GraphNeT 1.0.0 🎉

What's Changed

• Main kaggle comp by @RasmusOrsoe in #390
• Better selection in DatasetConfig by @asogaard in #391
• Dataset paths by @asogaard in #395
• added ability to overwrite distribution strategy by @RasmusOrsoe in #398
• Fix Docker image by @asogaard in #399
• Various minor fixes by @JostMigenda in #396
• Fix conversion of Upgrade I3-files by @asogaard in #407
• Move collate_fn to global scope by @asogaard in #408
• Model class fix by @RasmusOrsoe in #406
• bug fix by @RasmusOrsoe in #410
• removed typo by @RasmusOrsoe in #411
• predict fix for notebooks by @RasmusOrsoe in #414
• Bash script for training on multiple models by @MortenHolmRep in #426
• I3 Module Refactoring by @RasmusOrsoe in #416
• docstring update by @RasmusOrsoe in #427
• Updating documentation webpage by @asogaard in #409
• small fixes by @RasmusOrsoe in #429
• HLC/SLC and FADW & ATWD pulseflags by @RasmusOrsoe in #435
• EnsembleDataset by @RasmusOrsoe in #430
• Working base by @Aske-Rosted in #389
• Remove standard dataset cards by @asogaard in #437
• added model.inference() call by @RasmusOrsoe in #443
• Update logging by @asogaard in #436
• added max_table_count to merger by @RasmusOrsoe in #444
• Add discard_empty_events flag to the pulse cleaner module by @kayla-leonard in #449
• added fix by @RasmusOrsoe in #455
• add padding_value for pulseflag extraction by @RasmusOrsoe in #458
• added I3NTMuonLabelExtractor by @RasmusOrsoe in #459
• Fix error with batch_size=1 by @asogaard in #461
• Allow for accessing source code for lambda functions loaded from config by @asogaard in #463
• Add unit tests for graphnet.utilities.argparse module by @asogaard in #468
• added I3QUESOExtractor by @RasmusOrsoe in #469
• Remove len(data) assertion for SQLiteDataconverter save method by @RasmusOrsoe in #465
• Update READMEs in PISA examples folder by @asogaard in #471
• Restrict pytorch-lightning and pandas to < 2.0 until breaking changes are resolved by @asogaard in #476
• Remove sqlite_selection.py by @asogaard in #486
• Add a more thorough check for whether icetray is installed by @asogaard in #482
• Replace ConcatDataset with EnsembleDataset by @asogaard in #481
• Simplify examples by @RasmusOrsoe in #456
• Add files via upload by @RasmusOrsoe in #488
• fix typos in GETTING_STARTED.md by @JostMigenda in #489
• Add key property to Label add simplify Dataset.add_label by @asogaard in #485
• Update figure with missing connection line by @asogaard in #491
• Changed df.append to pd.concat to be compatible with pandas 2.0 by @AMHermansen in #490
• Load multiple parquet files by @The-Ludwig in #499
• Issue fix #496 by @axel-ponten in #498
• Multi-class classification example script by @MortenHolmRep in #494
• add more labels for extraction by @RasmusOrsoe in #501
• elasticity =/= inelasticity by @MoustHolmes in #502
• Update citations and README for v1.0 by @asogaard in #503

New Contributors

• @JostMigenda made their first contribution in #396
• @kayla-leonard made their first contribution in #449
• @AMHermansen made their first contribution in #490
• @The-Ludwig made their first contribution in #499
• @axel-ponten made their first contribution in #498
• @MoustHolmes made their first contribution in #502

Full Changelog: v0.2.4...v1.0.0

Tag for updated Docker image

What's Changed

• Implemented root dir variable for graphnet by @MortenHolmRep in #370
• Limit awkward-array version, due to breaking changes by @asogaard in #376
• Support macOS installation by @asogaard in #377
• Dataconverter fix by @RasmusOrsoe in #383
• Allow for selection=None in make_train_validation_dataloader by @asogaard in #384
• Restructure examples by @asogaard in #378
• Update Docker image by @asogaard in #387

Full Changelog: v0.2.3...v0.2.4

Tag for first analysis setup

What's Changed

• Changed i3featureextractor.py to also extract the event datetime from frame I3EventHeader. by @Peterandresen12 in #306
• Add feature to cluster by time windows by @mhaminh in #272
• 3d von mises fisher [draft] by @RasmusOrsoe in #308
• Bug-fix for dynedge Refactor by @RasmusOrsoe in #311
• Bugfix of create_table method by @RasmusOrsoe in #312
• Add relative DOM efficiency to unit test by @asogaard in #317
• Implementing model configuration and adding abstract base Model class by @asogaard in #309
• Re-compile JOSS paper draft on pushes to main rather than to some nam… by @asogaard in #318
• Add missing init file in graphnet.deployment by @asogaard in #319
• Add colorlog as required dependency by @asogaard in #321
• Update title of API docs and auto-detect type hints by @asogaard in #323
• Add a few options to Sphinx for better handling of type hints by @asogaard in #325
• Weight fitting refactor by @RasmusOrsoe in #313
• Update Parquet file indexing by @asogaard in #332
• Remove plots; to be implemented in graphnet-team/analyses by @asogaard in #333
• Update README.md by @MortenHolmRep in #334
• added i3pisaextractor by @RasmusOrsoe in #339
• Fix order of tables when writing to SQLite by @asogaard in #342
• Create citation.cff by @MortenHolmRep in #335
• Update citation file by @asogaard in #343
• Fix for I3PulseNoiseTruthFlag extractor issue by @RasmusOrsoe in #340
• Improve code quality and checking by @asogaard in #310
• added files by @RasmusOrsoe in #344
• Public test data by @asogaard in #345
• Fix Logger typing by @asogaard in #346
• Add configurable coarsening by @asogaard in #347
• Add dedicated Detector class for IceCube-86/main array by @asogaard in #348
• Update README.md by @RasmusOrsoe in #349
• Refactor configs and add Configurable class by @asogaard in #350
• Allow for querying entire column, rather than just one index at a time by @asogaard in #351
• Define final decorator once by @asogaard in #354
• Add DatasetConfig by @asogaard in #353
• Remove call to inference() that resulted in wrong validation loss values by @asogaard in #355
• Simplify training by @asogaard in #356
• Position reconstruction by @asogaard in #361
• Update direction reco example by @asogaard in #364
• Energy Task preventing over/under-flow by @Aske-Rosted in #352
• Enable multi-GPU training by @asogaard in #365
• Multi-class classification implementation by @MortenHolmRep in #329
• Remove obsolete unit test by @asogaard in #368
• Fix Build Icetray CI/CD job by @asogaard in #371
• SQLite utility methods by @asogaard in #366

New Contributors

• @Peterandresen12 made their first contribution in #306
• @Aske-Rosted made their first contribution in #352

Full Changelog: v0.2.2...v0.2.3

Tag for JOSS submission

What's Changed

• Update references by @asogaard in #291
• Coarsening cleanup by @RasmusOrsoe in #293
• Direction example by @RasmusOrsoe in #295
• I3extract modification by @MortenHolmRep in #296
• Update GCD-file finding by @asogaard in #299
• Move studies by @asogaard in #298
• Uniform weight fitter by @RasmusOrsoe in #279
• Implement I3GenericExtractor by @asogaard in #266
• Refactor modules by @asogaard in #304
• Update of README file with research/analysis ideas list by @troelspetersen in #292
• JOSS paper by @asogaard in #305
• New i3 extractors for high energy, small bug fixes by @RasmusOrsoe in #283
• Restructure models by @asogaard in #302
• Parquet to sqlite converter by @RasmusOrsoe in #277

New Contributors

• @troelspetersen made their first contribution in #292

Full Changelog: v0.2.1...v0.2.2

Tag for organisation transfer

What's Changed

• Parquet interface by @RasmusOrsoe in #217
• ParquetDataConverter and refactoring DataConverter by @asogaard in #215
• Upgrade astropy, uninstall PyYAML by @asogaard in #238
• Mitigate soft_unicode error in docs by limiting MarkupSafe version by @asogaard in #239
• Restrict docker to icecube/graphnet by @asogaard in #240
• Update installation instructions by @asogaard in #242
• added knn_graph_batch by @RasmusOrsoe in #236
• Pisa weight fitter by @RasmusOrsoe in #237
• Close SQLite connection after construction to avoid problem with multiprocessing by @asogaard in #243
• Weighted Loss by @RasmusOrsoe in #234
• Fix coarsening by @asogaard in #245
• Add default value of loss_weight by @asogaard in #246
• Optimise coarsening by @asogaard in #249
• Fix validate features by @RasmusOrsoe in #250
• Properly negate Data.x for min-pooling by @asogaard in #252
• Tweaks to improve experience with specified package versions by @asogaard in #261
• Reconstruct batch attributes by @RasmusOrsoe in #265
• Fix torch import in graphnet.data by @asogaard in #273
• Create W&B directory if it doesn't exist by @asogaard in #276

Full Changelog: v0.2.0...v0.2.1

Tag for DOI

What's Changed

• Track cascade indicies by @kaareendrup in #166
• Changing order of operations (transform/log10) by @BozianuLeon in #165
• Update BC task to fit new 2D regime by @kaareendrup in #167
• Tidyness by @RasmusOrsoe in #157
• Deploying upgrade models by @asogaard in #170
• Minor fixes for graphnet I3Module by @asogaard in #171
• two changes to i3 extractor by @RasmusOrsoe in #168
• Euclidean Graph Builder w/ weights by @BozianuLeon in #172
• Fix for neutrinos/noise in database creation by @BozianuLeon in #174
• Fixing build issue by @asogaard in #181
• New i3 converter/sqlite_dataconverter method by @BozianuLeon in #178
• Oscillation Study (1/3): InSQLitePipeline Class by @RasmusOrsoe in #179
• Pulse noise training by @RasmusOrsoe in #190
• Docstrings by @asogaard in #191
• Fixes from Tim by @asogaard in #192
• Filters In truth table, osc_weight in retro table by @BozianuLeon in #184
• Fixed call of get_predictions(...) by @timephy in #198
• DOM-wise coarsening by @asogaard in #196
• Pre-commits for black and flake8 by @asogaard in #200
• Making code base black and flake8 compliant by @asogaard in #202
• Data unit testing by @asogaard in #203
• Actions fix by @asogaard in #205
• CI/CD fixes by @asogaard in #206
• Specify torch version, which fixed #199 by @timephy in #210
• Logging by @asogaard in #207
• Oscillation Study (2/3): Pisa integration by @RasmusOrsoe in #180
• (3/3) oscillation study plots - new by @RasmusOrsoe in #219
• Container by @asogaard in #220
• Fix FilterMask bug by @BozianuLeon in #195
• PR new setup by @MortenHolmRep in #212
• Change trigger to give write permission from fork by @asogaard in #222
• Updating build-icetray action by @asogaard in #223
• Code for Marc Jacquart's analysis by @asogaard in #218

New Contributors

• @timephy made their first contribution in #198
• @MortenHolmRep made their first contribution in #212

Full Changelog: v0.1.2...v0.2.0

IceCube Upgrade functionality implemented

What's Changed

• converted labels from booleans to integers. by @RasmusOrsoe in #115
• Rename by @asogaard in #98
• Workflow by @asogaard in #119
• changed i3-file check to be compatible with lvl2 by @RasmusOrsoe in #117
• Double bang classification by @RasmusOrsoe in #118
• Upgrade functionality by @RasmusOrsoe in #116
• Clean up by @asogaard in #120
• upgrade hotfix by @RasmusOrsoe in #121
• Bug fix(es) by @asogaard in #123
• Use versioneer for automatically bookkeeping versions based on tags by @asogaard in #126
• Experiment tracking by @asogaard in #127
• Add dataset class to test stability against perturbations by @mhaminh in #142
• Train on Upgrade MC by @asogaard in #141
• mixed beans by @RasmusOrsoe in #145
• fixed float error for binary classification by @RasmusOrsoe in #147
• Implementation of Inverse Transform at Inference Step by @kaareendrup in #125
• Only commit badges on PRs by @asogaard in #149
• Radialgraphbuild by @BozianuLeon in #151
• functionality from paper branch to make upgrade sensitivity studies by @RasmusOrsoe in #154
• Upgrade noise pulsemap by @RasmusOrsoe in #153
• Stopmuon func in utils, add labels to i3extractor by @BozianuLeon in #148
• Allow for multiple task labels, e.g. for vertex reconstruction by @asogaard in #158
• Adding matplotlib to setup.py by @BozianuLeon in #160
• Upgrade reco by @asogaard in #155
• Ensure targets always have two dimensions by @asogaard in #163
• Installation and contributing guides by @asogaard in #162
• added noise upgrade study by @RasmusOrsoe in #161

New Contributors

• @mhaminh made their first contribution in #142

Full Changelog: v0.1.1...v0.1.2

Reference version for low-energy performance paper

What's Changed

• Added i3 extractor and general dataconverter by @RasmusOrsoe in #1
• Added Google-style Docstrings by @RasmusOrsoe in #13
• Configure package according to proposed directory structure by @asogaard in #12
• Setup workflow with unit tests and linting by @asogaard in #17
• Refactor SQLiteDataConverter and DataConverter by @asogaard in #18
• Add Sphinx automatic API documentation as part of build pipeline by @asogaard in #24
• Auto-add pylint and coverage badges to README by @asogaard in #27
• Restructure I3Extractor class to make it more modular and object-oriented by @asogaard in #30
• Add torch Dataset-inheriting class for SQLite "backend" by @asogaard in #32
• Update licence in setup file and add badge by @asogaard in #31
• Major changes by @RasmusOrsoe in #33
• Make additional installs (e.g. pytorch) user-specific by @asogaard in #36
• Build pipeline by @asogaard in #35
• Added trainer for multiple datasets (robustness test) by @RasmusOrsoe in #37
• Added persistent_workers as argument by @RasmusOrsoe in #39
• Minimal conda environments file by @asogaard in #38
• Robustness changes by @RasmusOrsoe in #50
• Modularising models by @asogaard in #40
• Update test_model_training_sqlite.py by @RasmusOrsoe in #66
• Upgrade MC by @asogaard in #67
• Reproducibility by @asogaard in #68
• Added binned resolution plot script by @RasmusOrsoe in #69
• Changed save_results to use .state_dict and created corresponding load_model function by @kaareendrup in #70
• Resolve duplications by @asogaard in #74
• Add convenience class methods for saving and loading models by @asogaard in #75
• Comparison study of new vs. legacy model implementations by @asogaard in #73
• Minor tweaks by @asogaard in #79
• Small updates to vonmisesfisher study training script by @RasmusOrsoe in #80
• Stable vmf by @asogaard in #82
• Fixed error related to new model saving/loading method and early stopping method by @RasmusOrsoe in #81
• Containerisation by @asogaard in #83
• Linting improvements by @asogaard in #86
• Adding BinaryClassification Task, BCELoss and GetEventNumbers in utils by @BozianuLeon in #92
• Implementing pytorch-lightning by @asogaard in #94
• Training on upgrade MC by @asogaard in #96
• Vertex reconstruction, add tasks and constant by @kaareendrup in #95

New Contributors

• @RasmusOrsoe made their first contribution in #1
• @asogaard made their first contribution in #12
• @kaareendrup made their first contribution in #70
• @BozianuLeon made their first contribution in #92

Full Changelog: https://github.com/icecube/graphnet/commits/v0.1.1