IPKM/mdevaluate
6
1
Fork 0
Code Issues 1 Pull Requests Activity

added full_output option to non gaussian parameter so that correct statistics can be done afterwards

Browse Source
This commit is contained in:
robrobo
2025-07-17 18:45:11 +02:00
parent 0eff84910b
commit 0ffce2f17a
1 changed files with 24 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
src/mdevaluate/correlation.py
View File

@@ -18,7 +18,7 @@ def log_indices(first: int, last: int, num: int = 100) -> np.ndarray:
return np.unique(np.int_(ls) - 1 + first) return np.unique(np.int_(ls) - 1 + first)
@autosave_data(2) @autosave_data(nargs=2, kwargs_keys=('selector', 'segments', 'skip', 'window', 'average', 'points',), version=1.0)
def shifted_correlation( def shifted_correlation(
function: Callable, function: Callable,
frames: Coordinates, frames: Coordinates,
@@ -460,6 +460,7 @@ def non_gaussian_parameter(
end_frame: CoordinateFrame, end_frame: CoordinateFrame,
trajectory: Coordinates = None, trajectory: Coordinates = None,
axis: str = "all", axis: str = "all",
full_output = False,
) -> float: ) -> float:
r""" r"""
Calculate the non-Gaussian parameter. Calculate the non-Gaussian parameter.
@@ -472,30 +473,41 @@ def non_gaussian_parameter(
else: else:
vectors = displacements_without_drift(start_frame, end_frame, trajectory) vectors = displacements_without_drift(start_frame, end_frame, trajectory)
if axis == "all": if axis == "all":
r = (vectors**2).sum(axis=1) r2 = (vectors**2).sum(axis=1)
dimensions = 3 dimensions = 3
elif axis == "xy" or axis == "yx": elif axis == "xy" or axis == "yx":
r = (vectors[:, [0, 1]]**2).sum(axis=1) r2 = (vectors[:, [0, 1]]**2).sum(axis=1)
dimensions = 2 dimensions = 2
elif axis == "xz" or axis == "zx": elif axis == "xz" or axis == "zx":
r = (vectors[:, [0, 2]]**2).sum(axis=1) r2 = (vectors[:, [0, 2]]**2).sum(axis=1)
dimensions = 2 dimensions = 2
elif axis == "yz" or axis == "zy": elif axis == "yz" or axis == "zy":
r = (vectors[:, [1, 2]]**2).sum(axis=1) r2 = (vectors[:, [1, 2]]**2).sum(axis=1)
dimensions = 2 dimensions = 2
elif axis == "x": elif axis == "x":
r = vectors[:, 0] ** 2 r2 = vectors[:, 0] ** 2
dimensions = 1 dimensions = 1
elif axis == "y": elif axis == "y":
r = vectors[:, 1] ** 2 r2 = vectors[:, 1] ** 2
dimensions = 1 dimensions = 1
elif axis == "z": elif axis == "z":
r = vectors[:, 2] ** 2 r2 = vectors[:, 2] ** 2
dimensions = 1 dimensions = 1
else: else:
raise ValueError('Parameter axis has to be ether "all", "x", "y", or "z"!') raise ValueError('Parameter axis has to be ether "all", "x", "y", or "z"!')
mean_r = np.mean(r) m2 = np.mean(r2)
if mean_r == 0.0: m4 = np.mean(r2**2)
if m2 == 0.0:
if full_output:
return 0.0, 0.0, 0.0
else:
return 0.0 return 0.0
return (np.mean(r**2) / ((1 + 2 / dimensions) * (mean_r ** 2))) - 1
alpha_2 = (m4 / ((1 + 2 / dimensions) * m2**2)) - 1
if full_output:
return alpha_2, m2, m4
else:
return alpha_2