Skip to main content

Gaussian Mixture Model clustering in size-and-shape space

Project description

GMM Positions

Overview

This is a package to perform Gaussian Mixture Model (GMM) clustering on particle positions (in ). Like other GMM schemes, the user must specify the number of clusters and a cluster initialization scheme (defaults to random). This is specified in the object initialization line, analagous to how it is done for the sklean GaussianMixture package. There are two choices for the form of the covariance but those are specified by calling different fit functions. See preprint (https://arxiv.org/abs/2112.11424) for additional details.

Installation

The package can be installed using pip

pip install shapeGMM

or downloaded and installed with

python setup.py install

Usage

This package is designed to mimic the usage of the sklearn package. You first initiliaze the object and then fit. Predict can be done once the model is fit. Fit and ppredict functions take particle position trajectories as input in the form of a (n_frames, n_atoms, 3) numpy array.

Initialize:

from shapeGMM import gmm_shapes

sgmm_object = gmm_shapes.ShapeGMM(n_clusters,verbose=True)

Fit:

Uniform (spherical, uncorrelated) covariance:

aligned_trajectory = sgmm_object.fit_uniform(training_set_positions)

Weighted (Kronecker product) covariance:

aligned_trajectory = sgmm_object.fit_weighted(training_set_positions)

Predict:

Uniform (spherical, uncorrelated) covariance:

clusters, aligned_traj, log_likelihood = sgmm_object.predict_uniform(full_trajectory_positions)

Weighted (Kronecker product) covariance:

clusters, aligned_traj, log_likelihood = sgmm_object.predict_weighted(full_trajectory_positions)

Description of Contents

Test Cases

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

shapeGMM-0.0.5.tar.gz (13.6 kB view hashes)

Uploaded Source

Built Distributions

shapeGMM-0.0.5-py3.8.egg (51.7 kB view hashes)

Uploaded Source

shapeGMM-0.0.5-py3-none-any.whl (24.3 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page