water_isotherm_workchains

Implemented workchain

gcmc_restart

A simple RASPA loading workchain with the following features:

• blocks inaccessible pockets
• performs on longer RASPA GCMC run with initialization in the beginning and then multiple shorter GCMC runs. This allows for two things:
  • it is easier to retrieve good statistics
  • it is easier to compare with the GCMC/MD workchain

gcmc_md

A workchain that cycles between short MD trajectories and short GCMC runs with the intuition that in some cases collective dynamics from MD is needed to 'disturb' a configuration where GCMC has a hard time in inserting new particles.

gcmc_md_monitor_rdf (development branch)

In development.

Notes

• My development version of the RASPA plugin need to be used to retrieve statistics about the MC moves and the RDFs, you can install it with pip install git+https://github.com/kjappelbaum/aiida-raspa.git@develop (warning! this might case problems in your older workflows. You might consider creating a special enviornment)
• The settings are not optimized but rather used for a "prove of concept"
• Make sure to expand the unitcells before you use the workchain. The workchain also implements the expansion using the orthogonal widths, but it is not tested, especially, I do not know how RASPA deals with the charge loop in this case. You can use Daniele Ongari's manage_crystal to do this.
• If you do not want the RDF output to explode use 'RemoveAtomNumberCodeFromLabel': 'yes'. The RASPA manual states that the charges are still used correctly and our tests show that this is indeed the case

Usage

1. Read the notes
2. pip install the workflows with pip install git+https://github.com/kjappelbaum/water_isotherm_workchains
3. restart the daemon verdi daemon restart

to use the examples it might be easier to

1. git clone the repository
2. cd water_isotherm workchains & pip install .

Known issues

• The output out the workchain is comparatively large as we save all RDFs for all simulations this can lead to problems if you have limited memory and want to safe into the database (i.e. in a Virtual Quantum Mobile machine we had issues whereas we had no problems in a 'real' machine)

Settings for the study

The folder files_4_study contains the runscript, the structures with charges and the force field definitions.

Value	Setting
systems	al-fumarate, mil-160, uio-66, mil-125-nh2
probe radius / A	3.1589/2.
force field	UFF with all interactions
water model	TIP4P 2005
partial charge derivation method	DDEC / EqEq (for MIL-125-NH2)
number repeats	30
number initialization cycles	20 000
cycles first GCMC	5 000
cycles short GCMC	1 000
temperature GCMC / K	298.0
temperature MD / K	298.0
timestep MD / fs	0.0005
cycles MD	15 000
pressures / Pa	00.0001E5, 00.001E5, 00.002E5, 00.004E5, 00.006E5, 00.008E5, 00.011E5, 00.014E5, 00.016E5, 00.018E5, 00.021E5, 00.023E5, 00.026E5, 00.0298E5, 00.036E5, 00.04E5
cutoff / A	13
tail-correction	yes, since RASPA uses switching potential there is no problem in MD

In total this means we run 5 000 * 20 + 1 000 * 30 * 20 = 700 000 MC steps. And we run 30 * 0.0005 fs * 15 000 * 20 = 4.5 ps of MD trajectory.