v0.1.10

feat(io): add timeit func for psrfits io load
refactor(dedispersion): Eliminate spectrum type check for multi-parameter search optimization
perf(search): optimize parameter iteration and caching logic with pathlib and itertools.product

v0.1.9

refactor(frbdetector): remove Numba acceleration code to simplify dependencies
perf(search): Optimize spectrum data loading flow to reduce redundant IO and memory usage
docs: add todo
docs: update citation section with published paper link
fix(plotter): improve time precision by removing np.round in _calculate_spectrum_time_window
feat(plotter): Add candidate info saving and improve code structure - Added gencand property to TaskConfig for boolean configuration.

v0.1.8

fix(generate): Remove unused import and optimize string concatenation

- Remove unused import `back` from turtle module
- Optimize string concatenation in print statement for missed candidate files

v0.1.7

fix(pyproject): set astropy>=6.17 and scipy>=1.15.3 to support python 3.10+

v0.1.6

@lintian233 released this Oct 20, 2025
· 7 commits to refs/heads/main since this release

🧾 Changes

fix(rfi): Optimize RFI processing slice logic to ensure full coverage of all samples required for dedispersion.

refactor(plotter): Enhance frequency downsampling algorithm and improve spectral plotting stability.

perf(plotter): Optimize subband matrix calculation and integrate RFI & IQRM masking for faster rendering.

feat(specplot): Add baseline removal for integrated and subband spectra to improve contrast.

refactor(cpucal): Enhance dedispersion safety and add savetype support for flexible output control.

refactor(cli): Simplify monitor logic, add include_last config, and improve error handling and output.

feat(astroflow): Add minimum file-age filtering to avoid incomplete file processing.

v0.1.5

feat: Added GPU implementation of IQRM
io(psrfits): Accelerated memory access and array copying

v0.1.4

feat(core): Add CUDA performance timing instrumentation
fix(search): Pass maskfile parameter to dedispersion calls
perf(analysis): Optimize FRB SNR calculation window
feat(filterbank): add I/O timing and throughput metrics

v0.1.3

fix(rfi): Remove duplicate RFI marking initialization and calls

feat(dedispersion): Implement subband dedispersion with CUDA optimization #7

v0.1.2

feat: Add GPU-accelerated RFI marking and decouple CPU/GPU implementations

refactor: Separate RFI marker into dedicated CPU/GPU modules
fix: Correct RFI application timing in CUDA dedispersion pipeline
perf: Implement parallel CUDA kernels for RFI zeroing operations
chore: Clean up debug logs and improve parameter handling

v0.1.1

• Initial release of pulseflow.
• High-performance radio astronomy single pulse detection library.
• Includes support for YOLOv11n model.

Assets

• yolov11n_0816_v1.pt: YOLOv11n model weights for inference."