AENET Solver

Overview

AenetSolver is a solver class that inherits from odatse.solver.SolverBase in ODAT-SE. It runs AENET’s predict.x as a subprocess, substitutes parameters into a structure template, and calculates interatomic energies.

How the Solver Works

When the evaluate method is called, the following steps are performed:

Create a working directory Log{step}_{iset}/
Copy predict.in and the ANN potential file
Replace placeholders in the template file (e.g., value_01) with parameter values to generate the input structure file
Execute predict.x
Extract the total energy and the number of atoms from predict.out
Return the energy per atom (eV/atom) as the R-factor

Source Code Structure

src/AenetSolver/
  __init__.py        # Export of the Solver class
  _main.py           # CLI entry point (odatse.initialize + algorithm dispatch)
  aenet.py           # Solver class implementation

Input Files

input.toml

This is the standard ODAT-SE TOML configuration file.

[base]
dimension = 1
output_dir = "output"

[solver]
name = "aenet"

[solver.config]
aenet_exec_file = "predict.x"           # Path to predict.x
aenet_ann_potential = "N.5t-5t.ann"      # ANN potential file
aenet_template_file = "template.xsf"    # Template structure file (default)
aenet_input_file = "structure_data.xsf"  # Generated input file name (default)
bulk_output_file = "predict.in"          # AENET configuration file (default)

[solver.param]
string_list = ["value_01"]              # Placeholders in the template

[solver.post]
remove_work_dir = false                 # Whether to remove the working directory after calculation

[algorithm]
name = "mapper"                          # Algorithm to use
label_list = ["z"]

[algorithm.param]
min_list = [0.8]
max_list = [1.4]
num_list = [101]

solver.config Section

Parameter	Default	Description
`aenet_exec_file`	`predict.x`	Path to the AENET predict.x executable
`aenet_ann_potential`	`N.10t-10t.ann`	Pre-trained ANN potential file
`aenet_template_file`	`template.xsf`	Structure template file
`aenet_input_file`	`structure_data.xsf`	Name of the generated structure input file
`bulk_output_file`	`predict.in`	AENET configuration file

solver.param Section

Parameter	Default	Description
`string_list`	`["value_01", "value_02"]`	List of placeholders to be replaced in the template

Template File

The template file (template.xsf) is a structure file in XSF format, where placeholders are placed at the positions of the optimization parameters.

ATOMS
N             0.0000000000        0.0000000000        0.0000000000
N             0.0000000000        0.0000000000        value_01

In this example, value_01 corresponds to the N-N bond distance and is optimized by the algorithm.

predict.in

This is the configuration file for AENET prediction calculations.

TYPES
1
N

NETWORKS
  N N.5t-5t.ann

FORCES

FILES
1
structure_data.xsf

How to Run

CLI execution:

odatse-aenet input.toml

MPI parallel execution:

mpiexec -np 4 odatse-aenet input.toml

Direct execution from a script:

python3 src/main.py input.toml