v0.241217

What's Changed

• Omit notebook subfolders in docs by @lrlunin in #549
• Fetch docker names without github token by @lrlunin in #555
• Add logo to readme and docs by @schuenke in #583
• Fix and modification for TrajType by @ckolbPTB in #584
• Move scaling to KTrajectory by @fzimmermann89 in #582
• Release v0.241217 by @schuenke in #586

Full Changelog: v0.241210...v0.241217

v0.241210

What's Changed

• [pre-commit] pre-commit autoupdate by @pre-commit-ci in #573
• Fix trajectory scaling in KTrajectoryPulseq by @schuenke in #551
• Add python 3.10 to badge and fix project keywords by @lrlunin in #550
• Memory efficient reshape by @fzimmermann89 in #557
• Release v0.241210 by @schuenke in #576

Full Changelog: v0.241126...v0.241210

v0.241126

What's Changed

• Select k-space data based on n_coils by @ckolbPTB in #309
• Fix formatting in warning string by @ckolbPTB in #514
• Add human readable ids to test by @fzimmermann89 in #485
• Add CartesianSamplingOp to FourierOp by @ckolbPTB in #482
• Exclude data outside of encoding_matrix by @ckolbPTB in #234
• Switch to adjoint as backward in FourierOP by @fzimmermann89 in #516
• Make randperm reproducible in tests by @fzimmermann89 in #517
• Fix operator norm device by @Stef-Martin in #520
• Add coil combination using PCA by @ckolbPTB in #470
• Fix typing issues by @fzimmermann89 in #515
• [RE-OPEN] Include .ipynb to sphinx-docs via myst-nb by @lrlunin in #524
• Add tests for autodiff of operators by @ckolbPTB in #405
• Implement Gram Shortcut for FourierOp by @fzimmermann89 in #503
• Remove MSEDataConsistency by @rkcatarina in #450
• Swap order of target and weight in functionals by @fzimmermann89 in #535
• Examples typos by @lrlunin in #449
• Fill range by @fzimmermann89 in #540
• Add apply (out of place) by @fzimmermann89 in #547
• Fix DCF and FourierOp edgecase bugs by @fzimmermann89 in #553
• Fix gram and adjoint of adjoint linear operator by @fzimmermann89 in #541
• Release v0.241126 by @fzimmermann89 in #561

New Contributors

• @lrlunin made their first contribution in #524
• @rkcatarina made their first contribution in #450

Full Changelog: v0.241112...v0.241126

v0.241112

What's Changed

• Fix unrolled CG autograd by @fzimmermann89 in #492
• Add matrix of LinearOperators by @fzimmermann89 in #432
• [pre-commit] pre-commit autoupdate by @pre-commit-ci in #498
• Add PCA-based compression operator by @fzimmermann89 in #181
• Fix CartesianSamplingOp by @ckolbPTB in #483
• Add reduce_view by @fzimmermann89 in #476
• Add apply_ to dataclasses by @fzimmermann89 in #505
• fix test installation from TestPyPi workflow by @schuenke in #499
• Remove check-docstring-first pre-commit hook by @fzimmermann89 in #508
• Revert NamedTemporaryFile ContextManager in example notebooks by @schuenke in #500
• Adapt KHeader parameters by @fzimmermann89 in #506
• Release v0.241112 by @fzimmermann89 in #510

New Contributors

• @pre-commit-ci made their first contribution in #498

Full Changelog: v0.241029...v0.241112

v0.241029

What's Changed

• Bump actions/github-script from 6 to 7 in the all-actions group by @dependabot in #448
• Sparsify documentation by @fzimmermann89 in #440
• Add L1Norm Functional by @fzimmermann89 in #397
• Add separable sum of proximable functions by @fzimmermann89 in #414
• Add ScaledFunctional implementing scalar*Functional by @fzimmermann89 in #411
• Add multiple axes unsqueeze and broadcasting to the right by @fzimmermann89 in #460
• Python 3.10 support by @fzimmermann89 in #461
• Add improper Rotations / Rotoflections by @fzimmermann89 in #441
• Hotfix Python3.10 support by @fzimmermann89 in #464
• Extend contributor guide by @fzimmermann89 in #458
• Add arithmetic to SpatialDimension by @fzimmermann89 in #417
• Add pre-commit check for all by @fzimmermann89 in #471
• Fix PyTest and Deployment workflow triggers by @schuenke in #467
• Remove superfluous .forward by @fzimmermann89 in #472
• Remove superfluous .forward by @ckolbPTB in #475
• Add RegularizedIterativeSENSEReconstruction by @ckolbPTB in #388
• Release 0.241029 by @fzimmermann89 in #491

Full Changelog: v0.241015...v0.241029

v0.241015

What's Changed

• description added by @ckolbPTB in #18
• Initial work on KData and ISMRMRD loading by @fzimmermann89 in #29
• fix names in from_ismrmrd_encodingLimitsType() by @schuenke in #31
• Make KData a frozen dataclass by @fzimmermann89 in #36
• Fix mypy ignore missing imports by @fzimmermann89 in #45
• Fix mypy #2 by @fzimmermann89 in #48
• Remove _get_shape from Trajectory base by @fzimmermann89 in #43
• Add torch, numpy to pre-commit/mypy venv by @fzimmermann89 in #49
• switch from autopep8 to black pre-commit hook by @schuenke in #55
• Add codespell to pre-commit by @fzimmermann89 in #56
• Remove 'clear' dictionary from codespell check by @fzimmermann89 in #57
• Ismrmrd test data by @ckolbPTB in #51
• switch CI pipeline from unittest to pytest by @schuenke in #59
• Initial IData and IHeader implementations by @fzimmermann89 in #30
• Update VSCode settings & recommended extensions by @schuenke in #60
• Pytest warnings by @schuenke in #66
• RPE trajectory by @ckolbPTB in #62
• Fix #38: Set trajectory_dimensions always to 3 by @fzimmermann89 in #75
• Refactor ktraj to a dataclass with kz,ky,kx & add DCF calculation by @fzimmermann89 in #76
• Change phantom to torch by @ckolbPTB in #77
• Sim coil sensitivities by @ckolbPTB in #79
• Fix #73: Let test pipeline fail if tests fail by @fzimmermann89 in #74
• Run tests in parallel by @schuenke in #89
• QData and QHeader implementation by @schote in #40
• CsmData with Walsh-Method implementation by @ckolbPTB in #83
• Fixing DCF logic bugs by @fzimmermann89 in #86
• hotfix vscode pytest-xdist issues (#87) by @fzimmermann89 in #101
• Documentation by @schote in #99
• Hotfix: Docu pipeline by @schote in #103
• Implementation of ktraj shaped like original raw data by @ckolbPTB in #107
• Implementation of KTrajectoryRadial2D by @schote in #105
• Implementation of Sensitivityop by @guastinimara in #110
• Testing by @schuenke in #108
• Ismrmrd traj by @ckolbPTB in #113
• Noise measurement data class and prewhitening algo added by @ckolbPTB in #114
• Remove oversampling and fft added by @ckolbPTB in #116
• KTrajectoryPulseq by @JoHa0811 in #111
• docu for users and devs extended by @ckolbPTB in #121
• Operator Template and Nufft by @fzimmermann89 in #104
• reading in of all dcm files from folder by @ckolbPTB in #124
• Reshape instead of view by @ckolbPTB in #127
• update supported python version to 3.10 & 3.11 by @schuenke in #133
• clean up by @ckolbPTB in #137
• Define trajectories to fit encoding_matrix and encoding_limits by @ckolbPTB in #140
• Cartesian trajectory by @ckolbPTB in #142
• fixed issue #141 by @SimoneHufnagel in #145
• Signal models added: Inversion and saturation recovery, MOLLI by @Stef-Martin in #129
• WASABI(TI) signal models by @schuenke in #130
• Fftshift for FourierOp by @ckolbPTB in #143
• Hotfix #143: zero_pad_or_crop broken for dim=None by @fzimmermann89 in #149
• Gpu support by @ckolbPTB in #136
• Hotfix for PyTorch 2.2 by @ckolbPTB in #152
• Disable cuda tests in VScode Test Explorer by @schuenke in #155
• Enable data which does not allow for a reshape into k2 and k1 by @ckolbPTB in #153
• Allow more than one other dim in KTrajectory by @fzimmermann89 in #160
• Bump maximum allowed python to 3.12 by @fzimmermann89 in #163
• Synthetic Sugar for Operators (needs python 3.11) by @fzimmermann89 in #128
• Fix ignore flags by @fzimmermann89 in #159
• Clean up code by @schuenke in #157
• Switch to Processes for voronoi cell volume calculation by @fzimmermann89 in #170
• Framework py -> ipynb -> docu by @ckolbPTB in #118
• Cartesian sampling operator by @ckolbPTB in #146
• Update README.md by @fzimmermann89 in #167
• HOTIFX: Binder to use main branch by @ckolbPTB in #176
• Limit xsd version by @fzimmermann89 in #177
• Cart sampling index int64 by @ckolbPTB in #182
• Lbfgs box by @schuenke in #131
• Same ndim for all AcqInfo fields by @ckolbPTB in #191
• Fix operator chaining by @fzimmermann89 in #189
• Return copy in remove readout os by @fzimmermann89 in #204
• Split data and select subset by @ckolbPTB in #117
• Switch to Ruff for linting and formatting by @schuenke in #193
• docu update by @ckolbPTB in #214
• Docker tests by @ckolbPTB in #175
• fix link by @ckolbPTB in #219
• move sensitivity operator tests to using dotproduct_adjointness_test by @fzimmermann89 in #220
• Generalized matrix operator by @fzimmermann89 in #212
• Conjugate Gradient by @koflera in #195
• Fix 221: Cartesian sampling op for random undersampling with varying pattern along other by @fzimmermann89 in #223
• Add separable nd-filters and sliding window by @fzimmermann89 in #161
• Clean up by @ckolbPTB in #218
• Support bipolar readout by @ckolbPTB in #222
• Fix #216 by @fzimmermann89 in #225
• Add from_array_zxy, zyx read-only property and tests to SpatialDimension by @fzimmermann89 in #238
• Move kernel in filters to input device. Fixes #239 by @fzimmermann89 in #240
• Rotation class by @fzimmermann89 in #209
• allow batch dimension in signal models by @fzimmermann89 in #226
• Detection tolerance for KTrajectoryPulseq by @schuenke in #232
• Restructure algorithms & Move to direct imports in src by @schuenke in #215
• Example script for pulseq radial 2D golden angle acquisition by @schuenke in #235
• Switch to direct import in einsumop by @fzimmermann89 in #246
• Read in multiple dicom files by @ckolbPTB in #248
• mono-exponential decay model added by @ckolbPTB in #245
• Add operator-tensor multiplication and type hints by @fzimmermann89 in #262
• Add .to/.cuda/cpu to KNoise by @fzimmermann89 in #263
• add rss to idata by @fzimmermann89 in #264
• switch in einsumop to einops notation by @fzimmermann89 in #265
• add as_operator to DcfData and CsmData by @fzimmermann89 in #266
• fix imports and add py.typed marker by @fzimmermann89 in #267
• Improve docstrings and variable names by @fzimmermann89 in #268
• Do not check notebooks for spelling mistakes by @ckolbPTB in #277
• Magnitude and Phase operator by @ckolbPTB in #273
• add fourierop.from_kdata by @fzimmermann89 in #269
• Allow multiple other dimensions in prewhiten data by @fzimmermann89 in #272
• add direct reconstruction by @fzimmermann89 in #270
• Bump pytorch and ismrmrd version by @fzimmermann89 ...

v0.241011

What's Included

• description added by @ckolbPTB in #18
• Initial work on KData and ISMRMRD loading by @fzimmermann89 in #29
• fix names in from_ismrmrd_encodingLimitsType() by @schuenke in #31
• Make KData a frozen dataclass by @fzimmermann89 in #36
• Fix mypy ignore missing imports by @fzimmermann89 in #45
• Fix mypy #2 by @fzimmermann89 in #48
• Remove _get_shape from Trajectory base by @fzimmermann89 in #43
• Add torch, numpy to pre-commit/mypy venv by @fzimmermann89 in #49
• switch from autopep8 to black pre-commit hook by @schuenke in #55
• Add codespell to pre-commit by @fzimmermann89 in #56
• Remove 'clear' dictionary from codespell check by @fzimmermann89 in #57
• Ismrmrd test data by @ckolbPTB in #51
• switch CI pipeline from unittest to pytest by @schuenke in #59
• Initial IData and IHeader implementations by @fzimmermann89 in #30
• Update VSCode settings & recommended extensions by @schuenke in #60
• Pytest warnings by @schuenke in #66
• RPE trajectory by @ckolbPTB in #62
• Fix #38: Set trajectory_dimensions always to 3 by @fzimmermann89 in #75
• Refactor ktraj to a dataclass with kz,ky,kx & add DCF calculation by @fzimmermann89 in #76
• Change phantom to torch by @ckolbPTB in #77
• Sim coil sensitivities by @ckolbPTB in #79
• Fix #73: Let test pipeline fail if tests fail by @fzimmermann89 in #74
• Run tests in parallel by @schuenke in #89
• QData and QHeader implementation by @schote in #40
• CsmData with Walsh-Method implementation by @ckolbPTB in #83
• Fixing DCF logic bugs by @fzimmermann89 in #86
• hotfix vscode pytest-xdist issues (#87) by @fzimmermann89 in #101
• Documentation by @schote in #99
• Hotfix: Docu pipeline by @schote in #103
• Implementation of ktraj shaped like original raw data by @ckolbPTB in #107
• Implementation of KTrajectoryRadial2D by @schote in #105
• Implementation of Sensitivityop by @guastinimara in #110
• Testing by @schuenke in #108
• Ismrmrd traj by @ckolbPTB in #113
• Noise measurement data class and prewhitening algo added by @ckolbPTB in #114
• Remove oversampling and fft added by @ckolbPTB in #116
• KTrajectoryPulseq by @JoHa0811 in #111
• docu for users and devs extended by @ckolbPTB in #121
• Operator Template and Nufft by @fzimmermann89 in #104
• reading in of all dcm files from folder by @ckolbPTB in #124
• Reshape instead of view by @ckolbPTB in #127
• update supported python version to 3.10 & 3.11 by @schuenke in #133
• clean up by @ckolbPTB in #137
• Define trajectories to fit encoding_matrix and encoding_limits by @ckolbPTB in #140
• Cartesian trajectory by @ckolbPTB in #142
• fixed issue #141 by @SimoneHufnagel in #145
• Signal models added: Inversion and saturation recovery, MOLLI by @Stef-Martin in #129
• WASABI(TI) signal models by @schuenke in #130
• Fftshift for FourierOp by @ckolbPTB in #143
• Hotfix #143: zero_pad_or_crop broken for dim=None by @fzimmermann89 in #149
• Gpu support by @ckolbPTB in #136
• Hotfix for PyTorch 2.2 by @ckolbPTB in #152
• Disable cuda tests in VScode Test Explorer by @schuenke in #155
• Enable data which does not allow for a reshape into k2 and k1 by @ckolbPTB in #153
• Allow more than one other dim in KTrajectory by @fzimmermann89 in #160
• Bump maximum allowed python to 3.12 by @fzimmermann89 in #163
• Synthetic Sugar for Operators (needs python 3.11) by @fzimmermann89 in #128
• Fix ignore flags by @fzimmermann89 in #159
• Clean up code by @schuenke in #157
• Switch to Processes for voronoi cell volume calculation by @fzimmermann89 in #170
• Framework py -> ipynb -> docu by @ckolbPTB in #118
• Cartesian sampling operator by @ckolbPTB in #146
• Update README.md by @fzimmermann89 in #167
• HOTIFX: Binder to use main branch by @ckolbPTB in #176
• Limit xsd version by @fzimmermann89 in #177
• Cart sampling index int64 by @ckolbPTB in #182
• Lbfgs box by @schuenke in #131
• Same ndim for all AcqInfo fields by @ckolbPTB in #191
• Fix operator chaining by @fzimmermann89 in #189
• Return copy in remove readout os by @fzimmermann89 in #204
• Split data and select subset by @ckolbPTB in #117
• Switch to Ruff for linting and formatting by @schuenke in #193
• docu update by @ckolbPTB in #214
• Docker tests by @ckolbPTB in #175
• fix link by @ckolbPTB in #219
• move sensitivity operator tests to using dotproduct_adjointness_test by @fzimmermann89 in #220
• Generalized matrix operator by @fzimmermann89 in #212
• Conjugate Gradient by @koflera in #195
• Fix 221: Cartesian sampling op for random undersampling with varying pattern along other by @fzimmermann89 in #223
• Add separable nd-filters and sliding window by @fzimmermann89 in #161
• Clean up by @ckolbPTB in #218
• Support bipolar readout by @ckolbPTB in #222
• Fix #216 by @fzimmermann89 in #225
• Add from_array_zxy, zyx read-only property and tests to SpatialDimension by @fzimmermann89 in #238
• Move kernel in filters to input device. Fixes #239 by @fzimmermann89 in #240
• Rotation class by @fzimmermann89 in #209
• allow batch dimension in signal models by @fzimmermann89 in #226
• Detection tolerance for KTrajectoryPulseq by @schuenke in #232
• Restructure algorithms & Move to direct imports in src by @schuenke in #215
• Example script for pulseq radial 2D golden angle acquisition by @schuenke in #235
• Switch to direct import in einsumop by @fzimmermann89 in #246
• Read in multiple dicom files by @ckolbPTB in #248
• mono-exponential decay model added by @ckolbPTB in #245
• Add operator-tensor multiplication and type hints by @fzimmermann89 in #262
• Add .to/.cuda/cpu to KNoise by @fzimmermann89 in #263
• add rss to idata by @fzimmermann89 in #264
• switch in einsumop to einops notation by @fzimmermann89 in #265
• add as_operator to DcfData and CsmData by @fzimmermann89 in #266
• fix imports and add py.typed marker by @fzimmermann89 in #267
• Improve docstrings and variable names by @fzimmermann89 in #268
• Do not check notebooks for spelling mistakes by @ckolbPTB in #277
• Magnitude and Phase operator by @ckolbPTB in #273
• add fourierop.from_kdata by @fzimmermann89 in #269
• Allow multiple other dimensions in prewhiten data by @fzimmermann89 in #272
• add direct reconstruction by @FZimm...

v0.241010

What's Changed

• description added by @ckolbPTB in #18
• Initial work on KData and ISMRMRD loading by @fzimmermann89 in #29
• fix names in from_ismrmrd_encodingLimitsType() by @schuenke in #31
• Make KData a frozen dataclass by @fzimmermann89 in #36
• Fix mypy ignore missing imports by @fzimmermann89 in #45
• Fix mypy #2 by @fzimmermann89 in #48
• Remove _get_shape from Trajectory base by @fzimmermann89 in #43
• Add torch, numpy to pre-commit/mypy venv by @fzimmermann89 in #49
• switch from autopep8 to black pre-commit hook by @schuenke in #55
• Add codespell to pre-commit by @fzimmermann89 in #56
• Remove 'clear' dictionary from codespell check by @fzimmermann89 in #57
• Ismrmrd test data by @ckolbPTB in #51
• switch CI pipeline from unittest to pytest by @schuenke in #59
• Initial IData and IHeader implementations by @fzimmermann89 in #30
• Update VSCode settings & recommended extensions by @schuenke in #60
• Pytest warnings by @schuenke in #66
• RPE trajectory by @ckolbPTB in #62
• Fix #38: Set trajectory_dimensions always to 3 by @fzimmermann89 in #75
• Refactor ktraj to a dataclass with kz,ky,kx & add DCF calculation by @fzimmermann89 in #76
• Change phantom to torch by @ckolbPTB in #77
• Sim coil sensitivities by @ckolbPTB in #79
• Fix #73: Let test pipeline fail if tests fail by @fzimmermann89 in #74
• Run tests in parallel by @schuenke in #89
• QData and QHeader implementation by @schote in #40
• CsmData with Walsh-Method implementation by @ckolbPTB in #83
• Fixing DCF logic bugs by @fzimmermann89 in #86
• hotfix vscode pytest-xdist issues (#87) by @fzimmermann89 in #101
• Documentation by @schote in #99
• Hotfix: Docu pipeline by @schote in #103
• Implementation of ktraj shaped like original raw data by @ckolbPTB in #107
• Implementation of KTrajectoryRadial2D by @schote in #105
• Implementation of Sensitivityop by @guastinimara in #110
• Testing by @schuenke in #108
• Ismrmrd traj by @ckolbPTB in #113
• Noise measurement data class and prewhitening algo added by @ckolbPTB in #114
• Remove oversampling and fft added by @ckolbPTB in #116
• KTrajectoryPulseq by @JoHa0811 in #111
• docu for users and devs extended by @ckolbPTB in #121
• Operator Template and Nufft by @fzimmermann89 in #104
• reading in of all dcm files from folder by @ckolbPTB in #124
• Reshape instead of view by @ckolbPTB in #127
• update supported python version to 3.10 & 3.11 by @schuenke in #133
• clean up by @ckolbPTB in #137
• Define trajectories to fit encoding_matrix and encoding_limits by @ckolbPTB in #140
• Cartesian trajectory by @ckolbPTB in #142
• fixed issue #141 by @SimoneHufnagel in #145
• Signal models added: Inversion and saturation recovery, MOLLI by @Stef-Martin in #129
• WASABI(TI) signal models by @schuenke in #130
• Fftshift for FourierOp by @ckolbPTB in #143
• Hotfix #143: zero_pad_or_crop broken for dim=None by @fzimmermann89 in #149
• Gpu support by @ckolbPTB in #136
• Hotfix for PyTorch 2.2 by @ckolbPTB in #152
• Disable cuda tests in VScode Test Explorer by @schuenke in #155
• Enable data which does not allow for a reshape into k2 and k1 by @ckolbPTB in #153
• Allow more than one other dim in KTrajectory by @fzimmermann89 in #160
• Bump maximum allowed python to 3.12 by @fzimmermann89 in #163
• Synthetic Sugar for Operators (needs python 3.11) by @fzimmermann89 in #128
• Fix ignore flags by @fzimmermann89 in #159
• Clean up code by @schuenke in #157
• Switch to Processes for voronoi cell volume calculation by @fzimmermann89 in #170
• Framework py -> ipynb -> docu by @ckolbPTB in #118
• Cartesian sampling operator by @ckolbPTB in #146
• Update README.md by @fzimmermann89 in #167
• HOTIFX: Binder to use main branch by @ckolbPTB in #176
• Limit xsd version by @fzimmermann89 in #177
• Cart sampling index int64 by @ckolbPTB in #182
• Lbfgs box by @schuenke in #131
• Same ndim for all AcqInfo fields by @ckolbPTB in #191
• Fix operator chaining by @fzimmermann89 in #189
• Return copy in remove readout os by @fzimmermann89 in #204
• Split data and select subset by @ckolbPTB in #117
• Switch to Ruff for linting and formatting by @schuenke in #193
• docu update by @ckolbPTB in #214
• Docker tests by @ckolbPTB in #175
• fix link by @ckolbPTB in #219
• move sensitivity operator tests to using dotproduct_adjointness_test by @fzimmermann89 in #220
• Generalized matrix operator by @fzimmermann89 in #212
• Conjugate Gradient by @koflera in #195
• Fix 221: Cartesian sampling op for random undersampling with varying pattern along other by @fzimmermann89 in #223
• Add separable nd-filters and sliding window by @fzimmermann89 in #161
• Clean up by @ckolbPTB in #218
• Support bipolar readout by @ckolbPTB in #222
• Fix #216 by @fzimmermann89 in #225
• Add from_array_zxy, zyx read-only property and tests to SpatialDimension by @fzimmermann89 in #238
• Move kernel in filters to input device. Fixes #239 by @fzimmermann89 in #240
• Rotation class by @fzimmermann89 in #209
• allow batch dimension in signal models by @fzimmermann89 in #226
• Detection tolerance for KTrajectoryPulseq by @schuenke in #232
• Restructure algorithms & Move to direct imports in src by @schuenke in #215
• Example script for pulseq radial 2D golden angle acquisition by @schuenke in #235
• Switch to direct import in einsumop by @fzimmermann89 in #246
• Read in multiple dicom files by @ckolbPTB in #248
• mono-exponential decay model added by @ckolbPTB in #245
• Add operator-tensor multiplication and type hints by @fzimmermann89 in #262
• Add .to/.cuda/cpu to KNoise by @fzimmermann89 in #263
• add rss to idata by @fzimmermann89 in #264
• switch in einsumop to einops notation by @fzimmermann89 in #265
• add as_operator to DcfData and CsmData by @fzimmermann89 in #266
• fix imports and add py.typed marker by @fzimmermann89 in #267
• Improve docstrings and variable names by @fzimmermann89 in #268
• Do not check notebooks for spelling mistakes by @ckolbPTB in #277
• Magnitude and Phase operator by @ckolbPTB in #273
• add fourierop.from_kdata by @fzimmermann89 in #269
• Allow multiple other dimensions in prewhiten data by @fzimmermann89 in #272
• add direct reconstruction by @fzimmermann89 in #270
• Bump pytorch and ismrmrd version by @fzimmermann89 ...