data/md_simulation/experimental.py

import argparse
import os
import random
from pathlib import Path
from typing import Union
import numpy as np
import torch
from ase import Atoms, units

from ase.calculators.calculator import Calculator
from ase.io import read
from ase.md import MDLogger

# torch.set_default_dtype(torch.float64)
from ase.md.nptberendsen import NPTBerendsen
from ase.md.velocitydistribution import MaxwellBoltzmannDistribution
from ase.optimize import FIRE

from checkpoint import multitask_from_checkpoint

from tqdm import tqdm
from Utils import ASEcalculator
from base import ForceRegressionTask
from ase.units import GPa      ## 1 GPa = 1 / 160.21766208 eV/ų.
from scipy.stats import linregress
import matplotlib.pyplot as plt
import pandas as pd



def get_density(atoms: Atoms) -> float:
    amu_to_grams = 1.66053906660e-24  # 1 amu = 1.66053906660e-24 grams
    angstrom_to_cm = 1e-8  # 1 Å = 1e-8 cm
    mass_amu = atoms.get_masses().sum()
    mass_g = (
        mass_amu * amu_to_grams
    )  # Get the volume of the atoms object in cubic angstroms (ų)
    volume_A3 = atoms.get_volume()
    volume_cm3 = volume_A3 * (angstrom_to_cm**3)  # 1 ų = 1e-24 cm³
    density = mass_g / volume_cm3

    return density

def elastic_tensor_calculation(atoms, calculator,filename):
  atoms.calc = calculator

  # Minimize the structure before stress calculations
  dyn = FIRE(atoms)
  dyn.run(fmax=0.05, steps=1000)  # Converge forces below 0.01 eV/Å

  # Define small strain range
  eps = 1e-4  # Maximum strain
  n_points = 20  # Number of points from -eps to +eps
  strain_values = np.linspace(-eps, eps, n_points)

  Cij = np.zeros((6, 6))  # Elastic tensor storage

  # Define strain matrices for Voigt notation (6 independent strains)
  strain_matrices = [
      [[1, 0, 0], [0, 0, 0], [0, 0, 0]],  # e_xx
      [[0, 0, 0], [0, 1, 0], [0, 0, 0]],  # e_yy
      [[0, 0, 0], [0, 0, 0], [0, 0, 1]],  # e_zz
      [[0, 0, 0], [0, 0, 0.5], [0, 0.5, 0]],  # e_yz
      [[0, 0, 0.5], [0, 0, 0], [0.5, 0, 0]],  # e_xz
      [[0, 0.5, 0], [0.5, 0, 0], [0, 0, 0]],  # e_xy
  ]

  # Labels for the strain components in Voigt notation
  voigt_labels = ['11', '22', '33', '23', '13', '12']

  # Compute reference stress
  ref_stress = atoms.get_stress(voigt=True)
  elastic_data=[]

  for i, strain_matrix in enumerate(strain_matrices):
      stresses = np.zeros((n_points, 6))
      intercepts=np.zeros((1,6))
      r_values=np.zeros((1,6))
      std_errs=np.zeros((1,6))

      for j, strain in enumerate(strain_values):
          strained_atoms = atoms.copy()
          deformation_matrix = np.eye(3) + strain * np.array(strain_matrix)
          strained_atoms.set_cell(atoms.cell @ deformation_matrix, scale_atoms=True)

          strained_atoms.calc = calculator
          dyn = FIRE(strained_atoms)
          dyn.run(fmax=0.05, steps=1000)  # Minimize structure
          stresses[j, :] = strained_atoms.get_stress(voigt=True) - ref_stress
          
          elastic_data.append([strain] + list(stresses[j, :]))

      # Perform linear regression to find the best slope
      for k in range(6):
          slope, intercept, r_value, p_value, std_err = linregress(strain_values, stresses[:, k])

          Cij[i, k] = slope
          intercepts[:,k]=intercept
          r_values[:,k]=r_value
          std_errs[:,k]=std_err


  # Convert the elastic tensor to GPa
  Cij_GPa = Cij / GPa # Convert the elastic tensor to GPa (Cij_GPa)
  
#   print("Elastic Stiffness Tensor (Cij) in GPa:")
#   print(np.array2string(Cij_GPa, precision=2, suppress_small=True,
#                          formatter={'float_kind': lambda x: f"{x:6.2f}"}))

# Save data as CSV using pandas
  columns = ["Strain"] + [f"Stress_{label}" for label in voigt_labels]
  df = pd.DataFrame(elastic_data, columns=columns)
  df.to_csv(filename, index=False)
  
  return Cij_GPa


def write_xyz(filepath: Union[str , Path], atoms: Atoms) -> None:
    """Writes ovito xyz file"""
    R = atoms.get_positions()
    species = atoms.get_atomic_numbers()
    cell = atoms.get_cell()

    with open(filepath, "w") as f:
        f.write(str(R.shape[0]) + "\n")
        flat_cell = cell.flatten()
        f.write(
            f'Lattice="{flat_cell[0]} {flat_cell[1]} {flat_cell[2]} {flat_cell[3]} {flat_cell[4]} {flat_cell[5]} {flat_cell[6]} {flat_cell[7]} {flat_cell[8]}" Properties=species:S:1:pos:R:3 Time=0.0'
        )
        for i in range(R.shape[0]):
            f.write(
                "\n"
                + str(species[i])
                + "\t"
                + str(R[i, 0])
                + "\t"
                + str(R[i, 1])
                + "\t"
                + str(R[i, 2])
            )


def replicate_system(atoms: Atoms, replicate_factors: np.ndarray) -> Atoms:
    """
    Replicates the given ASE Atoms object according to the specified replication factors.
    """
    nx, ny, nz = replicate_factors
    original_cell = atoms.get_cell()
    original_positions = (
        atoms.get_scaled_positions() @ original_cell
    )  # Scaled or Unscaled ?
    original_numbers = atoms.get_atomic_numbers()
    x_cell, y_cell, z_cell = original_cell[0], original_cell[1], original_cell[2]
    new_numbers = []
    for i in range(nx):
        for j in range(ny):
            for k in range(nz):
                new_numbers += [original_numbers]
    pos_after_x = np.concatenate([original_positions + i * x_cell for i in range(nx)])
    pos_after_y = np.concatenate([pos_after_x + i * y_cell for i in range(ny)])
    pos_after_z = np.concatenate([pos_after_y + i * z_cell for i in range(nz)])
    new_cell = [nx * original_cell[0], ny * original_cell[1], nz * original_cell[2]]
    new_atoms = Atoms(
        numbers=np.concatenate(new_numbers),
        positions=pos_after_z,
        cell=new_cell,
        pbc=atoms.get_pbc(),
    )
    return new_atoms


def symmetricize_replicate(curr_atoms: int, max_atoms: int, box_lengths: np.ndarray):
    replication = [1, 1, 1]
    atom_count = curr_atoms
    lengths = box_lengths
    while atom_count < (max_atoms // 2):
        direction = np.argmin(box_lengths)
        replication[direction] += 1
        lengths[direction] = box_lengths[direction] * replication[direction]
        atom_count = curr_atoms * replication[0] * replication[1] * replication[2]
    return replication, atom_count
    # Create new Atoms object


def minimize_structure(atoms: Atoms, fmax: float = 0.05, steps: int = 10) -> Atoms:
    """
    Perform energy minimization on the given ASE Atoms object using the FIRE optimizer.

    Parameters:
    atoms (ase.Atoms): The Atoms object to be minimized.
    fmax (float): The maximum force tolerance for the optimization (default: 0.01 eV/Å).
    steps (int): The maximum number of optimization steps (default: 1000).

    Returns:
    ase.Atoms: The minimized Atoms object.
    """
    dyn = FIRE(atoms, trajectory=None)
    dyn.run(fmax=fmax, steps=steps)
    return atoms


class TestArgs:
    runsteps = 50000
    model_name = "mace"  ##[mace, faenet, tensornet]
    model_path="/home/civil/phd/cez218288/scratch/Universal_potential/No_SAM/lightning_logs/version_0/checkpoints/epoch=99-step=109400.ckpt"
    timestep = 1.0
    results_dir="/home/civil/phd/cez218288/scratch/Universal_potential/Sim_output"
    input_dir="/home/civil/phd/cez218288/scratch/Universal_potential/Data/exp_1"
    device = "cuda"
    max_atoms = 200  # Replicate upto max_atoms (Min. will be max_atoms/2) (#Won't reduce if more than max_atoms)
    trajdump_interval = 10
    minimize_steps = 200
    thermo_interval = 10


config = TestArgs()


def run_simulation(
    calculator: Calculator,
    atoms: Atoms,
    pressure: float = 0.000101325,  # GPa
    temperature: float = 298,
    timestep: float = 0.1,
    steps: int = 10,
    SimDir: Union[str , Path]= Path.cwd(),
):
    # Define the temperature and pressure
    init_conf = atoms
    init_conf.set_calculator(calculator)
    # Initialize the NPT dynamics
    MaxwellBoltzmannDistribution(init_conf, temperature_K=temperature)

    dyn = NPTBerendsen(
        init_conf,
        timestep=timestep * units.fs,
        temperature_K=temperature,
        pressure_au=pressure * units.bar,
        compressibility_au=4.57e-5 / units.bar,
    )

    dyn.attach(
        MDLogger(
            dyn,
            init_conf,
            os.path.join(SimDir, "Simulation_thermo.log"),
            header=True,
            stress=True,
            peratom=False,
            mode="w",
        ),
        interval=config.thermo_interval,
    )

    density = []
    angles = []
    lattice_parameters = []

    def write_frame():
        dyn.atoms.write(
            os.path.join(SimDir, f"MD_{atoms.get_chemical_formula()}_NPT.xyz"),
            append=True,
        )

        cell = dyn.atoms.get_cell()

        lattice_parameters.append(cell.lengths())  # Get the lattice parameters
        angles.append(cell.angles())  # Get the angles
        density.append(get_density(atoms))

    dyn.attach(write_frame, interval=config.trajdump_interval)

    counter = 0
    for k in tqdm(range(steps), desc="Running dynamics integration.", total=steps):
        dyn.run(1)
        counter += 1

    density = np.array(density)
    angles = np.array(angles)
    lattice_parameters = np.array(lattice_parameters)

    # Calculate average values
    avg_density = np.mean(density)
    avg_angles = np.mean(angles, axis=0)
    avg_lattice_parameters = np.mean(lattice_parameters, axis=0)
    return avg_density, avg_angles, avg_lattice_parameters


def main(args, config):
    Loaded_model = ForceRegressionTask.load_from_checkpoint(config.model_path).to(config.device)
    calculator = ASEcalculator(Loaded_model, config.model_name)
    # calculator = MACECalculator(model_paths=config.model_path, device=config.device, default_dtype='float64')
    # calculator.model.double()  # Change model weights type to double precision (hack to avoid error)
    cif_files_dir = config.input_dir
    # output_file = config.out_dir
    import pandas as pd

    dirs = os.listdir(cif_files_dir)
    # for k in range(len(Dirs)):
    #     print(k,Dirs[k])
    folder = dirs[args.index]
    print("readong_folder number:", folder)

    # List to hold the data
    data = []
    folder_path = os.path.join(cif_files_dir, folder)

    if os.path.isdir(folder_path):
        for file in os.listdir(folder_path):
            file_path = os.path.join(folder_path, file)
            Temp, Press = file.split("_")[2:4]
            Temp, Press = float(Temp), float(Press)
            # TrajPath=os.path.join(config.traj_folder,"_".join(file.split("_")[:2])+'_Trajectory.xyz')

            # try:
            atoms = read(file_path)

            # Replicate_system
            replication_factors, size = symmetricize_replicate(
                len(atoms),
                max_atoms=config.max_atoms,
                box_lengths=atoms.get_cell_lengths_and_angles()[:3],
            )
            atoms = replicate_system(atoms, replication_factors)

            # Minimize the structure
            atoms.set_calculator(calculator)
            atoms = minimize_structure(atoms)

            # Calculate density and cell lengths and angles
            density = get_density(atoms)
            cell_lengths_and_angles = atoms.get_cell_lengths_and_angles().tolist()
            sim_dir = os.path.join(
                config.results_dir, f"{args.index}_Simulation_{file}"
            )
            print("SIMDIR:", sim_dir)
            elastic_file=os.path.join(sim_dir,f'elastic_plot_{file}.csv')
            os.makedirs(sim_dir, exist_ok=True)
            elastic_tensor=elastic_tensor_calculation(atoms,calculator, elastic_file)
            # Run the simulation
            avg_density, avg_angles, avg_lattice_parameters = run_simulation(
                calculator,
                atoms,
                pressure=Press,
                temperature=Temp,
                timestep=config.timestep,
                steps=config.runsteps,
                SimDir=sim_dir,
            )
            print(avg_density)
            # Append the results to the data list
            data.append(
                [file[:-4], density]
                + cell_lengths_and_angles
                + [avg_density]
                + avg_lattice_parameters.tolist()
                + avg_angles.tolist()
                + [elastic_tensor[i,j] for i in range(6) for j in range(6)]
            )
            # Create a DataFrame
            columns = [
                "Filename",
                "Exp_Density (g/cm³)",
                "Exp_a (Å)",
                "Exp_b (Å)",
                "Exp_c (Å)",
                "Exp_alpha (°)",
                "Exp_beta (°)",
                "Exp_gamma (°)",
                "Sim_Density (g/cm³)",
                "Sim_a (Å)",
                "Sim_b (Å)",
                "Sim_c (Å)",
                "Sim_alpha (°)",
                "Sim_beta (°)",
                "Sim_gamma (°)",
            ]+ [f"c{i+1}{j+1}" for i in range(6) for j in range(6)]
            df = pd.DataFrame(data, columns=columns)

            # Save the DataFrame to a CSV file
            df.to_csv(os.path.join(sim_dir, "Data.csv"), index=False)
            # print(f"Data saved to {output_file}")
            # except:
            #     print("filename", file_path)

    # Create a DataFrame
    # columns = ["Filename", "Exp_Density (g/cm³)", "Exp_a (Å)", "Exp_b (Å)", "Exp_c (Å)", "Exp_alpha (°)", "Exp_beta (°)", "Exp_gamma (°)"
    #            ,"Sim_Density (g/cm³)", "Sim_a (Å)", "Sim_b (Å)", "Sim_c (Å)", "Sim_alpha (°)", "Sim_beta (°)", "Sim_gamma (°)"]
    # df = pd.DataFrame(data, columns=columns)

    # # Save the DataFrame to a CSV file
    # df.to_csv(output_file, index=False)
    # print(f"Data saved to {output_file}")


if __name__ == "__main__":
    config = TestArgs()
    # Seed for the Python random module
    random.seed(123)
    np.random.seed(123)
    torch.manual_seed(123)
    if torch.cuda.is_available():
        torch.cuda.manual_seed(123)
        torch.cuda.manual_seed_all(123)  # if you are using multi-GPU.
    parser = argparse.ArgumentParser(description="Run MD simulation with MACE model")
    parser.add_argument("--index", type=int, default=0, help="index of folder")
    # parser.add_argument("--init_conf_path", type=str, default="example/lips20/data/test/botnet.xyz", help="Path to the initial configuration")
    # parser.add_argument("--device", type=str, default="cuda", help="Device: ['cpu', 'cuda']")
    # parser.add_argument("--input_dir", type=str, default="./", help="folder path")
    # parser.add_argument("--out_dir", type=str, default="out_dir_sl/neqip/lips20/exp.csv", help="Output path")
    # parser.add_argument("--results_dir", type=str, default="out_dir_sl/neqip/lips20/", help="Output  directory path")

    # parser.add_argument("--temp", type=float, default=300, help="Temperature in Kelvin")
    # parser.add_argument("--pressure", type=float, default=1, help="pressure in atm")
    # parser.add_argument("--timestep", type=float, default=1.0, help="Timestep in fs units")
    # parser.add_argument("--runsteps", type=int, default=1000, help="No. of steps to run")
    # parser.add_argument("--sys_name", type=str, default='System', help="System name")
    # parser.add_argument("--traj_folder", type=str, default="/home/civil/phd/cez218288/Benchmarking/MDBENCHGNN/mace_universal_2.0/EXP/Quartz/a.xyz")

    args = parser.parse_args()
    main(args, config)