Initial commit: PS4-centered Li/I density analysis

.gitignore

@@ -0,0 +1,14 @@
+# Ignore everything by default.
+*
+
+# Keep Git control files.
+!.gitignore
+!README.org
+
+# Keep analysis scripts.
+!dump2cube.py
+!cube2mayavi.py
+
+# Keep selected example input/output files.
+!050Li3PS4-050LiI.lammpstrj
+!050Li3PS4-050LiI_PS4_I.cube

README.org

@@ -0,0 +1,194 @@
+#+TITLE: Li2S-P2S5-LiI Glass: PS4-Centered Ion Density Analysis
+#+AUTHOR: Minami Sakuma
+#+OPTIONS: toc:2 num:nil
+
+* Overview
+
+This repository contains Python scripts for analyzing and visualizing
+Li and I spatial probability distributions around =PS4^{3-}= units in
+Li2S-P2S5-LiI glass trajectories.
+
+The workflow consists of two steps:
+
+1. =dump2cube.py=
+   - Reads a LAMMPS trajectory.
+   - Aligns each =PS4^{3-}= tetrahedron using the nearest I^- ion.
+   - Accumulates Li and I positions in the aligned coordinate system.
+   - Outputs three-dimensional probability-density data in Gaussian cube format.
+
+2. =cube2mayavi.py=
+   - Reads the generated cube file.
+   - Visualizes the spatial probability density as an isosurface using Mayavi.
+   - Displays the reference =PS4^{3-}= tetrahedron.
+
+The scripts are intended to analyze the local geometrical relationship
+between I^- ions and =PS4^{3-}= units in Li2S-P2S5-LiI glasses.
+
+* Files
+
+| File | Description |
+|---+---|
+| =dump2cube.py= | Converts a LAMMPS trajectory into Li/I probability-density cube files. |
+| =cube2mayavi.py= | Visualizes a cube file with Mayavi. |
+| =050Li3PS4-050LiI.lammpstrj= | Example LAMMPS trajectory. |
+| =050Li3PS4-050LiI_PS4_I.cube= | Example I^- probability-density cube file. |
+
+* Requirements
+
+The scripts require Python 3 and the following packages:
+
+- NumPy
+- Mayavi
+- VTK
+- Traits
+- PyQt5 or PySide6, depending on the Mayavi installation
+
+Example installation using conda:
+
+#+begin_src bash
+conda create -n ps4-density python=3.10
+conda activate ps4-density
+conda install -c conda-forge numpy mayavi pyqt
+#+end_src
+
+Alternatively, NumPy can be installed using pip:
+
+#+begin_src bash
+pip install numpy
+#+end_src
+
+Mayavi installation is generally more stable with conda-forge.
+
+* Input Trajectory Format
+
+=dump2cube.py= assumes a LAMMPS trajectory containing the following
+atom columns:
+
+#+begin_example
+ITEM: ATOMS id type element mol x y z
+#+end_example
+
+The trajectory must contain at least the following elements:
+
+- Li
+- P
+- S
+- I
+
+The script assumes that P and S atoms belonging to the same
+=PS4^{3-}= unit share the same molecule ID (=mol=).
+
+* Analysis Procedure
+
+For each trajectory frame:
+
+1. Each P atom is selected as the center of a reference =PS4^{3-}= unit.
+2. The four nearest I^- ions are identified.
+3. The nearest I^- ion is used to define the orientation of the =PS4^{3-}= unit.
+4. The =PS4^{3-}= tetrahedron is rotated into a common reference frame.
+5. Li positions within the cutoff distance are accumulated.
+6. The positions of the four nearest I^- ions are accumulated.
+7. Three-dimensional histograms are written as cube files.
+
+The reference orientation is defined as follows:
+
+- The S atom farthest from the nearest I^- ion is aligned with the z axis.
+- A second S atom is used to fix the rotation around the z axis.
+
+* Usage
+
+** Generate Cube Files
+
+#+begin_src bash
+python dump2cube.py \
+  -i 050Li3PS4-050LiI.lammpstrj \
+  -m 160 160 160 \
+  -cut 8
+#+end_src
+
+Arguments:
+
+| Argument | Description |
+|---+---|
+| =-i=, =--trjfile= | Input LAMMPS trajectory file. |
+| =-m=, =--mesh= | Number of grid points in x, y, and z directions. |
+| =-cut=, =--cutoff= | Spatial cutoff radius in angstrom. |
+
+Expected output files:
+
+#+begin_example
+050Li3PS4-050LiI_PS4_Li.cube
+050Li3PS4-050LiI_PS4_I.cube
+#+end_example
+
+** Visualize I^- Probability Density
+
+#+begin_src bash
+python cube2mayavi.py \
+  -i 050Li3PS4-050LiI_PS4_I.cube \
+  -atom I \
+  -iso 1.26483e-09
+#+end_src
+
+** Visualize Li+ Probability Density
+
+#+begin_src bash
+python cube2mayavi.py \
+  -i 050Li3PS4-050LiI_PS4_Li.cube \
+  -atom Li \
+  -iso 1.0e-09
+#+end_src
+
+The appropriate isovalue depends on the trajectory length, mesh size,
+cutoff radius, and probability-density distribution. It should therefore
+be adjusted for each dataset.
+
+* Output
+
+The cube files contain:
+
+- A reference =PS4^{3-}= tetrahedron:
+  - P atom at the origin
+  - Four S atoms in the aligned coordinate system
+- A three-dimensional spatial probability-density field for Li or I
+
+The cube files can be visualized using:
+
+- Mayavi
+- VMD
+- PyMOL
+- ParaView
+- Other software supporting Gaussian cube files
+
+* Visualization Colors
+
+The default visualization settings in =cube2mayavi.py= are:
+
+| Object | Color |
+|---+---|
+| P | Purple |
+| S | Yellow |
+| Li probability density | Blue |
+| I probability density | Red |
+| P-S bonds | Gray |
+
+* Notes
+
+- The trajectory is treated using periodic boundary conditions.
+- Coordinates are converted to fractional coordinates before alignment.
+- The current implementation assumes an orthorhombic simulation box for
+  the trajectory parsing procedure.
+- The script includes a triclinic-cell lattice conversion function, but
+  the exact input format should be checked before applying it to
+  triclinic LAMMPS trajectories.
+- Large trajectory and cube files can exceed the standard GitHub file-size
+  limit. Git LFS is recommended when files are larger than 100 MB.
+
+* Citation
+
+If this repository is used in research, please cite the corresponding
+publication or presentation describing the Li2S-P2S5-LiI glass analysis.
+
+* License
+
+This repository is intended for academic research use.

cube2mayavi.py

@@ -0,0 +1,246 @@
+#!/usr/bin/env python
+# ./cube2mayavi.py -i 050Li3PS4-050LiI_PS4_I.cube -atom I -iso 1.26483e-09
+import numpy as np
+from mayavi import mlab
+ang_borr = 0.5291772109217
+
+
+def read_cube(filename):
+    """
+    .cubeファイルを読み込み、原子座標と3次元ボクセルデータを返す関数
+    """
+    with open(filename, 'r') as f:
+        # ヘッダーの読み飛ばし（最初の2行はコメント）
+        f.readline()
+        f.readline()
+
+        # 3行目: 原子の数 と 原点座標
+        line = f.readline().split()
+        natoms = abs(int(line[0]))  # 原子の数
+        origin = np.array([float(x) for x in line[1:4]])  # 原点
+
+        # 4-6行目: グリッドの分割数(N)とベクトル(X, Y, Z軸)
+        line = f.readline().split()
+        nx, x_vec = int(line[0]), np.array([float(x) for x in line[1:4]])
+        line = f.readline().split()
+        ny, y_vec = int(line[0]), np.array([float(x) for x in line[1:4]])
+        line = f.readline().split()
+        nz, z_vec = int(line[0]), np.array([float(x) for x in line[1:4]])
+
+        # 原子座標の読み込み
+        atoms = []
+        for _ in range(natoms):
+            line = f.readline().split()
+            atoms.append([float(x) for x in line[2:5]])
+        atoms = np.array(atoms)
+
+        # ボクセルデータの読み込み
+        data = []
+        for line in f:
+            data.extend([float(x) for x in line.split()])
+
+    vol_data = np.array(data).reshape(nx, ny, nz)
+
+    return vol_data, atoms, origin, (x_vec, y_vec, z_vec)
+
+
+def draw_radial_scale(max_dist=10.0, step=1.0, axis='x'):
+    """
+    原点から動径方向に目盛り（定規）を描画する
+    """
+    if axis == 'y':
+        vec = np.array([0, 1, 0])
+        tick_dir = np.array([0.2, 0, 0])
+    elif axis == 'z':
+        vec = np.array([0, 0, 1])
+        tick_dir = np.array([0.2, 0, 0])
+    else:  # default x
+        vec = np.array([1, 0, 0])
+        tick_dir = np.array([0, 0, 0.2])
+
+    # メインの直線を引く
+    end_point = vec * max_dist
+    mlab.plot3d([0, end_point[0]], [0, end_point[1]], [0, end_point[2]],
+                tube_radius=0.02, color=(0, 0, 0))
+
+    # 目盛りと数字を配置
+    for r in np.arange(step, max_dist + step, step):
+        pos = vec * r
+        t_start = pos - tick_dir
+        t_end = pos + tick_dir
+
+        mlab.plot3d([t_start[0], t_end[0]],
+                    [t_start[1], t_end[1]],
+                    [t_start[2], t_end[2]],
+                    tube_radius=0.02, color=(0, 0, 0))
+
+        text_pos = pos + tick_dir * 1.5
+        mlab.text3d(text_pos[0], text_pos[1], text_pos[2],
+                    f"{int(r)}", scale=0.25, color=(0, 0, 0))
+
+
+def change_view(azimuth=30, elevation=60):
+    mlab.view(azimuth=azimuth, elevation=elevation)
+
+
+def Li_visualize(filename, iso_val=None, radius_cutoff=5.0):
+    print(f"Reading {filename}...")
+    vol_data, atoms, origin, vectors = read_cube(filename)
+    atoms, origin_ang = atoms * ang_borr, origin * ang_borr
+
+    dx = np.linalg.norm(vectors[0]) * ang_borr
+    dy = np.linalg.norm(vectors[1]) * ang_borr
+    dz = np.linalg.norm(vectors[2]) * ang_borr
+
+    nx, ny, nz = vol_data.shape
+    i, j, k = np.mgrid[0:nx, 0:ny, 0:nz]
+
+    X = origin_ang[0] + i * dx
+    Y = origin_ang[1] + j * dy
+    Z = origin_ang[2] + k * dz
+    R = np.sqrt(X**2 + Y**2 + Z**2)
+
+    vol_inner = np.where(R <= 5, vol_data, 0.0)
+    vol_outer = np.where((R < 9) & (R > 5), vol_data, 0.0)
+
+    mlab.figure(bgcolor=(1, 1, 1), size=(800, 600))
+
+    if args.iso_val is None:
+        iso_val = np.std(vol_data) * 2.0
+    else:
+        iso_val = args.iso_val
+
+    print(f"Plotting isosurface at value: +/- {iso_val:.4f}")
+    print("max_value: ", np.max(vol_data))
+    print("min_value: ", np.min(vol_data))
+
+    # --- 確率密度 内側 ---
+    src_in = mlab.pipeline.scalar_field(vol_inner)
+    src_in.spacing = [dx, dy, dz]
+    src_in.origin = origin_ang
+    obj_in = mlab.pipeline.iso_surface(src_in, contours=[iso_val],
+                                       opacity=0.4, color=(0, 0.2, 0.8))
+
+    # --- 確率密度 外側 ---
+    src_out = mlab.pipeline.scalar_field(vol_outer)
+    src_out.spacing = [dx, dy, dz]
+    src_out.origin = origin_ang
+    obj_out = mlab.pipeline.iso_surface(src_out, contours=[iso_val],
+                                        opacity=0.2, color=(0.7, 0.1, 0.1))
+
+    # --- 原子の描画 ---
+    mlab.points3d(atoms[0, 0], atoms[0, 1], atoms[0, 2],
+                  scale_factor=0.8, color=(0.576, 0.439, 0.8), resolution=20)
+    mlab.points3d(atoms[1:, 0], atoms[1:, 1], atoms[1:, 2],
+                  scale_factor=0.8, color=(1, 1, 0), resolution=20)
+
+    # --- 結合の描画 ---
+    p_coord = atoms[0]
+    s_coords = atoms[1:]
+    for s_coord in s_coords:
+        mlab.plot3d([p_coord[0], s_coord[0]],
+                    [p_coord[1], s_coord[1]],
+                    [p_coord[2], s_coord[2]],
+                    tube_radius=0.1, color=(0.6, 0.6, 0.6), opacity=1.0)
+
+    mlab.points3d(0, 0, 0, scale_factor=6.6,
+                  mode='sphere', color=(0, 0, 0), opacity=0.05, resolution=50)
+    mlab.points3d(0, 0, 0, scale_factor=14,
+                  mode='sphere', color=(0, 0, 0), opacity=0.05, resolution=50)
+
+    mlab.axes()
+    return obj_in, obj_out
+
+
+def I_visualize(filename, iso_val=None, cutoff=8):
+    """
+    1つのIのcubeファイルを読み込んで可視化する
+    """
+    print(f"Reading {filename}...")
+    vol_data, atoms, origin, vectors = read_cube(filename)
+    atoms, origin_ang = atoms * ang_borr, origin * ang_borr
+
+    dx = np.linalg.norm(vectors[0]) * ang_borr
+    dy = np.linalg.norm(vectors[1]) * ang_borr
+    dz = np.linalg.norm(vectors[2]) * ang_borr
+
+    mlab.figure(bgcolor=(1, 1, 1), size=(800, 600))
+
+    if iso_val is None:
+        if 'args' in globals() and hasattr(args, 'iso_val') and args.iso_val is not None:
+            iso_val = args.iso_val
+        else:
+            iso_val = np.std(vol_data) * 2.0
+
+    print(f"Plotting isosurface at value: +/- {iso_val:.4f}")
+
+    # --- カットオフによるデータの球状マスク処理 ---
+    if cutoff is not None:
+        nx, ny, nz = vol_data.shape
+        x = origin_ang[0] + np.arange(nx) * dx
+        y = origin_ang[1] + np.arange(ny) * dy
+        z = origin_ang[2] + np.arange(nz) * dz
+        X, Y, Z = np.meshgrid(x, y, z, indexing='ij')
+        R = np.sqrt(X**2 + Y**2 + Z**2)
+        vol_data[R > cutoff] = 0.0
+
+    print("max_value: ", np.max(vol_data))
+    print("min_value: ", np.min(vol_data))
+
+    # --- 確率密度 ---
+    src_in = mlab.pipeline.scalar_field(vol_data)
+    src_in.spacing = [dx, dy, dz]
+    src_in.origin = origin_ang
+
+    # 赤色で表示
+    obj = mlab.pipeline.iso_surface(src_in, contours=[iso_val], opacity=0.3,
+                                    color=(0.850, 0.058, 0.058))
+
+    # --- 原子の描画 ---
+    mlab.points3d(atoms[0, 0], atoms[0, 1], atoms[0, 2],
+                  scale_factor=0.8, color=(0.576, 0.439, 0.8), resolution=20)
+    mlab.points3d(atoms[1:, 0], atoms[1:, 1], atoms[1:, 2],
+                  scale_factor=0.8, color=(1, 1, 0), resolution=20)
+
+    # --- 結合の描画 ---
+    p_coord = atoms[0]
+    s_coords = atoms[1:]
+    for s_coord in s_coords:
+        mlab.plot3d([p_coord[0], s_coord[0]],
+                    [p_coord[1], s_coord[1]],
+                    [p_coord[2], s_coord[2]],
+                    tube_radius=0.1, color=(0.6, 0.6, 0.6), opacity=1.0)
+
+    mlab.points3d(0, 0, 0, scale_factor=11,
+                  mode='sphere', color=(0, 0, 0), opacity=0.05, resolution=50)
+
+    return obj
+
+
+# --- 実行部分 ---
+if __name__ == "__main__":
+
+    import argparse
+
+    description = """This is a test program"""
+    par = argparse.ArgumentParser(description=description)
+
+    par.add_argument('-i', '--infiles', default="", required=True, nargs="+",
+                     help='input file')
+    par.add_argument('-atom', '--atom', default="", required=True,
+                     choices=['Li', 'I'],
+                     help='atom')
+    par.add_argument('-iso', '--iso_val', required=False, type=float,
+                     help='しきい値')
+    args = par.parse_args()
+
+    if args.atom == "Li":
+        obj_in, obj_out = Li_visualize(args.infiles[0])
+        mlab.show()
+    elif args.atom == "I":
+        if len(args.infiles) > 1:
+            print("Warning: Multiple files were passed, but only the first one will be read.")
+        obj = I_visualize(args.infiles[0])
+        mlab.show()
+    else:
+        print("atom error")

dump2cube.py

@@ -0,0 +1,300 @@
+#!/usr/bin/env python
+# ./dump2cube.py -i 050Li3PS4-050LiI.lammpstrj -m 160 160 160 -cut 8
+import numpy as np
+import argparse
+import re
+import os
+import time
+
+description = """This is a test program"""
+par = argparse.ArgumentParser(description=description)
+
+par.add_argument('-i', '--trjfile', default="", required=True,
+                 help='input trjfile')
+par.add_argument('-m', '--mesh', default=(30, 30, 30), required=False,
+                 nargs=3, type=int, help='mesh grid')
+par.add_argument('-cut', '--cutoff', default=7, required=False,
+                 type=float, help='mesh grid')
+
+args = par.parse_args()
+ang_borr = 0.5291772109217
+dirname = os.path.dirname(__file__)
+
+
+class LammpsTrj():
+    def __init__(self):
+        """
+        trjファイルを読み込んで、原子数、lattice、ステップ数を格納
+        """
+        with open(args.trjfile) as o:
+            data = o.read().split()
+        self.atoms = int(data[7])  # 原子数
+        data = np.array(data).reshape(-1, self.atoms*7+32)
+        self.lattices = data[:, 17:23].astype(float)  # (step, 6)
+        self.data = data[:, 32:]  # 座標データ
+        self.steps = data.shape[0]
+        self.mesh = args.mesh
+        self.cutoff = args.cutoff
+
+    def setLattice(self, lat):
+        """
+        latticeの形によって操作を分岐
+        self.M, self.M_を作成
+        """
+        if lat.shape[0] == 6:
+            M = np.array([[lat[1]-lat[0], 0, 0],
+                          [0, lat[3]-lat[2], 0],
+                          [0, 0, lat[5]-lat[4]]]).astype(float)
+        if lat.shape[0] == 9:
+            xlo_bound, xhi_bound, xy = lat[0, 0], lat[0, 1], lat[0, 2]
+            ylo_bound, yhi_bound, xz = lat[1, 0], lat[1, 1], lat[1, 2]
+            zlo_bound, zhi_bound, yz = lat[2, 0], lat[2, 1], lat[2, 2]
+            xlo = xlo_bound - np.min([0.0, xy, xz, xy+xz])
+            xhi = xhi_bound - np.max([0.0, xy, xz, xy+xz])
+            ylo = ylo_bound - np.min([0.0, yz])
+            yhi = yhi_bound - np.max([0.0, yz])
+            zlo = zlo_bound
+            zhi = zhi_bound
+            lx = xhi - xlo
+            ly = yhi - ylo
+            lz = zhi - zlo
+            a = lx
+            b = np.sqrt(ly**2 + xy**2)
+            c = np.sqrt(lz**2 + xz**2 + yz**2)
+            alpha = np.arccos((xy*xz + ly*yz)/b/c)
+            beta = np.arccos(xz/c)
+            gamma = np.arccos(xy/b)
+            v1 = [a, 0, 0]
+            v2 = [b*np.cos(gamma), b*np.sin(gamma), 0]
+            v3 = [c*np.cos(beta),
+                  c*(np.cos(alpha)-np.cos(beta)*np.cos(gamma))/np.sin(gamma),
+                  c*np.sqrt(1+2*np.cos(alpha)*np.cos(beta)*np.cos(gamma)
+                            - np.cos(alpha)**2-np.cos(beta)**2
+                            - np.cos(gamma)**2 / np.sin(gamma))]
+            M = np.array([v1, v2, v3])
+        return M
+
+    def getOnestep(self, step):
+        """
+        引数のstepにおける座標を、原子ごとにself.elems(辞書)に格納する
+        self.elemsの座標は分率座標
+        """
+        step_data = self.data[step, :].reshape(self.atoms, -1)  # (atoms, 6)
+        self.M = self.setLattice(self.lattices[step])
+        self.M_ = np.linalg.inv(self.M)
+        self.elems = {}
+        for e in np.unique(step_data[:, 2]):
+            d = step_data[step_data[:, 2] == e, 3:].astype(float)
+            dxyz = d[:, 0:3]
+            dxyz = dxyz @ self.M_
+            dxyz = dxyz - np.floor(dxyz)
+            d[:, 0:3] = dxyz
+            self.elems[e] = d  # 元素ごとに結晶座標とmol番号を格納
+        return self.elems
+
+    def makeCube(self, trjfile):
+        """
+        すべてのP原子について処理を行う。
+        P原子ごとに最も近い4つのI原子を選出し、回転・アライメントを行う。
+        """
+        # Li用の座標リスト
+        self.li_coords_list = []
+
+        # I用のリスト (全てまとめて格納)
+        self.i_coords_list = []
+
+        # ヒストグラム結果格納用
+        self.li_hist = None
+        self.i_hist = None
+
+        count_p = 0  # 処理したP原子の総数カウント用
+
+        for step in range(self.steps):
+            print(f"processing {step} step")
+            self.elems = self.getOnestep(step)
+            elems_ = {k: v.copy() for k, v in self.elems.items()}  # copy
+
+            # 全てのP原子に対してループ処理を行う
+            for p_data in elems_["P"]:
+
+                # 全てのI原子の中から、このP原子に最も近い4つのI原子を特定する
+                i_diffs = elems_["I"][:, 0:3] - p_data[0:3]
+                i_diffs = i_diffs - np.around(i_diffs)  # 周期境界条件の考慮(分率)
+                i_diffs_abs = i_diffs @ self.M         # 絶対座標へ変換
+                distances = np.linalg.norm(i_diffs_abs, axis=1)
+
+                sorted_indices = np.argsort(distances)
+                closest_Is = elems_["I"][sorted_indices[:4]]  # 近い順に4つ抽出
+
+                nearest_I_coords = closest_Is[0]  # 回転定義に使用する「最も近いI」
+
+                # S原子の処理 (同じmol番号のSを取得)
+                s_data = elems_["S"][elems_["S"][:, 3] == p_data[3]]
+
+                # 1つのPS4に注目して、Pを原点に配置するための距離計算
+                si_diff = s_data[:, 0:3] - nearest_I_coords[0:3]
+                si_diff = si_diff - np.around(si_diff)
+                si_diff = np.linalg.norm(si_diff, axis=1)
+
+                # 最も遠いSをz軸上に配置するため、そのインデックスを取得
+                n_S_idx = np.argmax(si_diff)
+                # その他のインデックスを取得
+                s_indices = [0, 1, 2, 3]
+                s_indices.remove(n_S_idx)
+                n2_S_idx = s_indices[0]
+
+                s_xyz = s_data[:, 0:3] - p_data[0:3]
+                s_xyz = s_xyz - np.round(s_xyz)
+                s_xyz = s_xyz @ self.M  # 絶対座標へ
+
+                # 1つのSをyz平面上に配置(z軸周り回転)
+                theta = np.arctan2(s_xyz[n_S_idx][0], s_xyz[n_S_idx][1])
+                self.Mat_z = np.array([[np.cos(-theta), np.sin(-theta), 0],
+                                       [-np.sin(-theta), np.cos(-theta), 0],
+                                       [0, 0, 1]])
+                s_xyz = (self.Mat_z @ s_xyz.T).T
+
+                # 1つのSをz軸上に配置(x軸周り回転)
+                theta2 = np.arctan2(s_xyz[n_S_idx][1], s_xyz[n_S_idx][2])
+                self.Mat_x = np.array([[1, 0, 0],
+                                       [0, np.cos(-theta2), np.sin(-theta2)],
+                                       [0, -np.sin(-theta2), np.cos(-theta2)]])
+                s_xyz = (self.Mat_x @ s_xyz.T).T
+
+                # もう1つのSをyz平面上に配置(z軸周り回転)
+                theta3 = np.arctan2(s_xyz[n2_S_idx][0], s_xyz[n2_S_idx][1])
+                self.Mat_z2 = np.array([[np.cos(-theta3), np.sin(-theta3), 0],
+                                        [-np.sin(-theta3), np.cos(-theta3), 0],
+                                        [0, 0, 1]])
+                s_xyz = (self.Mat_z2 @ s_xyz.T).T
+                self.s_xyz = s_xyz
+
+                # PS4座標保存 (Output用)
+                p_xyz = np.array([0, 0, 0])
+                self.ps4_coord = np.vstack((p_xyz, s_xyz))
+                self.ps4_coord = self.ps4_coord / ang_borr
+
+                # 座標回転関数の定義
+                def rotate_coords(coords_fractional):
+                    if len(coords_fractional) == 0:
+                        return np.array([])
+                    rot_xyz = coords_fractional[:, 0:3] - p_data[0:3]
+                    rot_xyz = rot_xyz - np.round(rot_xyz)
+                    rot_xyz = rot_xyz @ self.M
+                    rot_xyz = (self.Mat_z @ rot_xyz.T).T
+                    rot_xyz = (self.Mat_x @ rot_xyz.T).T
+                    rot_xyz = (self.Mat_z2 @ rot_xyz.T).T
+                    return rot_xyz
+
+                # ヒストグラム範囲
+                bounds = [[-(self.cutoff)/ang_borr,
+                           (self.cutoff)/ang_borr]] * 3
+
+                # --- Liの処理: 座標をリストに追加 ---
+                segment_xyz_Li = []
+                for xyz in self.elems["Li"]:
+                    diff = xyz[0:3] - p_data[0:3]
+                    diff = diff - np.around(diff)
+                    diff = diff @ self.M
+                    diff = np.linalg.norm(diff)
+                    for xyzdata in xyz[diff < self.cutoff, :]:
+                        segment_xyz_Li.append(xyzdata)
+                segment_xyz_Li = np.array(segment_xyz_Li, dtype=float)
+
+                if len(segment_xyz_Li) > 0:
+                    rotationed_Li = rotate_coords(segment_xyz_Li)
+                    self.li_coords_list.extend(rotationed_Li.tolist())
+
+                # --- Iの処理: 抽出した4つのIを回転させてまとめてリストに追加 ---
+                rotationed_Is = rotate_coords(closest_Is)
+                self.i_coords_list.extend(rotationed_Is.tolist())
+
+                count_p += 1
+
+        # ループ終了後にまとめてヒストグラム計算
+        print("Calculating Histograms...")
+        volume = (self.cutoff*2)**3
+        bounds = [[-(self.cutoff)/ang_borr, (self.cutoff)/ang_borr]] * 3
+
+        # Li
+        if len(self.li_coords_list) > 0:
+            print("num_Li: ", len(self.li_coords_list))
+            li_arr = np.array(self.li_coords_list) / ang_borr
+            li_hist, _ = np.histogramdd(li_arr, bins=self.mesh, range=bounds)
+            self.li_hist = li_hist.ravel() / len(self.li_coords_list) / volume
+            print("sum_Li: ", np.sum(self.li_hist) * volume)
+
+        # I (All 4 atoms combined)
+        if len(self.i_coords_list) > 0:
+            print("num_I: ", len(self.i_coords_list))
+            i_arr = np.array(self.i_coords_list) / ang_borr
+            i_hist, _ = np.histogramdd(i_arr, bins=self.mesh, range=bounds)
+            self.i_hist = i_hist.ravel() / len(self.i_coords_list) / volume
+            print("sum_I: ", np.sum(self.i_hist) * volume)
+
+    def outputCube(self):
+        output_start_time = time.time()
+        base = re.match(r"(\d{3}.*?LiI).*?",
+                        args.trjfile.split("/")[-1]).group(1)
+        self.atomsDic = {'I': '53', 'Li': '3', 'P': '15', 'S': '16'}
+
+        if self.li_hist is None and self.i_hist is None:
+            print("No data found.")
+            return
+
+        def get_header():
+            body = f"created from {__file__}, {args}\n"
+            body += "Contains the selected quantity on a FFT grid\n"
+            origin = [-(self.cutoff)/ang_borr] * 3
+            body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                5, *origin)
+            body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                self.mesh[0], (self.cutoff*2)/self.mesh[0]/ang_borr, 0, 0)
+            body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                self.mesh[1], 0, (self.cutoff*2)/self.mesh[1]/ang_borr, 0)
+            body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                self.mesh[2], 0, 0, (self.cutoff*2)/self.mesh[2]/ang_borr)
+            # Pの座標
+            body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                int(self.atomsDic["P"]), float(self.atomsDic["P"]), *self.ps4_coord[0])
+            # Sの座標
+            for s_coord in self.ps4_coord[1:5]:
+                body += "{:>5d}{:>12.7f}{:>12.7f}{:>12.7f}{:>12.7f}\n".format(
+                    int(self.atomsDic["S"]), float(self.atomsDic["S"]), *s_coord)
+            return body
+
+        # Liの出力
+        if self.li_hist is not None:
+            body = get_header()
+            for idx, r in enumerate(self.li_hist):
+                if idx % 6 == 5:
+                    body += "{:>13.5E}\n".format(r)
+                else:
+                    body += "{:>13.5E}".format(r)
+
+            outfile = f"{dirname}/{base}_PS4_Li.cube"
+            with open(outfile, "w") as o:
+                o.write(body)
+            print(f"{outfile} was created.")
+
+        # Iの出力 (まとめて1ファイルに出力)
+        if self.i_hist is not None:
+            body = get_header()
+            for idx, r in enumerate(self.i_hist):
+                if idx % 6 == 5:
+                    body += "{:>13.5E}\n".format(r)
+                else:
+                    body += "{:>13.5E}".format(r)
+
+            outfile = f"{dirname}/{base}_PS4_I.cube"
+            with open(outfile, "w") as o:
+                o.write(body)
+            print(f"{outfile} was created.")
+
+        output_end_time = time.time()
+        print(f"output_time : {output_end_time - output_start_time} s")
+
+
+trj = LammpsTrj()
+trj.makeCube(args.trjfile)
+trj.outputCube()