Dimer

Overview

The Dimer class is a core component of NET-Finder, responsible for performing transition state searches using the dimer method. This method is an efficient algorithm for finding saddle points on potential energy surfaces without requiring an initial guess for the transition state geometry.

Key Features

Implements the dimer method for transition state search
Supports various displacement methods for initial dimer orientation
Allows for customization of dimer parameters
Performs reaction path optimization from identified transition states

Basic Usage

Here’s a basic example of how to use the Dimer class:

from ase.io import read
from ase.calculators.emt import EMT
from net_finder.search import Dimer

# Load initial structure
atoms = read('initial_structure.vasp')

# Set up calculator
calc = EMT()

# Set up dimer parameters
dimer_params = {
    'displacement_method': 'gauss',
    'displacement_radius': 2.0,
    'dimer_separation': 0.01
}

# Create Dimer instance
dimer = Dimer(atoms, calc, output_dir='dimer_output', dimer_params=dimer_params)

# Run dimer method
reaction_path = dimer.run(fmax=0.05, steps=100)

Parameters

atoms (ase.Atoms): The initial atomic structure.
calc (ase.calculators.Calculator): The ASE calculator to use for energy and force calculations.
output_dir (str): Directory to store output files.
dimer_params (dict): Dictionary of dimer method parameters.
- displacement_method (str): Method for initial displacement (‘gauss’, ‘random’, etc.).
- displacement_radius (float): Radius for the initial displacement.
- dimer_separation (float): Separation distance between dimer images.
- fix_index (int, optional): Fix atoms with index less than this value.

Methods

This method performs the dimer search and returns a ReactionPath object.

Parameters:

fmax (float): Maximum force criterion for optimization.
steps (int): Maximum number of optimization steps.

Returns:

ReactionPath: Object containing the optimized reaction path.

Advanced Usage

For more advanced usage, you can customize the dimer parameters to suit your specific system:

dimer_params = {
    'displacement_method': 'random',
    'displacement_radius': 1.5,
    'dimer_separation': 0.005,
    'fix_index': 10  # Fix the first 10 atoms
}

dimer = Dimer(atoms, calc, output_dir='custom_dimer_output', dimer_params=dimer_params)

You can also access intermediate results and trajectories from the Dimer instance after running:

reaction_path = dimer.run(fmax=0.05, steps=100)

# Access transition state structure
ts_structure = reaction_path.transition_state

# Access reaction path energies
energies = reaction_path.energies

# Access full trajectory
trajectory = dimer.get_trajectory()

Notes

The dimer method can be sensitive to initial conditions. If you’re not finding the desired transition state, try adjusting the displacement_radius or using a different displacement_method.
For systems with rigid parts (e.g., a surface), use the fix_index parameter to keep part of the structure fixed during the search.
The run method can be computationally intensive for large systems or accurate calculators. Consider using parallel computing resources when available.