gridmlip 0.1__tar.gz → 0.2__tar.gz

{gridmlip-0.1/gridmlip.egg-info → gridmlip-0.2}/PKG-INFO

@@ -1,12 +1,12 @@
 Metadata-Version: 2.1
 Name: gridmlip
-Version: 0.1
+Version: 0.2
 Summary: Grid-based method for calculating the percolation barrier of mobile species using machine learning interatomic potentials
 Home-page: https://github.com/dembart/gridmlip
 Author: Artem Dembitskiy
 Author-email: art.dembitskiy@gmail.com
 License: MIT
-Keywords: MLIP,percolation barrier
+Keywords: MLIP,percolation barrier,irreducible mesh
 Platform: UNKNOWN
 Classifier: Programming Language :: Python :: 3
 Classifier: License :: OSI Approved :: MIT License
@@ -31,7 +31,6 @@ License-File: LICENSE
 
 **gridmlip** is a library for calculating percolation barriers of mobile species in solids using grid-based method with machine learning interatomic potentials (MLIPs).
 
-
 ## Installation
 
 ```
@@ -53,7 +52,7 @@ Here we describe the pipeline in general. For a specific example, see [Notebooks
 from gridmlip import Grid
 
 atomic_types_mapper = {3:0, 31:1, 17:2}
-grig = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
+grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
                     atomic_types_mapper=atomic_types_mapper # optional
                   )
 cfgs = grid.construct_configurations('data.cfg')
@@ -72,7 +71,7 @@ from gridmlip import Grid
 
 grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8)
 grid.read_processed_configurations('processed_data.cfg', format = 'cfg')
-grid.percolation_barriers()
+barriers = grid.percolation_barriers()
 ```
 
 #### Step #4: Write .grd or .cube file for visualization in VESTA 3.0
@@ -84,7 +83,8 @@ g.write_grd('test.grd')
 ## Notebooks
 
 
-- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb) 
+- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb)    
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14gabZ-u19K-_I_e7-g67N8-xtPIN36Nw#sandboxMode=true&scrollTo=5TySp6C6UBoZ)
 
 
 ### How to cite

{gridmlip-0.1 → gridmlip-0.2}/README.md

@@ -15,7 +15,6 @@
 
 **gridmlip** is a library for calculating percolation barriers of mobile species in solids using grid-based method with machine learning interatomic potentials (MLIPs).
 
-
 ## Installation
 
 ```
@@ -37,7 +36,7 @@ Here we describe the pipeline in general. For a specific example, see [Notebooks
 from gridmlip import Grid
 
 atomic_types_mapper = {3:0, 31:1, 17:2}
-grig = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
+grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
                     atomic_types_mapper=atomic_types_mapper # optional
                   )
 cfgs = grid.construct_configurations('data.cfg')
@@ -56,7 +55,7 @@ from gridmlip import Grid
 
 grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8)
 grid.read_processed_configurations('processed_data.cfg', format = 'cfg')
-grid.percolation_barriers()
+barriers = grid.percolation_barriers()
 ```
 
 #### Step #4: Write .grd or .cube file for visualization in VESTA 3.0
@@ -68,7 +67,8 @@ g.write_grd('test.grd')
 ## Notebooks
 
 
-- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb) 
+- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb)    
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14gabZ-u19K-_I_e7-g67N8-xtPIN36Nw#sandboxMode=true&scrollTo=5TySp6C6UBoZ)
 
 
 ### How to cite

gridmlip-0.2/gridmlip/_mesh.py

@@ -0,0 +1,139 @@
+import numpy as np
+from spglib import get_symmetry_dataset
+
+__author__ = "Artem Dembitskiy"
+
+class Mesh:
+    """
+    This class implements mesh generation and finding irreducible mesh
+    points given the rotations and translations symmetry operations.
+
+    Example
+    -------
+    
+    mesh = Mesh.from_atoms(atoms, resolution = 0.2, symprec = 0.1)
+    ir_grid_points, mapping = mesh.get_irreducible_mapping()
+    
+    """
+    
+    def __init__(self, mesh_shape, rotations, translations, symmetry_dataset=None):
+        """
+        Initialization.
+
+        
+        Parameters
+        ----------
+        
+        mesh_shape: tuple of integers, (nx, ny, nz)
+            shape of the mesh
+
+        rotations: list of numpy arrays of shape (3, 3)
+            rotation matrices
+
+        translations: list of numpy arrays of shape (3, 1)
+            translation vector
+        
+        symmetry_dataset: dict generated by spglib library, None by default (Optional)
+            dataset containing symmetry operations, used for tests
+        """
+        
+        self.mesh_shape = mesh_shape
+        self.rotations = rotations
+        self.translations = translations
+        self.symmetry_dataset = symmetry_dataset
+        self.mesh = self._generate_mesh()
+
+    @property
+    def addresses(self):
+        return self.mesh
+        
+    
+    @classmethod
+    def from_atoms(cls, atoms, resolution=0.2, symprec=None):
+        """
+        Create Mesh object  form ASE's Atoms object.
+
+        
+        Parameters
+        ----------
+        
+        atoms: ASE's Atoms object
+            atomic structure
+        
+        resolution: float, 0.2 by default
+            distance between points of the mesh in angstroms
+        symprec: None or float, None by default
+            symmetry precision
+            used by spglib' get_symmetry_dataset to find symmetry operations
+        """
+  
+        rotations = [np.eye(3, dtype=int)]
+        translations = [np.zeros(3)]
+        symmetry_dataset = None
+        a, b, c = np.linalg.norm(atoms.cell, axis = 1)
+        nx, ny, nz = int(a // resolution), int(b // resolution), int(c // resolution)
+        nx, ny, nz = [n + (n % 2 + 1) % 2 for n in (nx, ny, nz)] # make odd numbers
+        mesh_shape = (nx, ny, nz)
+        
+        if symprec:
+            spglib_cell = (atoms.cell, atoms.get_scaled_positions(), atoms.numbers)
+            symmetry_dataset = get_symmetry_dataset(spglib_cell, symprec=symprec)
+            rotations = symmetry_dataset.rotations
+            translations = symmetry_dataset.translations
+        
+        return cls(mesh_shape, rotations, translations, symmetry_dataset=symmetry_dataset)
+        
+    def _generate_mesh(self):
+        """Generate integer grid points (i,j,k) for a 3D mesh, including edges and faces."""
+        grid = np.mgrid[
+            0:self.mesh_shape[0],
+            0:self.mesh_shape[1],
+            0:self.mesh_shape[2]
+        ].reshape(3, -1).T
+        return grid  # shape (N,3), integer
+
+    
+    def _index_to_linear(self, indices):
+        """Convert 3D integer indices -> 1D linear indices for fast lookup."""
+        return np.ravel_multi_index(indices.T, self.mesh_shape)
+
+    
+    def _apply_sym_ops(self):
+        """Apply symmetry ops to integer grid points, wrapping by mesh_shape."""
+        all_images = []
+        mesh_shape = np.array(self.mesh_shape)
+        for R, t in zip(self.rotations, self.translations):
+            t_mesh = np.rint(t * (mesh_shape - 1)).astype(int)
+            transformed = (self.mesh @ R.T + t_mesh) % (mesh_shape - 1)
+            all_images.append(transformed)
+        return np.stack(all_images, axis=0)  # (nsym, npoints, 3)
+
+    
+    def get_irreducible_mapping(self):
+        """
+        Find symmetry-equivalent points on an integer 3D mesh.
+        
+        Returns
+        -------
+        indices of the irreducible points of the mesh and 
+        mapping indices to restore full mesh from irreducible points
+        
+        """
+        
+        npoints = len(self.mesh)
+        nsym = len(self.rotations)
+        
+        lin_indices = self._index_to_linear(self.mesh)
+        index_map = np.arange(npoints, dtype=int)
+    
+        all_images = self._apply_sym_ops()
+        
+        transformed_lin = self._index_to_linear(all_images.reshape(-1, 3))
+        transformed_lin = transformed_lin.reshape(nsym, npoints)
+        
+        eq_ids = index_map[transformed_lin]
+        eq_ids = np.concatenate([eq_ids, index_map[None, :]], axis=0)
+        irreducible_ids = eq_ids.min(axis=0)
+        
+        unique, inverse = np.unique(irreducible_ids, return_inverse=True)
+        return unique, inverse

gridmlip-0.2/gridmlip/core.py

@@ -0,0 +1,333 @@
+import numpy as np
+from tqdm import tqdm
+from ase.io import read, write, cube
+from pymatgen.io.ase import AseAtomsAdaptor
+from .utils import read_cfg, write_cfg
+from ._neighborhood import nn_list
+from ._percolation_analysis import Percolyze
+from ._mesh import Mesh
+
+
+
+__author__ = "Artem Dembitskiy"
+__version__ = "0.2"
+
+
+
+class Grid:
+
+    def __init__(self,
+                 atoms,
+                 specie,
+                 resolution=0.25,
+                 r_cut=5.0,
+                 r_min=1.5,
+                 symprec=0.1,
+                 atomic_types_mapper= None,
+                 empty_framework=True,
+                 verbose=True,
+                 ):
+        """ 
+        Initialization. 
+
+        Parameters
+        ----------
+        
+        atoms: ASE's Atoms object
+            atomic structure
+
+        specie: int
+            atomic number of the mobile specie
+
+        resolution: float, 0.2 by default
+            spacing between points (in Angstroms)
+
+        r_cut: float
+            cutoff radius to find the first nearest neighbor for each grid point
+
+        r_min: float
+            blocking sphere radius
+
+        atomic_types_mapper: dict, optional
+            mapper of atomic numbers into species used by MLIP
+
+        symprec: None or float, 0.1 by default
+            symmetry precision
+            used to get symmetry operations for finding the irreducible mesh points
+            may reduce number of MLIP calculations by factor of ~10-100
+            if None, the symmetry will not be used.
+
+        empty_framework: boolean, True by default
+            remove mobile types of interest from the structure
+        """
+
+        self.symprec = symprec
+        self.verbose = verbose
+        if atomic_types_mapper:
+            numbers = self._map_atomic_types(atomic_types_mapper, atoms.numbers)
+            atoms = atoms.copy()
+            atoms.numbers = numbers
+            specie = atomic_types_mapper[specie]
+        self.atoms = atoms.copy()
+        self.specie = specie
+        self.resolution = resolution
+        self.cell = self.atoms.cell
+        self.r_cut = r_cut
+        self.r_min = r_min
+        self.base = self.atoms[self.atoms.numbers != self.specie] if empty_framework else atoms.copy()
+        
+        self._mesh = Mesh.from_atoms(self.base,
+                               resolution = self.resolution,
+                               symprec = self.symprec)
+        self.mesh_shape = self._mesh.mesh_shape
+        self.ir_grid_points, self.mapping = self._mesh.get_irreducible_mapping()
+        self.mesh_frac = self._mesh.addresses / (np.array(self._mesh.mesh_shape) - 1)
+        self.mesh_cart = np.dot(self.mesh_frac, self.base.cell)
+        
+        self.min_dists, _ = nn_list(self.base.positions,
+                                    self.mesh_cart[self.ir_grid_points],
+                                    r_cut,
+                                    self.cell)
+        if self.verbose:
+            total_points = self.mesh_frac.shape[0]
+            irreducible_points = self.ir_grid_points.shape[0]
+            passed_filter = (self.min_dists > self.r_min).sum()
+            compression = total_points / passed_filter if passed_filter else np.nan
+        
+            print("Mesh Summary")
+            print("────────────────────────────────────")
+            print(f"  Mesh shape:                         {tuple(self.mesh_shape)}")
+            print(f"  Symops:                             {len(self._mesh.rotations)}")
+            print(f"  Total points:                       {total_points:,}")
+            print(f"  Irreducible points:                 {irreducible_points:,}")
+            print(f"  Irreducible points (r_min filter):  {passed_filter:,}")
+            print(f"  Compression ratio:                  {compression:6.2f}×")
+            print("────────────────────────────────────\n")
+
+        
+
+    def _map_atomic_types(self, atomic_types_mapper, numbers):
+        u,inv = np.unique(numbers,return_inverse = True)
+        return np.array([atomic_types_mapper[x] for x in u])[inv].reshape(numbers.shape)
+
+    
+    @classmethod
+    def from_file(cls, file,
+                  specie,
+                  resolution = 0.25,
+                  r_cut = 5.0,
+                  r_min = 1.5, 
+                  symprec = 0.1,
+                  atomic_types_mapper = None, 
+                  empty_framework = True,
+                 ):
+        """ 
+        Create Grid object from the file.
+
+        Parameters
+        ----------
+        
+        file: string
+            .xyz of .cfg file representing atomic structure
+
+        specie: int
+            atomic number of the mobile specie
+
+        resolution: float, 0.2 by default
+            spacing between points (in Angstroms)
+
+        r_cut: float
+            cutoff radius to find the first nearest neighbor for each grid point
+
+        r_min: float
+            blocking sphere radius
+
+        atomic_types_mapper: dict, optional
+            mapper of the species into atomic numbers
+
+        empty_framework: boolean, True by default
+            whether to remove mobile types of interest from the structure
+
+        symprec: None or float, 0.1 by default
+            symmetry precision
+            used to get symmetry operations for finding the irreducible mesh points
+            may reduce number of MLIP calculations by factor of 10-100
+            if None, the symmetry will not be used.
+
+        Returns
+        -------
+        Grid object
+        """
+
+        if file.split('.')[-1] == 'cfg':
+            atoms = read_cfg(file)[0]
+        else:
+            atoms = read(file)
+        return cls(atoms, specie, resolution=resolution,
+                   symprec=symprec,
+                   r_cut=r_cut, r_min=r_min, atomic_types_mapper=atomic_types_mapper,
+                   empty_framework=empty_framework
+                  )
+
+    
+
+    def construct_configurations(self, config_format = 'ase', filename = None):
+        """ 
+        Construct atomic configurations for further calculations.
+
+        Parameters
+        ----------
+        
+        filename: string (Optional)
+            path to save .xyz of .cfg file with atomic configurations
+        
+        config_format: string, "ase" by default, can be "pymatgen" or "ase"
+            data format of the created configurations (pymatgen's Structure of ASE's Atoms)
+
+        
+        Returns
+        -------
+        configurations: list of ASE's atoms object
+        """
+
+        configurations = []
+        if config_format == 'ase':
+            ir_mesh_cart = self.mesh_cart[self.ir_grid_points]
+            assert len(ir_mesh_cart) == len(self.min_dists)
+            for p, d in tqdm(zip(ir_mesh_cart, self.min_dists), desc = 'creating configurations'):
+                if d > self.r_min:
+                    framework = self.base.copy()
+                    framework.append(self.specie)
+                    framework.positions[-1] = p
+                    configurations.append(framework)
+        elif config_format == 'pymatgen':
+            base = AseAtomsAdaptor.get_structure(self.base)
+            ir_mesh_frac = self.mesh_frac[self.ir_grid_points]
+            for p, d in tqdm(zip(ir_mesh_frac, self.min_dists), desc = 'creating configurations'):
+                if d > self.r_min:
+                    framework = base.copy()
+                    framework.append(self.specie, coords = p)
+                    configurations.append(framework)
+        else:
+            raise ValueError(f"Wrong config_format {config_format}")
+        
+        if filename:
+            if config_format != 'ase':
+                raise ValueError('Only "ase" format is allowed for saving files')
+            if filename.split('.')[-1] == 'cfg':
+                write_cfg(filename, configurations)
+            else:
+                write(filename, configurations)
+        return configurations
+
+
+
+    def load_energies(self, energies):
+
+        """ 
+        Load energies obtained by any MLIP or structure-to-property model.
+        Should match order in created configurations.
+
+        Parameters
+        ----------
+        
+        energies: np.array
+            calculated energies for the created configurations
+        """
+
+        self.energies = energies
+        
+        ir_distribution = np.ones_like(self.min_dists) * np.inf
+        ir_distribution[self.min_dists > self.r_min] = self.energies
+        
+        self.distribution = np.nan_to_num(ir_distribution, copy = False, nan = np.inf)[self.mapping]
+        self.data = self.distribution.reshape(self.mesh_shape)
+
+
+    def read_processed_configurations(self, filename, file_format = 'xyz'):
+
+        """ 
+        Read processed (by any MLIP) atomic configurations
+        with calculated energies.
+
+        Parameters
+        ----------
+        
+        filename: string
+            path to the processed .xyz of .cfg file
+
+        file_format: string, 'xyz' by default
+            format of the file
+            if 'cfg' will use custom function read_cfg() to read MLIP 2.0 generated data
+            otherwise will use ASE's read()
+        """
+
+        if file_format == 'cfg':
+            atoms_list = read_cfg(filename)
+        else:
+            atoms_list = read(filename, index = ':')
+        energies = np.array([atoms.get_potential_energy() for atoms in atoms_list])
+        del atoms_list
+        self.load_energies(energies)
+
+
+
+    def write_cube(self, filename):
+
+        """
+        Write .cube file containing structural and MLIP distribution data.
+
+        Parameters
+        ----------
+
+        filename: str
+            file name to write .cube
+        """
+
+        data = self.data
+        nx, ny, nz = data.shape
+        with open(f'{filename}.cube', 'w') as f:
+            cube.write_cube(f, self.atoms, data = data[:nx-1, :ny-1, :nz-1])
+    
+
+
+    def write_grd(self, filename):
+        
+        """
+        Write MLIP distribution volumetric file for VESTA 3.0.
+
+        Parameters
+        ----------
+
+        filename: str
+            file name to write .grd
+        """
+
+        
+        data = self.data
+        voxels = data.shape[0] - 1, data.shape[1] - 1, data.shape[2] - 1
+        cellpars = self.cell.cellpar()
+        with open(f'{filename}.grd' , 'w') as report:
+            comment = '# MLIP data made with gridmlip package: https://github.com/dembart/gridmlip'
+            report.write(comment + '\n')
+            report.write(''.join(str(p) + ' ' for p in cellpars).strip() + '\n')
+            report.write(''.join(str(v) + ' ' for v in voxels).strip() + '\n')
+            for i in range(voxels[0]):
+                for j in range(voxels[1]):
+                    for k in range(voxels[2]):
+                        val = data[i, j, k]
+                        report.write(str(val) + '\n')
+
+
+
+    def percolation_barriers(self, encut = 10.0):
+        """
+        Calculate percolation barriers.
+
+        Parameters
+        ----------
+        encut: float, 10.0 by default
+            upper bound for searching the barrier
+        """
+        pl = Percolyze(self.data)
+        return pl.percolation_barriers(encut = encut)

{gridmlip-0.1 → gridmlip-0.2}/gridmlip/utils.py

@@ -24,7 +24,7 @@ def read_cfg(filename):
     configurations = []
     for i_s, i_e in tqdm(zip(np.where(np.array(text) == 'BEGIN_CFG\n')[0], np.where(np.array(text) == 'END_CFG\n')[0])):
         subtext = text[i_s:i_e]
-        grade = None
+        grade = None # not used
         results = {}
         for i, line in enumerate(subtext):
             if 'Size' in line:

{gridmlip-0.1 → gridmlip-0.2/gridmlip.egg-info}/PKG-INFO

@@ -1,12 +1,12 @@
 Metadata-Version: 2.1
 Name: gridmlip
-Version: 0.1
+Version: 0.2
 Summary: Grid-based method for calculating the percolation barrier of mobile species using machine learning interatomic potentials
 Home-page: https://github.com/dembart/gridmlip
 Author: Artem Dembitskiy
 Author-email: art.dembitskiy@gmail.com
 License: MIT
-Keywords: MLIP,percolation barrier
+Keywords: MLIP,percolation barrier,irreducible mesh
 Platform: UNKNOWN
 Classifier: Programming Language :: Python :: 3
 Classifier: License :: OSI Approved :: MIT License
@@ -31,7 +31,6 @@ License-File: LICENSE
 
 **gridmlip** is a library for calculating percolation barriers of mobile species in solids using grid-based method with machine learning interatomic potentials (MLIPs).
 
-
 ## Installation
 
 ```
@@ -53,7 +52,7 @@ Here we describe the pipeline in general. For a specific example, see [Notebooks
 from gridmlip import Grid
 
 atomic_types_mapper = {3:0, 31:1, 17:2}
-grig = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
+grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8, 
                     atomic_types_mapper=atomic_types_mapper # optional
                   )
 cfgs = grid.construct_configurations('data.cfg')
@@ -72,7 +71,7 @@ from gridmlip import Grid
 
 grid = Grid.from_file('your.cif', specie = 3, r_min = 1.8)
 grid.read_processed_configurations('processed_data.cfg', format = 'cfg')
-grid.percolation_barriers()
+barriers = grid.percolation_barriers()
 ```
 
 #### Step #4: Write .grd or .cube file for visualization in VESTA 3.0
@@ -84,7 +83,8 @@ g.write_grd('test.grd')
 ## Notebooks
 
 
-- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb) 
+- [Using TensorNet model pre-trained on MatPES](notebooks/TensorNet_MatPES.ipynb)    
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14gabZ-u19K-_I_e7-g67N8-xtPIN36Nw#sandboxMode=true&scrollTo=5TySp6C6UBoZ)
 
 
 ### How to cite

{gridmlip-0.1 → gridmlip-0.2}/gridmlip.egg-info/SOURCES.txt

@@ -5,6 +5,7 @@ pyproject.toml
 setup.cfg
 setup.py
 gridmlip/__init__.py
+gridmlip/_mesh.py
 gridmlip/_neighborhood.py
 gridmlip/_percolation_analysis.py
 gridmlip/cli.py

{gridmlip-0.1 → gridmlip-0.2}/setup.cfg

@@ -7,7 +7,7 @@ url = https://github.com/dembart/gridmlip
 description = gridmlip
 long_description = file: README.md
 long_description_content_type = text/markdown
-keywords = MLIP, percolation barrier
+keywords = MLIP, percolation barrier, irreducible mesh
 license = MIT
 classifiers = 
 	License :: OSI Approved :: BSD License

{gridmlip-0.1 → gridmlip-0.2}/setup.py

@@ -5,7 +5,7 @@ with open("README.md", "r") as fh:
 
 setuptools.setup(
      name='gridmlip',  
-     version='0.1',
+     version='0.2',
      py_modules = ["gridmlip"],
      install_requires = [
                          "pandas",
@@ -18,7 +18,7 @@ setuptools.setup(
      author="Artem Dembitskiy",
      author_email="art.dembitskiy@gmail.com",
      description="Grid-based method for calculating the percolation barrier of mobile species using machine learning interatomic potentials",
-     key_words = ['percolation-barrier', 'umlip', 'conductivity', 'diffusion'],
+     key_words = ['percolation-barrier', 'UMLIP', 'ionic conductivity', 'diffusion'],
      long_description=long_description,
      long_description_content_type="text/markdown",
      url="https://github.com/dembart/gridmlip",

gridmlip-0.1/gridmlip/core.py

@@ -1,274 +0,0 @@
-import numpy as np
-from tqdm import tqdm
-from ase.io import read, write, cube
-from .utils import read_cfg, write_cfg
-from ._neighborhood import nn_list
-from ._percolation_analysis import Percolyze
-
-
-
-__version__ = "0.1"
-
-
-
-class Grid:
-
-    def __init__(self, atoms, specie = None, resolution = 0.25,
-                 r_cut = 5.0, r_min = 1.5, atomic_types_mapper = None):
-
-        """ 
-        Initialization. 
-
-        Parameters
-        ----------
-        
-        atoms: ASE's Atoms object
-            atomic structure
-
-        specie: int
-            atomic number of the mobile specie
-
-        resolution: float, 0.2 by default
-            spacing between points (in Angstroms)
-
-        r_cut: float
-            cutoff radius to find the first nearest neighbor for each grid point
-
-        r_min: float
-            blocking sphere radius
-
-        atomic_types_mapper: dict, optional
-            mapper of atomic numbers into species used by MLIP
-        """
-
-        if atomic_types_mapper:
-            numbers = self._map_atomic_types(atomic_types_mapper, atoms.numbers)
-            atoms = atoms.copy()
-            atoms.numbers = numbers
-            specie = atomic_types_mapper[specie]
-        self.atoms = atoms.copy()
-        self.specie = specie
-        self.resolution = resolution
-        self.cell = self.atoms.cell
-        self._mesh(resolution)
-        self.base = self.atoms[self.atoms.numbers != self.specie]
-        self.min_dists, _ = nn_list(self.base.positions, self.mesh_cart, r_cut, self.cell)
-        self.r_cut = r_cut
-        self.r_min = r_min
-
-
-
-    def _map_atomic_types(self, atomic_types_mapper, numbers):
-        u,inv = np.unique(numbers,return_inverse = True)
-        return np.array([atomic_types_mapper[x] for x in u])[inv].reshape(numbers.shape)
-
-
-
-    def _mesh(self, resolution):
-        
-        """ 
-        This method creates grid of equidistant points in 3D
-        with respect to the input resolution. 
-
-        Parameters
-        ----------
-        resolution: float, 0.2 by default
-            spacing between points (in Angstroms)
-
-        Returns
-        -------
-        mesh_cart: np.array
-            cartesian coordinates of meshgrid
-        
-        """
-        
-        a, b, c, _, _, _ = self.cell.cellpar()
-        nx, ny, nz = int(a // resolution), int(b // resolution), int(c // resolution)
-        x = np.linspace(0, 1, nx) 
-        y = np.linspace(0, 1, ny) 
-        z = np.linspace(0, 1, nz)
-        self.mesh_frac = np.stack(np.meshgrid(x, y, z, indexing='ij'), axis=-1).reshape(-1, 3)
-        self.mesh_cart = self.cell.cartesian_positions(self.mesh_frac)
-        self.size = (nx, ny, nz)
-        return self.mesh_cart
-
-
-
-    @classmethod
-    def from_file(cls, file, specie = None, resolution = 0.25,
-                  r_cut = 5.0, r_min = 1.5, atomic_types_mapper = None):
-        """ 
-        Create Grid object from the file.
-
-        Parameters
-        ----------
-        
-        file: string
-            .xyz of .cfg file representing atomic structure
-
-        specie: int
-            atomic number of the mobile specie
-
-        resolution: float, 0.2 by default
-            spacing between points (in Angstroms)
-
-        r_cut: float
-            cutoff radius to find the first nearest neighbor for each grid point
-
-        r_min: float
-            blocking sphere radius
-
-        atomic_types_mapper: dict, optional
-            mapper of the species into atomic numbers
-
-        Returns
-        -------
-        Grid object
-        """
-
-        if file.split('.')[-1] == 'cfg':
-            atoms = read_cfg(file)[0]
-        else:
-            atoms = read(file)
-        return cls(atoms, specie, resolution=resolution,
-                   r_cut = r_cut, r_min = r_min, atomic_types_mapper =  atomic_types_mapper)
-
-
-
-    def construct_configurations(self, filename = None):
-        """ 
-        Construct atomic configurations for further calculations.
-
-        Parameters
-        ----------
-        
-        filename: string (Optional)
-            path to save .xyz of .cfg file with atomic configurations
-        
-        Returns
-        -------
-        configurations: list of ASE's atoms object
-        """
-        
-        configurations = []
-        for p, d in tqdm(zip(self.mesh_cart, self.min_dists), desc = 'creating configurations'):
-            if d > self.r_min:
-                framework = self.base.copy()
-                framework.append(self.specie)
-                framework.positions[-1] = p
-                configurations.append(framework)
-
-        if filename:
-            if filename.split('.')[-1] == 'cfg':
-                write_cfg(filename, configurations)
-            else:
-                write(filename, configurations)
-        return configurations
-
-
-
-    def load_energies(self, energies):
-
-        """ 
-        Load energies obtained by any MLIP or structure-to-property model.
-        Should match order in created configurations.
-
-        Parameters
-        ----------
-        
-        energies: np.array
-            calculated energies for the created configurations
-        """
-
-        self.energies = energies
-        self.distribution = np.ones_like(self.min_dists) * np.inf
-        self.distribution[self.min_dists > self.r_min] = self.energies
-        self.distribution = np.nan_to_num(self.distribution, copy = False, nan = np.inf)
-        self.data = self.distribution.reshape(self.size)
-
-
-    def read_processed_configurations(self, filename, format = 'xyz'):
-
-        """ 
-        Read processed (by any MLIP) atomic configurations
-        with calculated energies.
-
-        Parameters
-        ----------
-        
-        filename: string
-            path to the processed .xyz of .cfg file
-        """
-
-        if format == 'cfg':
-            atoms_list = read_cfg(filename)
-        else:
-            atoms_list = read(filename, index = ':')
-        self.energies = np.array([atoms.get_potential_energy() for atoms in atoms_list])
-        del atoms_list
-        self.distribution = np.ones_like(self.min_dists) * np.inf
-        self.distribution[self.min_dists > self.r_min] = self.energies
-        self.distribution = np.nan_to_num(self.distribution, copy = False, nan = np.inf)
-        self.data = self.distribution.reshape(self.size)
-
-
-
-    def write_cube(self, filename):
-
-        """
-        Write .cube file containing structural and MLIP distribution data.
-
-        Parameters
-        ----------
-
-        filename: str
-            file name to write .cube
-        """
-
-        data = self.data
-        nx, ny, nz = data.shape
-        with open(f'{filename}.cube', 'w') as f:
-            cube.write_cube(f, self.atoms, data = data[:nx-1, :ny-1, :nz-1])
-    
-
-
-    def write_grd(self, filename):
-        
-        """
-        Write MLIP distribution volumetric file for VESTA 3.0.
-
-        Parameters
-        ----------
-
-        filename: str
-            file name to write .grd
-        """
-
-        
-        data = self.data
-        voxels = data.shape[0] - 1, data.shape[1] - 1, data.shape[2] - 1
-        cellpars = self.cell.cellpar()
-        with open(f'{filename}.grd' , 'w') as report:
-            comment = '# MLIP data made with gridmlip package: https://github.com/dembart/gridmlip'
-            report.write(comment + '\n')
-            report.write(''.join(str(p) + ' ' for p in cellpars).strip() + '\n')
-            report.write(''.join(str(v) + ' ' for v in voxels).strip() + '\n')
-            for i in range(voxels[0]):
-                for j in range(voxels[1]):
-                    for k in range(voxels[2]):
-                        val = data[i, j, k]
-                        report.write(str(val) + '\n')
-
-
-
-    def percolation_barriers(self, encut = 10.0):
-        """
-        Calculate percolation barriers.
-
-        Parameters
-        ----------
-        encut: float, 10.0 by default
-            upper bound for searching the barrier
-        """
-        pl = Percolyze(self.data)
-        return pl.percolation_barriers(encut = encut)