Daniel Nagel

Scientific software from my research. Please refer to my Github pages for more details.

A machine-learning method for efficient free-energy estimation under periodic boundary conditions. It trains a diffusion-style score model on the Fokker-Planck dynamics, delivering accurate estimates in regimes where conventional approaches struggle.

#python #jax #diffusion-model #machine-learning #free-energy #statistical-physics

msmhelper

#creator40.8k11

A time-series analysis toolkit for building and validating Markov state models of molecular dynamics. It offers fast, Numba-accelerated routines for clustering, dynamical coring, and pathway analysis of long trajectories.

#python #numba #markov-state-models #molecular-dynamics #data-analysis

dcTMD

#contributor11.4k17

An analysis framework for dissipation-corrected targeted molecular dynamics. It reconstructs free-energy profiles and friction factors along a pulled reaction coordinate from nonequilibrium pulling simulations.

#python #molecular-dynamics #free-energy #nonequilibrium #data-analysis

MoSAIC

#creator38.3k29

A correlation-based feature selection method that groups collective coordinates by their mutual dependencies. It uncovers the essential degrees of freedom in high-dimensional data fully unsupervised, without prior assumptions.

#python #feature-selection #unsupervised-learning #clustering #data-analysis

NorMI

#creator14k23

A generalized Kraskov estimator for the normalized mutual information of multidimensional data. It quantifies nonlinear dependencies robustly, staying accurate even in high dimensions and with limited samples.

#python #information-theory #mutual-information #statistics #data-analysis