Today we are proud to release the first stable version of `sparse-ir`: a python library for optimal compression of many-body propagators on the imaginary (Euclidean) time axis as well as fast and stable diagrammatic computations.
Reasons to use IR basis functions and sparse sampling:
- The IR basis is a **provably optimal** basis for many-body propagators on the imaginary axis
- The IR basis comes with a **sparse, near-optimal set** of imaginary times and frequencies on which diagrammatic equations can be solved.
- The IR basis has an **intimate connection with the real-frequency** axis: it is a powerful preprocessor and preconditioner for analytic continuation.
Reasons to upgrade from the old `irbasis` library:
- sparse-ir computes bases for *arbitrary* cutoffs and kernels, usually within seconds
- sparse-ir provides battle-tested classes for sparse sampling, with fast and accurate fitting methods
- sparse-ir significantly improves upon the choice of sampling points, also allowing the use of symmetries
- sparse-ir packages objects for representing self-energies