The quantum database is a collection of open, standardized datasets for benchmarking numerical and machine learning techniques for quantum state reconstruction, tomography, and inference on various quantum condensed matter systems. We initiated this database in an effort to provide researchers in physics and machine learning with easily accessible data to benchmark their network architectures.
The following datasets are available now or in the near future:
- 2D Ising model spin configurations (classical)
- Z2 2D Ising Gauge theory bond configurations
- 1D Ising model with transverse field spin configurations (available soon)
- 2D Classical XY model (the
$O(2)$ rotor model) - 2D Spin 1/2 Quantum Heisenberg Model
The primary motivation for this database is to collect datasets that are interesting to study for the advancement of machine learning techniques ,but are also physically relevant models that can provide insight to our theoretic understanding of condensed matter models.
The datasets are structured such that both machine learning practitioners with little physics experience as well as physicists with little machine learning experience should be able to understand and use them.
Some of these datasets are straightforward, where others are more computationally intensive to generate. Not everyone has access to state-of-the-art numerical results, and having publically available datasets alleviates this issue. The computational methods involved in generating many of the datasets are non-trivial, and providing a high-quality, complete dataset saves time for researchers trying to get a start in the field.
Further contributions of datasets are encouraged! Please contact us if you have a dataset you wish to contribute.