UNIQORN -- The Universal Neural-network Interface for Quantum Observable Readout from N-body wavefunctions

UNIQORN is an implementation of machine learning tasks using the TensorFlow library. It's a GPLv3-licensed software currently hosted at gitlab.
The code performs regression and/or classification tasks on various observables and uses single-shot images -- the standard measurements of ultracold atomic systems -- as input data. An example on the regression of one-body and two-body densities from single-shot simulations obtained with MCTDH-X can be found here.

UNIQORN can regress quantum observables like density distributions or correlation functions, and more, at an accuracy beyond the commonly used averaging approaches and follows this workflow:

 

 

UNIQORN workflow

Hits: 4828