This tutorial guides you through running a single-point calculation using the aiida-mlip package.
- Install aiida-mlip, aiida-core, janus-core.
- Set up an AiiDA profile, computer (localhost or remote) and a code (a predefined path for janus-core).
This code can be run in a verdi shell or as a python script through verdi run command.
To run a geometry optimisation using aiida-mlip you need to define some inputs as AiiDA data types, to then pass to the calculation.
To start, you will need a structure to optimise. Let's assume the structure is a cif file /path/to/structure.cif. The input structure in aiida-mlip needs to be saved as a StructureData type:
from aiida.orm import StructureData
structure = StructureData(ase=read("/path/to/structure.cif"))Then we need to choose a model and architecture to be used for the calculation and save it as ModelData type, a specific data type of this plugin. In this example we use MACE with a model that we download from this URI: "https://github.com/stfc/janus-core/raw/main/tests/models/mace_mp_small.model", and we save the file in the cache folder (default="~/.cache/mlips/"):
from aiida_mlip.data.model import ModelData
uri = "https://github.com/stfc/janus-core/raw/main/tests/models/mace_mp_small.model"
model = ModelData.from_uri(uri, architecture="mace", cache_dir="/.cache/")If we already have the model saved in some folder we can save it as:
model = ModelData.from_local("/path/to/model", architecture="mace")Another parameter that we need to define as AiiDA type is the code. Assuming the code is saved as janus in the localhost computer, the code info that are needed can be loaded as follow:
from aiida.orm import load_code
code = load_code("janus@localhost")The other inputs can be set up as AiiDA Str. There is a default for every input except the structure and code. This is a list of possible inputs:
from aiida.orm import Bool, Float, Str
inputs = {
"code": code,
"model": model,
"structure": structure,
"architecture": Str(model.architecture),
"precision": Str("float64"),
"device": Str("cpu"),
"max_force": Float(0.1), # Specific to geometry optimisation: convergence criteria
"opt_cell_lengths": Bool(False), # Specific to geometry optimisation
"opt_cell_fully": Bool(True), # Specific to geometry optimisation: to optimise the cell
"metadata": {"options": {"resources": {"num_machines": 1}}},
}It's worth noting that the architecture is already defined within the model, accessible through the architecture property in the ModelData. Even if not explicitly provided as input, it will be automatically retrieved from the model. The parameters that are not specific to geometry optimisation are the same for the single point calculation.
The calculation must be set:
from aiida.plugins import CalculationFactory
geomoptCalculation = CalculationFactory("mlip.opt")In this case, since we are running a geometry optimisation, the entry point for the calculation is mlip.opt. For a single point calculation, the entry point would be mlip.sp.
Finally, run the calculation:
from aiida.engine import run_get_node
result, node = run_get_node(geomoptCalculation, **inputs)results is a dictionary of the available results obtained from the calculation:
In : print(result)
Out :
{'log_output': <SinglefileData: uuid: 058e153b-f5fb-4799-9686-cc6dcc6f5fbb (pk: 1133)>,
'xyz_output': <SinglefileData: uuid: 2e8e2f74-39e9-4d3a-a492-02bfa979373b (pk: 1134)>,
'std_output': <SinglefileData: uuid: a72f2836-1d20-40f6-bcce-d1b56e6b1ba4 (pk: 1135)>,
'results_dict': <Dict: uuid: 99328f3d-e371-477b-857e-bcbf3353883a (pk: 1136)>,
'traj_file': <SinglefileData: uuid: 66886994-b856-42f6-abea-af54a8d0eaf8 (pk: 1137)>,
'traj_output': <TrajectoryData: uuid: b487c8b2-4aca-4c75-b20c-f5d92b625bda (pk: 1138)>,
'final_structure': <StructureData: uuid: 320b9165-2233-41bc-b14d-b44d8f7f72f3 (pk: 1139)>,
'remote_folder': <RemoteData: uuid: 4cf9f0cd-20b2-4a47-8dbd-46dbd410a558 (pk: 1131)>,
'retrieved': <FolderData: uuid: 5601957c-da54-4cd5-9e01-8a215e8ac4cf (pk: 1132)>}If more information are needed on specifi outputs they can be called like:
In : result["traj_output"].numsteps
Out : 3
In : result["final_structure"].cell #prints cell parameters of the optimised structure
[[4.0223130461422, -8.6767214011906e-17, 2.7878898106399e-16],
[2.0111565230711, 3.4834252799327, 2.1832573300987e-16],
[2.0111565230711, 1.1611417599776, 3.2842048495961]]Each data type has some properties that can be explored. In these examples traj_output contains info on the structures at every step of the optimisation(as a TrajectoryData), while final_structure contains info on the optimised structure (as a StructureData). The properties numsteps and cell are specific to the respective data types.
while node is the node of the calculation
In : type(node)
Out : aiida.orm.nodes.process.calculation.calcjob.CalcJobNode
In : print(node)
Out: uuid: 1d46ad08-2ea7-4892-9dd6-0240b9aeda8b (pk: 1130) (aiida.calculations:mlip.opt)The calculation can also be interacted with through verdi cli. Use verdi process list to show the list of calculations.
verdi process list -a
PK Created Process label Process State Process status
---- --------- --------------- --- ---------------- ----------------------------------
1130 1m ago GeomOpt Finished [0]verdi node show 1130
Property Value
----------- ------------------------------------
type GeomOpt
state Finished [0]
pk 1130
uuid 1d46ad08-2ea7-4892-9dd6-0240b9aeda8b
label
description
ctime 2024-03-19 13:29:58.202562+00:00
mtime 2024-03-19 13:30:19.461601+00:00
computer [2] localhost
Inputs PK Type
--------------- ---- -------------
architecture 1121 Str
code 2 InstalledCode
device 1123 Str
opt_cell_fully 1126 Bool
log_filename 1128 Str
max_force 1124 Float
model 1119 ModelData
precision 1122 Str
structure 1120 StructureData
traj 1129 Str
opt_cell_lengths 1125 Bool
xyz_output_name 1127 Str
Outputs PK Type
--------------- ---- --------------
final_structure 1139 StructureData
log_output 1133 SinglefileData
remote_folder 1131 RemoteData
results_dict 1136 Dict
retrieved 1132 FolderData
std_output 1135 SinglefileData
traj_file 1137 SinglefileData
traj_output 1138 TrajectoryData
xyz_output 1134 SinglefileData
Log messages
---------------------------------------------
There are 1 log messages for this calculation
Run 'verdi process report 1130' to see themThe results can be examined using verdi calcjob commands, such as:
verdi calcjob res 1130
{
"cell": [
[
4.0223130461422,
-8.6767214011906e-17,
2.7878898106399e-16
],
[
2.0111565230711,
3.4834252799327,
2.1832573300987e-16
],
[
2.0111565230711,
1.1611417599776,
3.2842048495961
]
],
"forces": [
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
]
],
"info": {
"energy": -6.7615876501454,
"free_energy": -6.7615876501454,
"spacegroup": "P 1",
"stress": [
[
-0.0001451361211389,
-1.122474781947e-17,
8.992142627858e-18
],
[
-1.122474781947e-17,
-0.00014513612113885,
-1.0680451921731e-17
],
[
8.992142627858e-18,
-1.0680451921731e-17,
-0.00014513612113892
]
],
"unit_cell": "conventional"
},
"numbers": [
11,
17
],
"pbc": [
true,
true,
true
],
"positions": [
[
0.0,
0.0,
0.0
],
[
4.02231305,
2.32228352,
1.64210242
]
],
"spacegroup_kinds": [
0,
1
]
}verdi res contains the results dictionary, which in these calculations is a dictionary containing the xyz_output file content.