Welcome to dupin a Python package for detecting rare events in molecular simulations.
dupin is designed to provide an unopinionated Python API for partitioning temporal point cloud data into regions of stability and transition. Generally such data comes from molecular simulations or experimental imaging techniques. For example, if a researcher imaged gold nanoparticles nucleating into FCC crystal, dupin could help to partition the system into the initial fluid, transition, and crystal regions of the point cloud trajectory. Though, dupin attempts to be general and unopinionated we provide sensible defaults and do not sacrifice ease of use for customizability.
- Documentation: Read our API documentation.
- Installation Guide: Look at our installation guide.
- GitHub Repository: Check out the source code.
- Issue Tracker: File a bug report or pull request.
- HOOMD-blue: Molecular simulation engine
- freud: Molecular trajectory analysis
- ruptures: Change point decection library
- kneed: Elbow detection library
import dupin as du
import numpy as np
import ruptures as rpt
signal = np.load("signal.npy")
dynp = rpt.Dynp(custom_cost=du.detect.CostLinearFit())
detector = du.detect.SweepDetector(dynp, 8)
chps = detector.fit(signal)
This package was created with Cookiecutter based on a modified version of audreyr/cookiecutter-pypackage.