Version 1.0.1

Release Notes

• API changes of dependent libraries: ruamel.yaml
• Only import tensorflow if necessary
• Additional search paths in setup to find icecube headers

Full Changelog: v1.0.0...v1.0.1

Version 1.0.0

Release Notes:

• This release adds some more c++ utility for improved speed
• DNN data for complete detector can now be computed from within c++ as opposed to only per DOM
• Two new data formats: reduced_summary_statistics_data and total_dom_charge are added, which compute the input data entirely in c++. This allows to run networks with speeds of <1ms/event

Changes:

• only set data of container if writing to frame and necessary: improves runtime when not needed
• restructured code to use different calculation methods calculate_for_detector and calculate_per_dom. This allows to move the entire input data calculation to c++

Bug fixes:

• add support for older boost versions of py2

Added Functionality:

• added c++ function ext_boost.merge_pulses to merge pulses
• added c++ function ext_boost.combine_exclusions to combine exclusions
• Add option to directly fill DNN data container from within c++ via pybindings
• added data formats: reduced_summary_statistics_data and total_dom_charge

Version 0.2.0

Changes:

• Checking of pulse data settings is restructured: now allows for arbitrary settings

Added Functionality:

• DNN input data based on I3MCTree (for biased MuonGun simulations)
• Create bin exclusion lists for cascade-generator input data
• Added cascade-based DNN input data for cascade classification

Bug Fixes:

• Added compatibility with python3 and boost version in py3-v4.1.0 and py3-v4.0.1
• Only process DOMs with OMkeys.om <= 60
• Throw error when pulses are not in frame

Initial Release

Initial release of ic3-data