Hello everyone, I’m a newcomer to Abinit and just getting my feet wet with the code, though have a fair amount of experience with QE, VASP, and DFT in general. After converging the geometry of a few simple unit cells, I tried to test my understanding by converging a small yet “messy” unit cell with many degrees of freedom - monoclinic C2/m CuBr2. So far, whenever I allow unit cell geometry to change, there’s an error with the symmetry and I end up in space group P1. I am using the manual-suggested method of a two-step run that first optimizes internal degrees of freedom before allowing the cell to change. I suspect things might be due to some numerical problem with the choice of primitive unit cell vectors - I know that in other codes the software can be picky about exactly which ones are used. What’s strange is that I’m using the vectors suggested by running abistruct convert on a cif file, so I assumed that choice would be a good one.

I have tried:

- Increasing tolsym to 1d-4, but nothing much seemed to change
- Using a larger two formula unit cell with “nicer” lattice vectors and chkprim=0, but it seemed like Abinit wasn’t able to properly identify the space group
- When I ran the non-primitive cell without setting chkprim=0, it suggested a primitive unit cell with different lattice vectors than the ones given by abistruct. I tried those, and symmetry still broke in much the same way
- Using angdeg instead of setting rprim, but I got an unexpected error, “Number of bands nband= 18 > number of planewaves npw= 1”, which didn’t make sense to me since I had not changed ecut.

Things remain stable if I use optcell=1 and only allow volume changes to the cell, but that’s not a very satisfying solution. It’s also possible that the system is delicate and I might just need a higher ecut or kpoint mesh. I’ve included my .abi file below, which uses LDA and pseudopotentials from Pseudodojo (v4.1).

Thank you for any advice you are able to give! If you have any tips on how to generate the primitive cell from the .cif than abistruct (or what might be going wrong with angdeg) I am happy to hear them.

ndtset 2

natom 3

ntypat 2

typat 1 2 2

znucl 29 35

xred

0.0000000000 0.0000000000 0.0000000000

0.5056360000 0.5056360000 0.7612900000

0.4943640000 0.4943640000 0.2387100000

acell 1.0 1.0 1.0

rprim

7.8496356240 0.0000000000 -3.3648092894

5.0741829977 5.9891106465 -3.3648092894

0.0000000000 0.0000000000 13.0609347069

#pseudpotentials

pp_dirpath “.”

pseudos “Cu.psp8, Br.psp8”

ixc = 1

#Definition of the planewave basis set

ecut 20

#Definition of the k-point grid

kptopt 1

ngkpt 6 6 6

#nshiftk 1 # these are the defaults so we don’t need to include them

#shiftk 0.5 0.5 0.5

optcell1 0

optcell2 2

getxred2 -1

ionmov 2

ntime 10

dilatmx 1.05

ecutsm 0.5

tolmxf 1.0d-4

nstep 50

#Definition of the SCF procedure

toldfe 1.0d-6 #

diemac 12.0