of K5 and K6 always mix GM5 and GM6);generic: a generic field of the direction inside one irrep of CHILD (GM5: P2_1), and of the whole basis when it folds onto more than one irrep (K3(a;b): P3c1).
The identity irrep of CHILD (GM1) keeps every operation of the distorted structure: for an irrep of its distortion, the direction of the structural order parameter (0 degrees), and the other irreps of the basis are orthogonal to it (GM4: 90 degrees). For YMnO3, the K3(a;0.974a)-type direction aligned with the structural domain is GM1, P6_3cm (185.197, type I); the K3(0.282a;a)-type direction at 90 degrees to it is GM4, P6_3’cm’ (185.200); the exported
comp1andcomp2lie 60 degrees from these domains, mix GM1 and GM4 and give P3c1 (158.57), as every phase but the two aligned ones does.* K3(1) [AFM (q != 0)] on 185_PPOSCAR_YMnO3 *: every aligned field with the spins of the magnetic sites of CHILD (Mn1..Mn6in its atom order), its magnetic groups, the listing of all atoms and theMAGMOMline (or the Quantum ESPRESSO input with--format qe) for the cell of CHILD; the VESTA filePOSCAR_YMnO3_spin_K3_K_comp2_GM4.vestaand the POSCARS185_PPOSCAR_YMnO3_K(the cell of CHILD as given, in the atom order of theMAGMOMlines) go with it, next to the files of the parent cell.
How the alignment works
The cell relation comes from the structure mapping of crystod-group --correlate --supergroup-poscar (SymmetryModeAnalysis), expressed between
the cells of the two input files. For every basis, the real order-parameter
space V and the action of G1’ on it are those of * Order-parameter directions *. The fields of V placed on CHILD are decomposed with the
projection operators of the irreps of CHILD at GM (its spin representation on
the magnetic sites, ISO-IR labels in the frame of CHILD, a complex pair
merged): U_Γ is the subspace of V whose fields are of irrep Γ alone. For every
one-parameter direction, its domains (the distinct images of its line under
G1’) that lie in some U_Γ are the aligned fields; one row is kept per irrep
and magnetic space group on CHILD, the domain nearest to an exported
component of the direction. Two checks in section 05-07 are independent of this
construction: every aligned field lies in the span of the bases that
crystod-mag -c CHILD --qpoint GM finds for its irrep, and in the magnetic
cell of the parent its magnetic space group is the magnetic isotropy subgroup
of its direction (it is a domain of that direction).
At a time-reversal-invariant q the same holds without the phase: for AlNi3
(Pm-3m) at R with a two-cell child whose Al atoms carry an R-point
displacement (I4/mmm), R4+ folds onto GM1- + GM5-, the exported R4+ comp2
is already GM1- (comp2_GM1-, 0 degrees, I4/m’m’m’), and the (a,a,a)
direction has no domain in one irrep.
Options and requirements
q ≠ 0, and the cell of CHILD must be a magnetic cell of q: q . a’, q . b’, q . c’ integers for its lattice vectors (q folds onto GM of CHILD). At GM, run
crystod-magon CHILD itself; a q that does not fold (M of YMnO3 on the √3×√3 cell) stops with an ERROR that gives q . a’, q . b’, q . c’.CHILD is a primitive cell (POSCAR or CIF) of the same compound, with
sizetimes the magnetic atoms of the parent,sizethe number of primitive cells of the parent in it; a conventional cell or a supercell is refused (the PPOSCAR ofphonopy --symmetryis a primitive cell).The mapping and the groups on CHILD use the tolerance of
crystod-group --correlate(0.01 Å), or--tolerancewhen that is larger; a magnetic atom of CHILD more than 1.8 Å from its site stops the run.--classify-onlyprints the three blocks and writes no files;--format qegives the Quantum ESPRESSO lines of the aligned fields;--conventionalchanges only the files of the parent cell.When an operation of CHILD takes q to another arm of its star (a child of a multi-arm order), the fields of one basis are not all sums of fields of single irreps of CHILD; a NOTE says so and only the directions inside one irrep are aligned.
Python API
from phonopy.interface.vasp import read_vasp
from crystod import mag
result = mag.analyze_spin_basis(read_vasp("194_PPOSCAR_YMnO3"), "Mn", "K",
subgroup_poscar="185_PPOSCAR_YMnO3")
alignment = result.subgroup # mag.SubgroupAlignment
alignment.folding
# [('K2(1)', ['GM2', 'GM3']), ('K3(1)', ['GM1', 'GM4']), ...]
for row in alignment.rows: # mag.SubgroupSpinField
if row.kind == "aligned" and row.label.startswith("K3"):
print(row.tag, row.direction.direction_label, row.child_irrep_text,
row.subgroup, round(row.angle, 1))
# comp1_GM1 K3(a;0.974a) GM1 185.197 P6_3cm 60.0
# comp2_GM4 K3(0.282a;a) GM4 185.200 P6_3'cm' 60.0
print("\n".join(mag.format_subgroup_alignment(result)))
SubgroupAlignment carries child_file, child_symbol, child_number,
transformation (P and p), size, the cell of CHILD (lattice, symbols,
positions), site_atoms and site_sources (the site and cell of the
parent each magnetic atom takes its spin from), spin_matrix, folding,
identity_irrep, mulliken, rows and notes; child_spins(field) places
any field of the magnetic cell on CHILD. A SubgroupSpinField has kind
("component", "aligned", "generic", "none"), direction, tag,
child_irreps (label and weight), reference, group (on CHILD), angle,
coefficients, field (in the magnetic cell of the parent), spins (on the
atoms of CHILD) and vesta_name; mag.subgroup_table_rows gives the rows of
the table.
05-08. Phonon angular momentum near GM (--phonon-angular-momentum)#
Example directory: example/05-08_phonon_angular_momentum (testsuite section 05-08); after pip install: crystod-example 05-08
Can the phonons of a magnetic arrangement carry angular momentum, and how
does it vary with the wavevector near Γ? The angular momentum l(q) of a
phonon branch is an axial vector, odd under time reversal, and the magnetic
point group fixes the lowest-order polynomial in q that each component can
follow: s-wave (a constant, as for a net magnetization), p-wave (linear in
q), d-wave, f-wave and so on up to j-wave (degree 7), together with the nodal
planes through Γ on which a component must vanish. --phonon-angular-momentum
adds this to every component of crystod-mag --qpoint, and with
--classify-only gives one table that starts with the paramagnetic crystal
itself, so the option is useful for non-magnetic crystals as well.
See also
Theory: Phonon angular momentum near Gamma — the constraint l(A q) = B l(q), the polynomial reading, why the lowest degree and the number of enforced nodal planes differ, P T and parity, what symmetry does not tell, and the relation to the altermagnet wave of the spin splitting.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, magnetic element, q point |
|
terminal: in every component block, after |
the same, table only |
add |
the block |
cd example/05-08_phonon_angular_momentum
crystod-mag -c 136_PPOSCAR_TiO2 --element Ti --qpoint GM --classify-only --phonon-angular-momentum
* Phonon angular momentum allowed by symmetry *
basis component MPG l_x l_x ~ planes l_y l_y ~ planes l_z l_z ~ planes
paramagnetic - 4/mmm1' - none (P T) - - none (P T) - - none (P T) -
GM3+(1) [FM, dipole] z 4/mm'm' d q_x q_z 2 d q_y q_z 2 s const 0
GM5+(2) [FM, dipole] 1m10 mm'm' s const 0 s const 0 d q_z (q_x - q_y) 2
GM5+(2) [FM, dipole] 110 mm'm' s const 0 s const 0 d q_z (q_x + q_y) 2
GM5+(2) [FM, dipole] x mm'm' s const 0 d q_x q_y 2 d q_x q_z 2
GM5+(2) [AFM, octupole] 1m10 mm'm' s const 0 s const 0 d q_z (q_x - q_y) 2
GM5+(2) [AFM, octupole] 110 mm'm' s const 0 s const 0 d q_z (q_x + q_y) 2
GM5+(2) [AFM, octupole] x mm'm' d q_x q_y 2 s const 0 d q_y q_z 2
GM2+(1) [AFM, octupole] z 4'/mmm' d q_y q_z 2 d q_x q_z 2 d q_x q_y 2
l_i: component i of the phonon angular momentum l(q) of a branch near GM (an axial vector, odd under time reversal), from l(A q) = B l(q) with A = s R, B = s det(R) R (s = -1 with time reversal);
its column: the wave of the lowest degree n of a q-polynomial allowed for it (s, p, d, f, g, h, i, j for n = 0..7, the number beyond); l_i ~: the polynomials of that degree (a second one on the next line);
planes: only the nodal planes through GM pinned by a single operation (a mirror without time reversal containing e_i, a 2' along the normal when e_i is along it); fewer than n: the other nodes are not fixed in position by symmetry;
none (P T): P T makes l vanish at every q; s: a constant, the net magnetization.
Symmetry fixes the forms, not the magnitudes. paramagnetic: the grey group of the crystal, what ordinary (time-reversal-even) force constants give (odd degrees only, l(-q) = -l(q));
the even-degree textures of a magnet, and every form beyond the paramagnetic row, need a time-reversal-odd term in the lattice dynamics (e.g. spin-phonon coupling).
MPG as in the table above; q and l in the Cartesian axes of the input cell, the components read one at a time.
...
Reference (phonon angular momentum): L. Zhang and Q. Niu, Phys. Rev. Lett. 112, 085503 (2014).
How to read the output.
paramagnetic: the grey group of the crystal without spins (4/mmm1'), the symmetry of phonons computed from ordinary force constants. Rutile is centrosymmetric, so P T is in it and no component survives: the phonon angular momentum of rutile needs the magnetic order and a time-reversal-odd term in the lattice dynamics.l_x,l_y,l_z: the wave letter of the lowest allowed degree;l_i ~: the polynomials of that degree (Cartesian axes of the input cell; the letters s, p, d, f, g, h, i, j for the degrees 0 to 7 are the convention used here);planes: only the nodal planes through Γ pinned by a single operation, which can be fewer than the degree (the other nodes are then not fixed in position by symmetry). The altermagnetic GM2+ arrangement (moments along z,4'/mmm') allows the d-wavel_z ~ q_x q_ywith two planes, the same pattern as its spin splittingk_x k_y.s/const: a constant component, allowed exactly where the group allows a net magnetization along that axis (GM3+, the ferromagnet along z, hasl_zs-wave; the GM5+ octupole along x allows a magnetization along y and hasl_ys-wave).The number of planes can be smaller than the degree: the group
4'givesl_z ~ q_zwith no plane,4'/mthe d-wave pairq_x^2 - q_y^2,q_x q_ywith no plane.
Without --classify-only the same result is one line per component block:
crystod-mag -c 136_PPOSCAR_TiO2 --element Ti --qpoint GM --phonon-angular-momentum
* GM2+(1) [AFM, octupole] *
component z:
Ti1: S = [ 0.0000, 0.0000, 1.0000]
Ti2: S = [ 0.0000, 0.0000, -1.0000]
net moment: [0.0, 0.0, 0.0]
magnetic space group: P4_2'/mnm' (BNS 136.499, type III), 16 operations, 8 with time reversal
magnetic point group: 4'/mmm'
UNI symbol: P4_2'/mnm'; point group 4'/mmm'
allowed by symmetry: net magnetization none; electric polarization none; linear magnetoelectric effect none; piezomagnetism 2 components
collinear order: altermagnet (d-wave; spin splitting allowed by symmetry ~ k_x k_y; spin-degenerate planes perpendicular to [100], [010])
phonon angular momentum: l_x ~ q_y q_z (d-wave; zero on the planes perpendicular to [010], [001]); l_y ~ q_x q_z (d-wave; zero on the planes perpendicular to [100], [001]); l_z ~ q_x q_y (d-wave; zero on the planes perpendicular to [100], [010])
...
The three components are separated by ; ; a component whose polynomials of
the lowest degree span two dimensions lists both (l_z ~ q_x^2 - q_y^2, q_x q_y (d-wave)), and a vanishing one reads l_x = 0 (P T).
A noncentrosymmetric crystal: YMnO3 (P6_3cm)
The paramagnetic row is not empty when the crystal lacks inversion: the grey
group 6mm1' of ferroelectric YMnO3 allows p-wave in-plane components and
the j-wave (degree 7) l_z with seven planes, already with ordinary force
constants. The magnetic ground state GM4 (P6_3'cm', 6'mm') lowers l_z
to g-wave; a two-dimensional polynomial space continues on the next line.
crystod-mag -c 185_PPOSCAR_YMnO3 --element Mn --qpoint GM --classify-only --phonon-angular-momentum
* Phonon angular momentum allowed by symmetry *
basis component MPG l_x l_x ~ planes l_y l_y ~ planes l_z l_z ~ planes
paramagnetic - 6mm1' p q_y 1 p q_x 1 j q_x q_y q_z (3 q_x^4 - 10 q_x^2 q_y^2 + 3 q_y^4) 7
GM1(1) [AFM, 64pole] comp1 6mm p q_y 1 p q_x 1 i q_x q_y (3 q_x^4 - 10 q_x^2 q_y^2 + 3 q_y^4) 6
GM2(1) [FM, dipole] z 6m'm' p q_y 0 p q_x 0 s const 0
GM2(1) [AFM, quadrupole] comp1 6m'm' p q_y 0 p q_x 0 s const 0
GM4(1) [AFM, octupole] comp1 6'mm' p q_y 0 p q_x 1 g q_x q_z (q_x^2 - 3 q_y^2) 4
GM4(1) [AFM, hexadecapole] z 6'mm' p q_y 0 p q_x 1 g q_x q_z (q_x^2 - 3 q_y^2) 4
GM3(1) [AFM, octupole] comp1 6'm'm p q_y 1 p q_x 0 g q_y q_z (3 q_x^2 - q_y^2) 4
GM5(2) [AFM, quadrupole] comp1 m'm'2 p q_x 0 p q_x 0 s const 0
q_y q_y
GM5(2) [AFM, quadrupole] comp2 mm2 p q_x 0 p q_x 0 d 3 q_x^2 - 2sqrt3 q_x q_y - 3 q_y^2 2
q_y q_y
...
The in-plane components of a hexagonal group are read one at a time in the
Cartesian axes (the operations mix them), so l_x and l_y of one row can
have different numbers of planes. The mm2 of GM5 comp2 has its mirrors at
30 degrees from the Cartesian axes, hence the sqrt3 in its d-wave form.
Details: what is computed, layers, limits
For each degree n = 0, 1, …, 9 the homogeneous polynomial vector fields F(q) with F(A q) = B F(q) for every operation (A = s R, B = s det(R) R, s = -1 with time reversal) are found as a null space; the wave of l_i is the lowest n at which a field has a nonzero i-th component. A degree beyond j-wave is printed as its number (
8), and a component with no form up to degree 9 asnone up to degree 9(it does not happen in the groups tested).The enforced nodal planes of l_i come from single operations: a mirror without time reversal whose plane contains e_i, or a 2’ along the plane normal when e_i is along it.
The magnetic point group is that of the exported component (spglib on the magnetic cell, the operations in the Cartesian axes of the input cell); a group spglib does not name is still used through its operations (MPG
-).Layers (two-dimensional systems):
crystod.mag.phonon_angular_momentum(..., plane_normal=[0, 0, 1])restricts q to the plane; the nodes are then lines, and the group must map the normal to plus or minus itself. There is no command-line option for layers yet.Not printed with
--subgroup-poscarfor the aligned fields on the child, and not in the survey (it needs--qpoint);--parentrejects the option.
Python API
import numpy as np
from phonopy.interface.vasp import read_vasp
from crystod import mag
atoms = read_vasp("136_PPOSCAR_TiO2")
magmoms = np.zeros((len(atoms.numbers), 3))
magmoms[atoms.numbers == 22] = [[0, 0, 1], [0, 0, -1]]
info = mag.magnetic_space_group(atoms.cell, atoms.scaled_positions, atoms.numbers, magmoms)
result = mag.phonon_angular_momentum(info) # mag.PhononAngularMomentum
lz = result.component("z") # mag.AngularMomentumComponent
lz.degree, lz.letter, lz.forms, lz.nodal
# (2, 'd', ('q_x q_y',), ('[100]', '[010]'))
c4 = np.array([[0, -1, 0], [1, 0, 0], [0, 0, 1]], dtype=float)
group_4p = [np.eye(3), c4, c4 @ c4, c4.T] # the group 4'
flags = [False, True, False, True]
mag.phonon_angular_momentum(group_4p, flags).component("z").text()
# 'l_z ~ q_z (p-wave)'
mag.phonon_angular_momentum(group_4p, flags, plane_normal=[0, 0, 1]).component("z").text()
# 'l_z ~ q_x^2 - q_y^2, q_x q_y (d-wave)'
mag.phonon_angular_momentum_degrees(group_4p, flags, component="z")
# (1, 2, 3, 4, 5, 6, 7, 8, 9)
analysis = mag.analyze_spin_basis(atoms, "Ti", "GM")
print("\n".join(mag.format_phonon_angular_momentum(analysis)))
AngularMomentumComponent carries axis, degree, forms, nodal (plane
normals, or line directions for a layer), zero_reason ("P T", or for a
layer the mirror of the layer or a 2’ along its normal), two_dimensional
and max_degree, with letter, short_name, n_nodal, nodal_text() and
text(); PhononAngularMomentum has components, pt_symmetric,
plane_normal, in_plane and text(). phonon_angular_momentum_degrees
lists every allowed degree (the N-th order angular-momentum current response
to a temperature gradient needs degree N + 1).
mag.phonon_angular_momentum_rows and mag.paramagnetic_point_group give the
rows of the block.
05-09. Several magnetic elements (--element A B, --superpose)#
Example directory: example/05-09_multi_element (testsuite section 05-09); after pip install: crystod-example 05-09
Which magnetic arrangements does each irrep give when the spins sit on two
different elements (two Wyckoff orbits)? The spin space is then the direct sum
of the spin spaces of the elements, and an irrep can occur on both. --element Al Ni analyses every element as its single-element run does, prints the
decomposition per element and in total, checks that the copies of one irrep on
different elements transform with the same irrep matrices, and lists the
arrangements with one copy per element along every high-symmetry direction of
the listing ((a,0,0), (a,a,0), (a,a,a), … at Γ, the strata of section 05-06 at
q ≠ 0) with the relative signs (+,+) and (+,-). --superpose adds any listed
bases, of any irreps and elements, with given coefficients, the way
crystod-phonon --modulation --mode ... --amplitude ... adds modes.
See also
Theory: Spin bases on several elements — the direct sum over the orbits, the copies of one irrep, why a superposition along one direction keeps the group of each copy (the table of the single-element runs is complete), and what symmetry leaves free.
Input |
Command |
Output |
|---|---|---|
primitive POSCAR, two or more elements, q point |
|
terminal: the decomposition per element and in total, |
the same, table only |
add |
the blocks and the table, no listing and no files but the report |
listed bases and coefficients |
add |
|
two or more elements, no q point |
|
the survey: the decomposition of every element and of the total at every special point; the report |
cd example/05-09_multi_element
crystod-mag -c 221_PPOSCAR_AlNi3 --element Al Ni --qpoint GM --classify-only
* Spin (axial-vector) representation *
Spin (axial-vector) representation on Al, Ni sites: 12 dimensions
Decomposition (Al, 3 dimensions): GM4+(3)
Decomposition (Ni, 9 dimensions): 2 x GM4+(3) + GM5+(3)
Decomposition (total): 3 x GM4+(3) + GM5+(3)
* Symmetry-adapted spin bases *
Al GM4+(3): dim 3 [FM, dipole]
Ni GM4+(3): dim 3 [FM, dipole]
Ni GM4+(3): dim 3 [AFM, octupole]
Ni GM5+(3): dim 3 [AFM, octupole]
* Copies of one irrep on several elements *
GM4+(3): reference Al GM4+(3) [FM, dipole]
Ni GM4+(3) [FM, dipole]: same matrices on the rows
Ni GM4+(3) [AFM, octupole]: rows mapped by (-x, y, z)
Every copy of one irrep transforms with the same irrep matrices (on its rows, or through the map given), so a real superposition of copies along the same order-parameter direction, with any nonzero amplitudes, has the magnetic space group of each copy alone: the table irrep x direction -> magnetic space group is complete; the relative amplitudes of the copies are free parameters that symmetry does not fix.
20 combined arrangements below: one copy per element along every direction of the first element's copy, each copy with largest moment 1, with the relative signs (+,+), (+,-), ... of the printed moments (the sign of a basis vector is a convention).
* Magnetic groups and collinear classes *
no. element basis component MSG (BNS) type MPG net magnetization class MSG (UNI)
1 Al GM4+(3) [FM, dipole] z P4/mm'm' (123.345) III 4/mm'm' along z ferromagnet P4/mm'm'
...
16 Ni GM5+(3) [AFM, octupole] z P4'/mm'm (123.342) III 4'/mm'm none altermagnet (d-wave) P4'/mm'm
...
- Al+Ni GM4+(3) [Al FM, dipole + Ni FM, dipole] z (+,+) P4/mm'm' (123.345) III 4/mm'm' along z ferromagnet P4/mm'm'
- Al+Ni GM4+(3) [Al FM, dipole + Ni FM, dipole] z (+,-) P4/mm'm' (123.345) III 4/mm'm' along z ferrimagnet P4/mm'm'
...
- Al+Ni GM4+(3) [Al FM, dipole + Ni AFM, octupole] 111 (+,-) R-3m' (166.101) III -3m' along [111] noncollinear R-3m'
How to read the output.
Decomposition: per element and in total (the direct sum). The bases of every element are exactly those of its single-element run; their rows in the table are numbered (no.), the numbers--superposetakes.* Copies of one irrep on several elements *: for every irrep found on two or more elements, how the basis rows of each copy relate to those of the first copy.same matrices on the rows: the operations of the crystal act on both with the same matrices.rows mapped by (-x, y, z): the same irrep, but one row with the opposite sign convention (the Ni octupole copy of GM4+, whose exported111is therefore the domain [11-1]); the combined arrangements follow the direction through that map.The combined rows (
no.-, elementAl+Ni) take one copy per element along one direction of the first element’s copy (z,110,111, …;dir<k>and the ISOTROPY label at q ≠ 0), each copy with largest moment 1 and signed as its listed component;(+,+)and(+,-)are the relative signs of the moments as printed. Their magnetic space group is the one of each copy alone; only the class (ferromagnet, ferrimagnet, compensated) and the moments differ. AlNi3 gives 20 (GM4+ along 5 directions, Al with either Ni copy, two signs); GM5+ occurs on Ni only.
Two different elements are never exchanged by an operation of the crystal, so a pattern that looks like one antiferromagnet on two sites is a different magnetic group when the two sites carry different elements. In a Lieb lattice with Al and B on the edge sites (Pmmm), Al up and B down is Pm’m’m (47.252), the group of Al up and B up, compensated with l_z s-wave; with Al on both sites (P4/mmm) the same up-down pattern is the d-wave altermagnet P4’/mm’m (123.342) with l_z ~ q_x^2 - q_y^2:
crystod-mag -c POSCAR_Lieb_CuAlB --element Al B --qpoint GM --classify-only --phonon-angular-momentum
crystod-mag -c POSCAR_Lieb_CuAl2 --element Al --qpoint GM --classify-only --phonon-angular-momentum
- Al+B GM2+(1) [Al FM, dipole + B FM, dipole] z (+,+) Pm'm'm (47.252) III m'm'm along z ferromagnet Pm'm'm
- Al+B GM2+(1) [Al FM, dipole + B FM, dipole] z (+,-) Pm'm'm (47.252) III m'm'm along z compensated (unrelated sublattices) Pm'm'm
...
GM4+(1) [AFM, octupole] z P4'/mm'm (123.342) III 4'/mm'm none altermagnet (d-wave) P4'/mm'm
Superposition of listed bases (--superpose)
Any listed bases, of any irreps and elements, can be added with real coefficients; this is how a coupled (“hybrid”) arrangement of two irreps is built on a concrete structure. Every basis enters as listed (largest moment 1); the sum is scaled to largest moment 1.
crystod-mag -c 221_PPOSCAR_AlNi3 --element Al Ni --qpoint GM --classify-only --superpose 1 16 --coefficients 1 0.5 --no-files
* Superposition *
[1] Al GM4+(3) [FM, dipole] z: coefficient 1
[16] Ni GM5+(3) [AFM, octupole] z: coefficient 0.5
irrep content (squared weights): Al GM4+(3) [FM, dipole] 0.667 + Ni GM5+(3) [AFM, octupole] 0.333
Al1: S = [ 0.0000, 0.0000, 1.0000]
Ni1: S = [ 0.0000, 0.0000, 0.0000]
Ni2: S = [ 0.0000, 0.0000, 0.5000]
Ni3: S = [ 0.0000, 0.0000, -0.5000]
net moment over the magnetic cell: [0.0, 0.0, 1.0]
magnetic space group: Pm'm'm (BNS 47.252, type III), 8 operations, 4 with time reversal
magnetic point group: m'm'm
UNI symbol: Pm'm'm; point group m'm'm
allowed by symmetry: net magnetization along z; electric polarization none; linear magnetoelectric effect none; piezomagnetism 5 components
collinear order: ferrimagnet
irrep content: the squared projection of the sum on the space of every
basis (weights sum to 1). Without --classify-only the block is followed by
the MAGMOM line and the VESTA file is written. With one element the same
option works on the single-element listing; the numbers are then the rows of
its table of magnetic groups, counted from 1.
Details: what is combined, limits
The copies of one irrep are compared through the matrices of the operations of the crystal (with the translations of the magnetic cell) on their basis rows: the map T with D_copy(g) T = T D_first(g) closest to the identity is found as a null space; the identity means the same matrices. At a q that is not time-reversal invariant the maps between copies form a two-dimensional family (a relative phase), and the one closest to the identity is used; every member gives the same groups.
The combinations use the first copy of the first element as reference; with three elements there are 2^2 sign patterns, and the copies of every element are combined in all ways (AlNi3: Al with either Ni copy).
At most 48 combined arrangements get a VESTA file (all are in the table); a NOTE then points to
--superposefor the others.--subgroup-poscartakes one element. MCP access to several elements is not yet available.
Python API
from phonopy.interface.vasp import read_vasp
from crystod import mag
multi = mag.analyze_multi_element(read_vasp("221_PPOSCAR_AlNi3"), ["Al", "Ni"], "GM")
multi.decomposition # {'GM4+(3)': 3, 'GM5+(3)': 1}
[r.text() for r in multi.relations] # ['same matrices on the rows', 'rows mapped by (-x, y, z)']
a = multi.combined[1] # mag.CombinedArrangement
a.header, a.component_text, a.group.bns_number, a.collinear.short_name
# ("GM4+(3) [Al FM, dipole + Ni FM, dipole]", 'z (+,-)', '123.345', 'ferrimagnet')
s = mag.superpose_bases(multi, [1, 16], [1.0, 0.5]) # mag.Superposition
s.irrep_text()
print("\n".join(mag.format_multi_summary(multi)))
MultiElementAnalysis has elements, analyses (one SpinBasisAnalysis per
element), site_indices, decomposition, relations (CopyRelation:
label, reference, copy, map, identity, residual), combined,
numbered and notes; CombinedArrangement has label, copies, tag,
signs, magmoms (every atom of the magnetic cell), field, net_moment,
group, responses, collinear, direction_label and vesta_name.
superpose_bases also takes a single-element SpinBasisAnalysis, and with
vibrations= the SymmetryOnlyVibrations of the crystal it classifies the
sum.