Added new function for calculating the occupation matrix

This commit is contained in:
2023-12-08 17:20:06 +01:00
parent 5267f6abac
commit c40ea052b8
2 changed files with 74 additions and 23 deletions

View File

@@ -9,10 +9,61 @@ import cmath
import pandas as pd
import multiprocessing as mp
VALID_GEOMETRY = {"cylindrical", "slab"}
def occupation_matrix(trajectory, edge_length=0.05, segments=1000, skip=0.1, nodes=8):
frame_indices = np.unique(
np.int_(np.linspace(len(trajectory) * skip, len(trajectory) - 1, num=segments))
)
box = trajectory[0].box
x_bins = np.arange(0, box[0][0] + edge_length, edge_length)
y_bins = np.arange(0, box[1][1] + edge_length, edge_length)
z_bins = np.arange(0, box[2][2] + edge_length, edge_length)
bins = [x_bins, y_bins, z_bins]
# Trajectory is split for parallel computing
size = math.ceil(len(frame_indices) / nodes)
indices = [
np.arange(len(frame_indices))[i : i + size]
for i in range(0, len(frame_indices), size)
]
pool = mp.Pool(nodes)
results = pool.map(
partial(_calc_histogram, trajectory=trajectory, bins=bins), indices
)
pool.close()
matbin = np.sum(results, axis=0)
x = (x_bins[:-1] + x_bins[1:]) / 2
y = (y_bins[:-1] + y_bins[1:]) / 2
z = (z_bins[:-1] + z_bins[1:]) / 2
coords = np.array(np.meshgrid(x, y, z, indexing="ij"))
coords = np.array([x.flatten() for x in coords])
matbin_new = matbin.flatten()
occupation_df = pd.DataFrame(
{"x": coords[0], "y": coords[1], "z": coords[2], "occupation": matbin_new}
)
occupation_df = occupation_df.query("occupation != 0")
return occupation_df
def _calc_histogram(numberlist, trajectory, bins):
matbin = None
for index in range(0, len(numberlist), 1000):
try:
indices = numberlist[index : index + 1000]
except IndexError:
indices = numberlist[index:]
frames = np.concatenate(np.array([trajectory.pbc[i] for i in indices]))
hist, _ = np.histogramdd(frames, bins=bins)
if matbin is None:
matbin = hist
else:
matbin += hist
return matbin
def get_fel(
traj,
path,