Skip to content

What's New

2024

  • [Lei, Sep 23, 2024] New release sfft v1.6.1: add PureCupy modules to directly handle GPU arrays as inputs/outputs. Most improvements were developed during the NASA GPU Hackathon 2024.

  • [Lei, Sep 17, 2024] We are developing SFFT in NASA GPU Hackathon 2024 as part of the pipeline optimization efforts for the Roman Supernova PIT team.

  • [Lei, Mar 9, 2024] New release sfft v1.5.0 that added B-Spline SFFT as a main function.

  • [Lei, Jan 9, 2024] New release sfft v1.4.2+ that can support Python 3.10!

2023

  • [Lei, Dec 4, 2023] New tutorial jupyter notebooks are available now, find them in test directories!

2022

  • [Lei, Nov 9, 2022] A warning for users: As scikit-image has changed something in its function of hough detection since version 0.19.0, I recently found that the source selection in sfft will be affected by this upgrade. For the time being I would recommend users to install 0.16.2 <= scikit-image <= 0.18.3. I may add a constrain on scikit-image version in sfft 1.3.5.

  • [Lei, Oct 25, 2022] A warning message about the usage of sfft.MultiEasySparsePacket and sfft.MultiEasyCrowdedPacket is added in the related test scripts.

  • [Lei, Aug 19, 2022] The preprocessing in sparse-flavor-sfft is refined using an additional rejection of mild varaibles since version 1.3.0.

  • [Lei, May 24, 2022] The sfft is now optimized for multiple tasks since version 1.1.0.

  • [Lei, May 24, 2022] A few argument-names have been changed since version sfft v1.1.0, please see the test scripts.

  • [Lei, May 24, 2022] Locking file is removed since sfft v1.1.0, as I found it unreliable in our tests, i.e., -GLockFile is removed.

  • [Lei, May 24, 2022] The trial subtraction for refinement is removed since sfft v1.1.0. However, I add a post-subtraction check to search anomalies on the difference image using the same logic. One can feed the coordinates of the anomalies to sfft again as Prior-Banned sources to refine the subtraction (see -XY_PriorBan in sfft.MultiEasySparsePacket).