Feature/constraints

[cpignedoli]

This is the first draft allowing for the constraint "distance"
It forces the key ion_dynamics 'fire' in the 'relax' calculation.
Still, I think we should warn/enforce no_symm.
Following this line, I could add other constraints (angle and dihedral); @mbercx what do you think?

[AndresOrtegaGuerrero]

I think some changes and check should be done in PwRelaxWorkChain (src/aiida_quantumespresso/workflows/pw/relax.py), since in principle the CONSTRAINTS are applied to the relaxation workflow, depending on the RelaxType (src/aiida_quantumespresso/common/types.py)

[mbercx]

Is this necessary? If so, I would validate this in validate_inputs. We can also set this there if it's more user-friendly, and raise a warning.

EDIT: Seems you can also set damp or verlet, and there may be a use case for this?

https://www.quantum-espresso.org/Doc/INPUT_PW.html#idm1620

[cpignedoli]

@mbercx you are right, I overlooked the damp option we should support it as well.
verlet is for md type of calculation, I did not have this in mind at teh moment, I was focusing on relax

[mbercx]

Could you show me an example from an input file?

[cpignedoli]

Hi @mbercx, here is an example I ran (it fails due to symmetries, but the constraint is parsed).
In the first line of the constraint, you have '1 1e-6': the total number of constraints and the tolerance (unfortunately, only a global tolerance seems to be allowed)

&CONTROL
  calculation = 'relax'
  etot_conv_thr =   9.0000000000d-04
  forc_conv_thr =   1.0000000000d-03
  max_seconds =   4.1040000000d+04
  nstep = 50
  outdir = './out/'
  prefix = 'aiida'
  pseudo_dir = './pseudo/'
  tprnfor = .true.
  tstress = .true.
  verbosity = 'high'
/
&SYSTEM
  degauss =   2.7500000000d-02
  ecutrho =   4.0000000000d+02
  ecutwfc =   5.0000000000d+01
  ibrav = 0
  nat = 9
  nosym = .false.
  ntyp = 2
  occupations = 'smearing'
  smearing = 'cold'
  tot_charge =   0.0000000000d+00
  vdw_corr = 'none'
/
&ELECTRONS
  conv_thr =   3.6000000000d-09
  electron_maxstep = 80
  mixing_beta =   4.0000000000d-01
/
&IONS
  ion_dynamics = 'fire'
/
ATOMIC_SPECIES
O      15.999 O.pbesol-n-kjpaw_psl.0.1.UPF
Si     28.085 Si.pbesol-n-rrkjus_psl.1.0.0.UPF
ATOMIC_POSITIONS angstrom
Si           1.1911527876       2.0631371475       0.0000000000 1 1 1
Si           2.6106944249       0.0000000000       3.6186666668 1 1 1
Si          -1.3053472125       2.2609276935       1.8093333333 1 1 1
O           -0.8256234071       3.6317748754       2.5091163607 1 1 1
O           -0.2358975994       1.7931665587       0.6997830273 1 1 1
O            1.0615210065       3.2231882479       4.3184496942 1 1 1
O            3.5580210066       1.1008765932       2.9188836393 1 1 1
O            1.6708765929       0.6922899656       4.7282169727 1 1 1
O            2.2606024006       2.5308982824       1.1095503059 1 1 1
K_POINTS automatic
5 5 4 0 0 0
CONSTRAINTS
1 1e-6
distance 1 5 1.65
CELL_PARAMETERS angstrom
      4.9930000000       0.0000000000       0.0000000000
     -2.4965000000       4.3240648411       0.0000000000
      0.0000000000       0.0000000000       5.4280000000

[mbercx]

I think some changes and check should be done in PwRelaxWorkChain (src/aiida_quantumespresso/workflows/pw/relax.py), since in principle the CONSTRAINTS are applied to the relaxation workflow, depending on the RelaxType (src/aiida_quantumespresso/common/types.py)

Can you elaborate on the changes/checks you think are required? I'm not very familiar with CONSTRAINTS, so I'm trying to understand.

[mbercx]

Thanks @cpignedoli! Would be a nice addition. I've left a few questions so far. Could you show me an example snippet of usage? Then I can test it more easily, thanks!

[mbercx]

Note: I'll be adapting our devops stack soon, including our formatter/linter, so that might lead to merge conflicts.