### Get Help for Jobflow Runner Start Command

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

Displays detailed usage information and available options for the 'jf runner start' command.

```bash
jf runner start -h
```

--------------------------------

### Start MongoDB Instance

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Command to start a MongoDB database instance, binding to all network interfaces and specifying the database path and port.

```bash
mongod --bind_ip_all  --dbpath /this_is_the_path_to_your_db --quiet --port 27017
```

--------------------------------

### Start MongoDB using Configuration File

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Command to start a MongoDB instance using a specified configuration file and a custom socket prefix.

```bash
mongod -f path/to/mongod.conf --quiet --unixSocketPrefix path/to/store/.sock/file
```

--------------------------------

### MongoDB Configuration File Example

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

An example configuration file for MongoDB, specifying process management, network settings, storage, and logging.

```yaml
processManagement:
   fork: false
   pidFilePath: /dss/dsshome1/00/username/path/to/store/your/pidfile/mongod.pid
net:
   bindIp: 0.0.0.0
   port: 27017
storage:
   dbPath: /dss/dsshome1/00/username/path/to/store/your/db/mongo
systemLog:
   destination: file
   path: /dss/dsshome1/00/username/path/to/mongod.log
   logAppend: true
storage:
   journal:
      enabled: true
```

--------------------------------

### Install Autoplex from Source

Source: https://github.com/autoatml/autoplex/blob/main/docs/dev/dev_install.md

Clone the repository and install autoplex with all development dependencies. Note that non-Python programs for ACE potential fitting are not installed by this command and must be installed separately.

```bash
git clone https://github.com/autoatml/autoplex.git
cd autoplex
pip install -e .[strict,dev,tests,docs]
```

--------------------------------

### Autoplex Workflow Output Example

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/flows.md

Example output showing training and testing errors for an MLIP fit.

```bash
Training error of MLIP (eV/at.): 0.0049634
Testing error of MLIP (eV/at.): 0.0023569
```

--------------------------------

### Mock VASP Setup Note

Source: https://github.com/autoatml/autoplex/blob/main/tutorials/tutorial_phonon.ipynb

This comment indicates that VASP calculations will be mocked for the tutorial. Ensure that folders with pre-computed VASP output files are available for the notebook to execute correctly.

```python
# Please note that I am reusing the same relaxations here in several steps.

```

--------------------------------

### Start MongoDB Service with Homebrew

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Command to start the MongoDB community service using Homebrew. Use `brew services list` to check its status.

```bash
brew services start mongodb-community
```

--------------------------------

### Complete DFT vs ML Benchmark Workflow Setup

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/flows.md

Sets up the CompleteDFTvsMLBenchmarkWorkflow by defining material IDs for structures and benchmarks. This workflow automates VASP and phonopy calculations, MLIP fits, and benchmarks.

```python
mpbenchmark = ["mp-22905"]
for mpid in mpids:
    structure = mpr.get_structure_by_material_id(mpid)
    structure_list.append(structure)
for mpbm in mpbenchmark:
    bm_structure = mpr.get_structure_by_material_id(mpbm)
    benchmark_structure_list.append(bm_structure)

complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    apply_data_preprocessing=True,
).make(
    structure_list=structure_list, mp_ids=mpids, 
    benchmark_structures=benchmark_structure_list, benchmark_mp_ids=mpbenchmark)

complete_flow.name = "tutorial"
autoplex_flow = complete_flow
```

--------------------------------

### Iterative Phonon Workflow Setup

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/flows.md

Sets up an iterative version of the CompleteDFTvsMLBenchmarkWorkflow to converge phonon RMSE. It allows for different configurations in the first generation compared to subsequent ones.

```python
from mp_api.client import MPRester
from autoplex.auto.phonons.flows import CompleteDFTvsMLBenchmarkWorkflow, IterativeCompleteDFTvsMLBenchmarkWorkflow

mpr = MPRester(api_key='YOUR_MP_API_KEY')
structure_list = []
benchmark_structure_list = []
mpids = ["mp-22905"]  
# you can put as many mpids as needed e.g. mpids = ["mp-22905", "mp-1185319"] for all LiCl entries in the Materials Project
mpbenchmark = ["mp-22905"]
for mpid in mpids:
    structure = mpr.get_structure_by_material_id(mpid)
    structure_list.append(structure)
for mpbm in mpbenchmark:
    bm_structure = mpr.get_structure_by_material_id(mpbm)
    benchmark_structure_list.append(bm_structure)

complete_flow=IterativeCompleteDFTvsMLBenchmarkWorkflow(rms_max=0.2, max_iterations=4, 
                                                        complete_dft_vs_ml_benchmark_workflow_0=CompleteDFTvsMLBenchmarkWorkflow(
    apply_data_preprocessing=True, add_dft_phonon_struct=True,
), 
                                                        complete_dft_vs_ml_benchmark_workflow_1=CompleteDFTvsMLBenchmarkWorkflow(
    apply_data_preprocessing=True, add_dft_phonon_struct=False,
)
                                                        ).make(
    structure_list=structure_list, mp_ids=mpids, 
    benchmark_structures=benchmark_structure_list, benchmark_mp_ids=mpbenchmark)

complete_flow.name = "tutorial"
autoplex_flow = complete_flow
```

--------------------------------

### Atomate2 Configuration Example

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

An example of an atomate2.yaml configuration file. This file is used on the remote cluster to specify commands for computational packages like VASP and LOBSTER.

```yaml
VASP_CMD:  your hpc vasp_std cmd
VASP_GAMMA_CMD:  your hpc vasp_gam cmd
LOBSTER_CMD: your hpc lobster cmd
```

--------------------------------

### Install Julia and ACEpotentials Dependencies

Source: https://github.com/autoatml/autoplex/blob/main/README.md

Installs Julia version 1.9.2 and configures Julia with necessary registries and packages for ACEpotentials fitting.

```bash
curl -fsSL https://install.julialang.org | sh -s -- default-channel 1.9.2
```

```julia
using Pkg; Pkg.Registry.add("General"); Pkg.Registry.add(Pkg.Registry.RegistrySpec(url="https://github.com/ACEsuit/ACEregistry")); Pkg.add(Pkg.PackageSpec(;name="ACEpotentials", version="0.6.7")); Pkg.add("DataFrames"); Pkg.add("CSV")
```

--------------------------------

### Install pytest-xdist for Parallel Testing

Source: https://github.com/autoatml/autoplex/blob/main/docs/dev/dev_install.md

Install the pytest-xdist library to enable parallel test execution.

```bash
pip install pytest-xdist
```

--------------------------------

### Install Julia v1.9.2

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/installation/installation.md

Installs a specific version of Julia, which is required for fitting and validating ACEpotentials.

```bash
curl -fsSL https://install.julialang.org | sh -s -- default-channel 1.9.2
```

--------------------------------

### Install MongoDB using Homebrew

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Commands to install MongoDB Community Edition using Homebrew, including tapping the official MongoDB repository.

```bash
brew tap mongodb/brew
brew update
brew install mongodb-community
```

--------------------------------

### Finetune MACE-MP-0 with Custom Settings

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Configure the `CompleteDFTvsMLBenchmarkWorkflowMPSettings` for finetuning MACE-MP-0. This example sets custom volume scale factors, rattle/distort types, and detailed MACE fitting parameters including weights, epochs, and learning rate.

```python
complete_workflow_mace = CompleteDFTvsMLBenchmarkWorkflowMPSettings(
        ml_models=["MACE"],
        volume_custom_scale_factors=[0.95,1.00,1.05], rattle_type=0, distort_type=0,
        apply_data_preprocessing=True,
        ...
    ).make(
        structure_list=[structure],
        mp_ids=["mpid"],
        benchmark_mp_ids=["mpid"],
        benchmark_structures=[structure],
        fit_kwargs_list=[
            {
                "model": "MACE",
                "name": "MACE_final",
                "foundation_model": "large",
                "multiheads_finetuning": False,
                "r_max": 6,
                "loss": "huber",
                "energy_weight": 1000.0,
                "forces_weight": 1000.0,
                "stress_weight": 1.0,
                "compute_stress": True,
                "E0s": "average",
                "scaling": "rms_forces_scaling",
                "batch_size": 1,
                "max_num_epochs": 200,
                "ema": True,
                "ema_decay": 0.99,
                "amsgrad": True,
                "default_dtype": "float64",
                "restart_latest": True,
                "lr": 0.0001,
                "patience": 20,
                "device": "cpu",
                "save_cpu": True,
                "seed": 3
            }
        ],
    )
```

--------------------------------

### Configure M3GNet Fitting Hyperparameters

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Pass M3GNet fitting hyperparameters to the `make` method of `CompleteDFTvsMLBenchmarkWorkflow`. This example sets cutoff, batch size, epochs, and other M3GNet-specific parameters.

```python
complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    ml_models=["M3GNet"], ..., apply_data_preprocessing=True,
).make(...,
    structure_list=[structure],
    mp_ids=["mpid"],
    benchmark_mp_ids=["mpid"],
    benchmark_structures=[structure],
    fit_kwargs_list=[
        {
            "cutoff": 5.0,
            "threebody_cutoff": 4.0,
            "batch_size": 10,
            "max_epochs": 1000,
            "include_stresses": True,
            "hidden_dim": 128,
            "num_units": 128,
            "max_l": 4,
            "max_n": 4,
            "device": "cuda",
            "test_equal_to_val": True
        }
    ],
    ...,
    )
```

--------------------------------

### FireWorks Launchpad Configuration (YAML)

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Configure the launchpad to connect to your MongoDB instance. Ensure 'host', 'port', 'name', 'username', and 'password' match your MongoDB setup.

```yaml
host: login_node_name
port: 27017
name: database_name
username: admin
password: adminpassword
ssl_ca_file: null
logdir: null
strm_lvl: INFO
user_indices: []
wf_user_indices: []
```

--------------------------------

### Generate Jobflow Project

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

Run this command to generate an empty project configuration file in your home directory. This file is essential for jobflow-remote setup.

```bash
jf project generate --full YOUR_PROJECT_NAME
```

--------------------------------

### GAP Default Settings JSON

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Example of the default GAP fitting settings structure, including file naming conventions and parameter names for energy, forces, and virial.

```json
  "general": {
    "at_file": "train.extxyz",
    "energy_parameter_name": "REF_energy",
    "force_parameter_name": "REF_forces",
    "virial_parameter_name": "REF_virial",
    "gp_file": "gap_file.xml"
  },

```

--------------------------------

### Iterative DFT vs ML Benchmark Workflow Setup

Source: https://github.com/autoatml/autoplex/blob/main/tutorials/tutorial_phonon.ipynb

Configures an IterativeCompleteDFTvsMLBenchmarkWorkflow for comparing DFT and ML models. This workflow iteratively refines predictions and includes settings for data preprocessing, structure generation, and model fitting.

```python
from pymatgen.analysis.defects.workflows.builder import IterativeCompleteDFTvsMLBenchmarkWorkflow, CompleteDFTvsMLBenchmarkWorkflow

iteration_flow = IterativeCompleteDFTvsMLBenchmarkWorkflow(
    max_iterations=2, # with the current test data, you can switch between 1 and 2
    rms_max=0.2,
    complete_dft_vs_ml_benchmark_workflow_0=CompleteDFTvsMLBenchmarkWorkflow(
        symprec=1e-3,
        apply_data_preprocessing=True,
        add_dft_rattled_struct=True,
        add_dft_phonon_struct=True,
        volume_custom_scale_factors=[1.0],
        rattle_type=0,
        distort_type=0,
        rattle_std=0.1,  #
        benchmark_kwargs={"relax_maker_kwargs": {"relax_cell": False}},
        supercell_settings={
            "min_length": 10,
            "max_length": 15,
            "min_atoms": 10,
            "max_atoms": 300,
            "fallback_min_length": 9,
        },
        # settings that worked with a GAP
        split_ratio=0.33,
        regularization=False,
        separated=False,
        num_processes_fit=48,
        displacement_maker=phonon_displacement_maker,
        phonon_bulk_relax_maker=phonon_bulk_relax_maker,
        phonon_static_energy_maker=phonon_static_energy_maker,
        rattled_bulk_relax_maker=phonon_bulk_relax_maker,
        isolated_atom_maker=static_isolated_atom_maker,
    ),
    complete_dft_vs_ml_benchmark_workflow_1=CompleteDFTvsMLBenchmarkWorkflow(
        symprec=1e-3,
        apply_data_preprocessing=True,
        add_dft_phonon_struct=False,
        add_dft_rattled_struct=True,
        volume_custom_scale_factors=[1.0],
        rattle_type=0,
        distort_type=0,
        rattle_std=0.1,  # maybe 0.1
        benchmark_kwargs={"relax_maker_kwargs": {"relax_cell": False}},
        supercell_settings={
            "min_length": 10,
            "max_length": 15,
            "min_atoms": 10,
            "max_atoms": 300,
            "fallback_min_length": 9,
        },
        # settings that worked with a GAP
        split_ratio=0.33,
        regularization=False,
        separated=False,
        num_processes_fit=48,
        displacement_maker=phonon_displacement_maker,
        phonon_bulk_relax_maker=phonon_bulk_relax_maker,
        phonon_static_energy_maker=phonon_static_energy_maker,
        rattled_bulk_relax_maker=phonon_bulk_relax_maker,

```

--------------------------------

### Configure MACE Fitting Hyperparameters

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Pass MACE fitting hyperparameters to the `make` method of `CompleteDFTvsMLBenchmarkWorkflow`. This example configures model type, weights, cutoff, epochs, and other MACE-specific parameters.

```python
complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    ml_models=["MACE"], ..., apply_data_preprocessing=True,
).make(...,
    structure_list=[structure],
    mp_ids=["mpid"],
    benchmark_mp_ids=["mpid"],
    benchmark_structures=[structure],
    fit_kwargs_list=[
        {
            "model": "MACE",
            "config_type_weights": '{"Default": 1.0}',
            "hidden_irreps": "128x0e + 128x1o",
            "r_max": 5.0,
            "batch_size": 10,
            "max_num_epochs": 1500,
            "start_swa": 1200,
            "ema_decay": 0.99,
            "correlation": 3,
            "loss": "huber",
            "default_dtype": "float32",
            "device": "cuda"
        }
    ],
    ...
)
```

--------------------------------

### Start MongoDB with Custom Socket Prefix

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Use this command if you encounter errors related to socket file permissions. It allows specifying a custom directory for temporary socket files.

```bash
mongod --bind_ip_all  --dbpath /this_is_the_path_to/your_db --quiet --port 27017 --unixSocketPrefix /new/directory/for/temp/file
```

--------------------------------

### Install Autoplex with Pacemaker Dependencies

Source: https://github.com/autoatml/autoplex/blob/main/README.md

Installs Autoplex with Pacemaker support, along with specific versions of TensorFlow, TensorPotential, and python-ace. Ensure CMake is available before installing python-ace.

```bash
pip install autoplex[pacemaker]
```

```bash
pip install tensorflow==2.8.0
```

```bash
pip install --no-deps git+https://github.com/ICAMS/TensorPotential.git@1e44b2558356800ae070658c0bb856ff9bf74538
```

```bash
pip install --no-deps git+https://github.com/ICAMS/python-ace.git@d1c213a7d9c5b809a3ae83b2e5a916be26d921f0
```

--------------------------------

### Complete DFT vs ML Benchmark Workflow

Source: https://context7.com/autoatml/autoplex/llms.txt

Initiates a full phonon MLIP workflow. This includes generating DFT reference data, preprocessing and splitting the dataset, fitting MLIP models, and benchmarking them against DFT phonon band structures. Configure parameters like ML models, data augmentation, and fitting settings.

```python
from mp_api.client import MPRester
from autoplex.auto.phonons.flows import CompleteDFTvsMLBenchmarkWorkflow
from jobflow_remote import submit_flow

mpr = MPRester(api_key="YOUR_MP_API_KEY")
mpids = ["mp-22905"]           # LiCl rocksalt
mpbenchmark = ["mp-22905"]

structure_list = [mpr.get_structure_by_material_id(m) for m in mpids]
benchmark_structures = [mpr.get_structure_by_material_id(m) for m in mpbenchmark]

complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    ml_models=["GAP"],
    add_dft_phonon_struct=True,      # include phonopy-displaced supercells
    add_dft_rattled_struct=True,     # include rattled supercells
    n_structures=50,                 # number of rattled structures
    symprec=1e-4,
    rattle_std=0.01,
    distort_type=0,                  # 0=volume, 1=angle, 2=both
    volume_scale_factor_range=[0.95, 1.05],
    supercell_settings={"min_length": 15, "max_length": 20, "min_atoms": 50, "max_atoms": 500},
    apply_data_preprocessing=True,
    distillation=True,
    force_max=40.0,
    split_ratio=0.4,
    atom_wise_regularization=True,
    atomwise_regularization_parameter=0.1,
    force_min=0.01,
    num_processes_fit=48,
    summary_filename_prefix="results_",
).make(
    structure_list=structure_list,
    mp_ids=mpids,
    benchmark_structures=benchmark_structures,
    benchmark_mp_ids=mpbenchmark,
    fit_kwargs_list=[{
        "soap": {
            "delta": 1.0, "l_max": 12, "n_max": 10, "atom_sigma": 0.5,
            "zeta": 4, "cutoff": 5.0, "cutoff_transition_width": 1.0,
            "central_weight": 1.0, "n_sparse": 9000, "f0": 0.0,
            "covariance_type": "dot_product", "sparse_method": "cur_points",
        },
        "general": {
            "two_body": False, "three_body": False, "soap": True,
            "default_sigma": "{0.001 0.05 0.05 0.0}", "sparse_jitter": 1e-8,
        },
    }],
)

complete_flow.name = "LiCl_GAP_workflow"
resources = {"nodes": 3, "partition": "micro", "time": "2:00:00", "ntasks": 144}
print(submit_flow(complete_flow, worker="my_worker", resources=resources, project="autoplex"))

# Expected summary output (results_LiCl.txt):
# Potential  Structure  MPID      Displacement(Å)  RMSE(THz)  imagmodes(pot)  imagmodes(dft)
# GAP        LiCl       mp-22905  0.01             0.576      False           False
```

--------------------------------

### Compile and Install AIRSS

Source: https://github.com/autoatml/autoplex/blob/main/README.md

Downloads, compiles, and installs the AIRSS package version 0.9.3. Ensure you are in the correct directory after extraction.

```bash
curl -O https://www.mtg.msm.cam.ac.uk/files/airss-0.9.3.tgz; tar -xf airss-0.9.3.tgz; rm airss-0.9.3.tgz; cd airss; make ; make install ; make neat; cd ..
```

--------------------------------

### Complete DFT vs ML Phonon Benchmark Workflow

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/benchmark/benchmark.md

Sets up and runs a complete benchmark workflow comparing DFT and ML phonon calculations. It can utilize pre-existing DFT data or automatically execute VASP calculations if no references are provided. A mix of pre-existing and missing DFT references is not supported.

```python
from mp_api.client import MPRester
from autoplex.auto.phonons.flows import CompleteDFTvsMLBenchmarkWorkflow
from atomate2.common.schemas.phonons import PhononBSDOSDoc
from monty.serialization import loadfn

mpr = MPRester(api_key='YOUR_MP_API_KEY')
structure_list = []
benchmark_structure_list = []
mpids = ["mp-22905"]
mpbenchmark = ["mp-22905"]
dft_data = loadfn("/path/to/DFT/ref/data/PhononBSDOSDoc_mp_22905.json")
dft_reference: PhononBSDOSDoc = dft_data["output"]
for mpid in mpids:
    structure = mpr.get_structure_by_material_id(mpid)
    structure_list.append(structure)
for mpbm in mpbenchmark:
    bm_structure = mpr.get_structure_by_material_id(mpbm)
    benchmark_structure_list.append(bm_structure)

complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    apply_data_preprocessing=True,
).make(
    structure_list=structure_list, mp_ids=mpids, 
    benchmark_structures=benchmark_structure_list, benchmark_mp_ids=mpbenchmark, 
    dft_references=[dft_reference])
```

--------------------------------

### Install Autoplex with Strict Dependencies

Source: https://github.com/autoatml/autoplex/blob/main/README.md

Use this command to install Autoplex along with all necessary Python packages and dependencies for MLIP fits.

```bash
pip install autoplex[strict]
```

--------------------------------

### Configure DFT vs ML Benchmark Workflow

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Set up a benchmark workflow including ML models, data preprocessing, force thresholds, and split ratios. MLIP-specific fit parameters can be passed via `fit_kwargs_list`.

```python
complete_flow = CompleteDFTvsMLBenchmarkWorkflow(
    ml_models=["GAP", "MACE"], ..., 
    apply_data_preprocessing=True, 
    f_max=40.0, split_ratio=0.4,
    num_processes_fit=48,
    benchmark_kwargs={"relax_maker_kwargs": {"relax_cell": False, "relax_kwargs": ...}, "calculator_kwargs": {"device": "cpu"}}
).make(..., 
    fit_kwargs_list=[
        {"general": {"two_body": True, "three_body": False, "soap": False}},  # GAP parameters
        {"model": "MACE", "device": "cuda"}  # MACE parameters  
        # fit_kwargs_list has to have the same order as in ml_models
    ],
    ...  # put the other hyperparameter commands here as shown below
)
```

--------------------------------

### Create and Run Phonon Benchmark

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/benchmark/benchmark.md

This snippet demonstrates how to initialize and run a PhononBenchmarkMaker using a specified structure and DFT reference data. It configures the underlying relaxation and static energy makers with a GAP potential and then creates and runs the benchmark job.

```python
from atomate2.forcefields.jobs import ForceFieldRelaxMaker, ForceFieldStaticMaker
from atomate2.phonons.jobs import PhononMaker, PhononBenchmarkMaker
from jobflow import run_locally
from mp_api.client import MPRester
from pymatgen.core import Structure
from monty.serialization import loadfn

# Connect to the job store
# store.connect() # Assuming 'store' is already defined and connected

mpr = MPRester(api_key = 'YOUR_API_KEY')
mpid = "mp-22905"
structure =  mpr.get_structure_by_material_id(mpid)
dft_data = loadfn("/path/to/DFT/ref/data/PhononBSDOSDoc_mp_22905.json")
dft_reference: PhononBSDOSDoc = dft_data["output"]
potential_filename = "/path/to/GAP/file/gap_file.xml"

phojob = PhononMaker(
        bulk_relax_maker=ForceFieldRelaxMaker(calculator_kwargs={"args_str": "IP GAP", "param_filename": potential_filename},
        relax_cell=True, relax_kwargs={"interval": 500, "fmax": 0.00001}, steps=10000,
        force_field_name="GAP",
),
        phonon_displacement_maker=ForceFieldStaticMaker(calculator_kwargs={"args_str": "IP GAP", "param_filename": potential_filename},
        force_field_name="GAP",
),
        static_energy_maker=ForceFieldStaticMaker(calculator_kwargs={"args_str": "IP GAP", "param_filename": potential_filename}),
        store_force_constants=False, min_length=18,
        generate_frequencies_eigenvectors_kwargs={"units": "THz"}, force_field_name="GAP",
).make(structure=structure)
        
bm = PhononBenchmarkMaker(name="Benchmark").make(
    structure=structure, benchmark_mp_id = "mp-22905",
    displacement=0.01, atomwise_regularization_parameter=0.1,
    soap_dict={'n_sparse': 6000, 'delta': 0.5}, suffix="",  # exemplary values
    ml_phonon_task_doc = phojob.output, dft_phonon_task_doc = dft_reference)

comp_bm = write_benchmark_metrics(
            benchmark_structures=[structure],
            metrics=[[bm.output]],
        )

run_locally([phojob, bm, comp_bm], create_folders=True, store=store)
```

--------------------------------

### Successful LAMMPS Installation Output

Source: https://github.com/autoatml/autoplex/blob/main/README.md

This is the expected output when the LAMMPS Python interface is successfully installed and initialized. It confirms the version and threading settings.

```bash
LAMMPS (29 Aug 2024 - Update 1)
OMP_NUM_THREADS environment is not set. Defaulting to 1 thread. (src/comm.cpp:98)
  using 1 OpenMP thread(s) per MPI task
>>>
```

--------------------------------

### Initialize CompleteDFTvsMLBenchmarkWorkflow

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/flows.md

Construct the `CompleteDFTvsMLBenchmarkWorkflow` using `MPRester` to fetch structures. This workflow is designed for benchmarking machine-learned potentials against DFT results for phonon calculations.

```python
from mp_api.client import MPRester
from autoplex.auto.phonons.flows import CompleteDFTvsMLBenchmarkWorkflow

mpr = MPRester(api_key='YOUR_MP_API_KEY')
structure_list = []
benchmark_structure_list = []
mpids = ["mp-22905"]  

```

--------------------------------

### Combine Existing Dataset with RSS for New MLIP Training

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/rss/flow/quick_start/start.md

Illustrates how to initialize an RSS workflow that merges an existing dataset with newly generated RSS structures to train a new MLIP. This is useful when you have prior data or no trained MLIP yet.

```python
from autoplex.auto.rss.config import RssConfig
rss_config = RssConfig.from_file('path/to/your/config.yaml')
rss_job = RssMaker(name="your workflow name", 
                   rss_config=rss_config).make(
                   pre_database_dir='path/to/pre-existing/database')
```

--------------------------------

### Install GCC and Gfortran for AIRSS

Source: https://github.com/autoatml/autoplex/blob/main/README.md

Installs the necessary GCC and Gfortran compilers (version 5 and above) required for building AIRSS utilities on Ubuntu/Debian systems.

```bash
apt install -y build-essential gfortran
```

--------------------------------

### Configure CASTEP Static Calculation Parameters

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/rss/flow/input/input.md

Initialize a CastepStaticMaker with user-defined parameters for cutoff energy, XC functional, SCF convergence, and k-point settings. Ensure these parameters are tuned for your specific system to guarantee convergence.

```python
from autoplex.settings import RssConfig
from autoplex.auto.rss.flows import RssMaker
from autoplex.misc.castep.jobs import CastepStaticMaker
from autoplex.misc.castep.utils import CastepStaticSetGenerator

castep_maker = CastepStaticMaker(
    name="static_castep",
    input_set_generator=CastepStaticSetGenerator(
        user_param_settings={
            "cut_off_energy": 600.0,
            "xc_functional": "PBE",
            "max_scf_cycles": 1000,
            "elec_energy_tol": 1e-5,
            "smearing_scheme": "Gaussian",
            "smearing_width": 0.05,   
            "spin_polarized": False,
            "finite_basis_corr": "automatic",
            "perc_extra_bands": 100  
        },
        user_cell_settings={
            "kpoint_mp_spacing": 0.03, 
            "symmetry_generate": True,
        },
    ),
)
```

--------------------------------

### Load RSS Configuration and Create RSS Job

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/rss/flow/input/input.md

Load an RSS configuration from a file and then create an RssMaker instance. Ensure that the static energy maker and static energy maker for isolated atoms are provided as per the specific requirements of your project.

```python
rss_config = RssConfig.from_file('/path/to/rss/config/file') 

rss_job = RssMaker(name="Si_rss", 
                   rss_config=rss_config,
                   static_energy_maker=# code-specific static maker, see above
                   static_energy_maker_isolated_atoms=# code-specific static maker, see above
                   ).make()
```

--------------------------------

### Autoplex Workflow Results Summary Example

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/flows.md

Example summary of essential results from an Autoplex workflow run, including potential type, structure, RMSE, and parameters.

```text
Potential  Structure  MPID        Displacement (Å)  RMSE (THz)  imagmodes(pot)  imagmodes(dft)  Database type   (Hyper-)Parameters
GAP        LiCl       mp-22905    0.01              0.57608     False           False           full            atom-wise f=0.1: n_sparse = 6000, SOAP delta = 0.5
```

--------------------------------

### Generate RSS Data and Train MLIP

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/rss/flow/quick_start/start.md

This snippet shows how to initialize an RSS configuration, generate a new RSS dataset, preprocess and merge it with existing data, and then train an MLIP model. Ensure your configuration file path and pre-existing database path are correctly specified.

```python
from autoplex.settings import RssConfig
from autoplex.auto.rss.flows import RssMaker
from autoplex.fitting.common.flows import MLIPFitMaker
from autoplex.data.common.jobs import preprocess_data

rss_config = RssConfig.from_file('path/to/your/config.yaml')
rss_job = RssMaker(name="your workflow name", 
                   rss_config=rss_config).make()

data_preprocessing_job = preprocess_data(
        dft_ref_dir=rss_job.output["mlip_path"],  # The path to store the RSS dataset can be read from the previous job.
        pre_database_dir='path/to/pre-existing/database',  # The path to store the pre-existing dataset that you'd like to merge.
    )

fitting_job = MLIPFitMaker(
    mlip_type="MACE",              # Select the potential model you want to use for training.
    ref_energy_name="REF_energy",  # Define the name of the labels you are using.
    ref_force_name="REF_forces",
    ref_virial_name="REF_virial",
    apply_data_preprocessing=False,
    ).make(
    isolated_atom_energies=rss_job.output["isolated_atom_energies"],
    database_dir=data_preprocessing_job.output,
    device='cuda',
    **fit_kwargs,   # define the hyperparameters for potential training here.
    )

jobs = [rss_job, data_preprocessing_job, fitting_job]
```

--------------------------------

### Check for Errors in Jobflow Project Setup

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

This command performs a more comprehensive check of your jobflow-remote setup, including MongoDB connection verification. Use it to diagnose any issues.

```bash
jf project check --errors
```

--------------------------------

### Install autoplex with workflow managers

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/installation/installation.md

Installs autoplex along with optional support for workflow managers like FireWorks and jobflow-remote. This command enables advanced workflow capabilities.

```bash
pip install autoplex[strict,workflow-managers]
```

--------------------------------

### Build Documentation Locally

Source: https://github.com/autoatml/autoplex/blob/main/docs/dev/dev_install.md

Build the autoplex documentation using sphinx-build, outputting to the _build directory. The -W flag will treat warnings as errors.

```bash
sphinx-build -W docs _build
```

--------------------------------

### Start Jobflow Runner in Interactive Mode

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

Starts the Jobflow runner daemon in single-process interactive mode, enabling MFA login with OTP. Requires manual confirmation for host connection and subsequent password/OTP entry.

```bash
jf runner start -s -i
```

--------------------------------

### Troubleshoot Stuck Tests

Source: https://github.com/autoatml/autoplex/blob/main/docs/dev/dev_install.md

Prefix pytest with OMP_NUM_THREADS=1 to resolve issues where test execution gets stuck.

```bash
OMP_NUM_THREADS=1 pytest
```

--------------------------------

### Check Jobflow Project Setup

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/jobflowremote.md

Verify that your jobflow-remote project is set up correctly. This command checks worker configuration and connection status.

```bash
jf project check -w example_worker
```

--------------------------------

### Update User INCAR Settings

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Modifies the VASP INCAR settings for the Autoplex flow. This example specifically sets the NPAR parameter to 4.

```python
autoplex_flow = update_user_incar_settings(autoplex_flow, {"NPAR": 4})
```

--------------------------------

### Create atomate2 Directory Scaffold

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/mongodb.md

Creates the necessary directory structure for atomate2, including configuration and log directories.

```bash
mkdir -p atomate2/{config,logs}
```

--------------------------------

### Run Benchmark with Pre-existing DFT and GAP Potential

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/benchmark/benchmark.md

Template for running or repeating a phonon benchmark using pre-existing DFT calculations and a GAP potential. Requires DFT data in the form of a `PhononBSDOSDoc` task document object. Sets environment variables for threading and configures the job store.

```python
#!/usr/bin/env python

import os
from atomate2.common.schemas.phonons import PhononBSDOSDoc
from atomate2.forcefields.jobs import ForceFieldRelaxMaker, ForceFieldStaticMaker
from mp_api.client import MPRester
from autoplex.benchmark.phonons.flows import PhononBenchmarkMaker
from atomate2.forcefields.flows.phonons import PhononMaker
from autoplex.benchmark.phonons.jobs import write_benchmark_metrics
from monty.serialization import loadfn
from jobflow import SETTINGS
from jobflow import run_locally

os.environ["OMP_NUM_THREADS"] = "48"
os.environ["OPENBLAS_NUM_THREADS"] = "1"

store = SETTINGS.JOB_STORE

```

--------------------------------

### Add Julia ACEpotentials Dependencies

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/installation/installation.md

Configures Julia's package manager to add necessary registries and install the ACEpotentials package along with DataFrames and CSV.

```bash
julia -e 'using Pkg; Pkg.Registry.add("General"); Pkg.Registry.add(Pkg.Registry.RegistrySpec(url="https://github.com/ACEsuit/ACEregistry")); Pkg.add(Pkg.PackageSpec(;name="ACEpotentials", version="0.6.7")); Pkg.add("DataFrames"); Pkg.add("CSV")'
```

--------------------------------

### Set Autoplex Flow Name

Source: https://github.com/autoatml/autoplex/blob/main/docs/user/phonon/flows/fitting/fitting.md

Assigns a descriptive name to the Autoplex flow. This example names the flow 'small Sn test, test without phonon2'.

```python
autoplex_flow.name = "small Sn test, test without phonon2"
```

--------------------------------

### Benchmark MLIP Phonons Against DFT with PhononBenchmarkMaker

Source: https://context7.com/autoatml/autoplex/llms.txt

Benchmark a fitted MLIP against DFT phonon calculations using PhononBenchmarkMaker. This computes band-structure RMSE and generates diagnostic plots. Requires `ml_phonon_task_doc` and `dft_phonon_task_doc` from atomate2 runs.

```python
from autoplex.benchmark.phonons.flows import PhononBenchmarkMaker
from jobflow import run_locally

# ml_phonon_task_doc and dft_phonon_task_doc are PhononBSDOSDoc objects
# obtained from atomate2 PhononMaker runs
benchmark_job = PhononBenchmarkMaker().make(
    ml_model="GAP",
    structure=structure,          # pymatgen Structure
    benchmark_mp_id="mp-22905",
    ml_phonon_task_doc=ml_doc,
    dft_phonon_task_doc=dft_doc,
    displacement=0.01,
    atomwise_regularization_parameter=0.1,
    soap_dict={"n_sparse": 6000, "delta": 1.0},
    suffix="",
)

run_locally(benchmark_job, create_folders=True)
# Outputs: RMSE (THz), band structure comparison PNG, q-point RMSE PNG
```