DARTsort 0.5.28__tar.gz → 0.5.30__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {dartsort-0.5.28 → dartsort-0.5.30}/.github/workflows/ci.yml +2 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/PKG-INFO +4 -4
- {dartsort-0.5.28 → dartsort-0.5.30}/environment.yml +1 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/pyproject.toml +3 -4
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/agglomerate.py +14 -4
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/cluster_util.py +3 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/clustering.py +1 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/mixture.py +2 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/refine_util.py +14 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/config.py +8 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/analysis.py +2 -1
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/hybridkit.py +1 -1
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/noise_recording_tools.py +2 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/localize/localize_util.py +1 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/peel_base.py +4 -39
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/subtract.py +10 -12
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/subtract_util.py +0 -28
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/_multichan_denoiser_kit.py +1 -1
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/mixture_classifier.py +38 -71
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/reduction.py +0 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/data_util.py +8 -6
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/drift_util.py +1 -1
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/internal_config.py +9 -5
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/multiprocessing_util.py +2 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/preprocess_util.py +3 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/registration_util.py +22 -14
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/waveform_util.py +1 -1
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/mixture.py +11 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/unit.py +18 -4
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/PKG-INFO +4 -4
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/requires.txt +2 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_clustering.py +7 -3
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_drift_util.py +23 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_simkit.py +7 -2
- {dartsort-0.5.28 → dartsort-0.5.30}/.github/workflows/ci_windows.yml +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/.github/workflows/deploy.yml +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/.github/workflows/docs.yml +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/.gitignore +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/README.md +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/docs/index.md +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/docs/main_api.md +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/docs/toolkit_api.md +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/notebook/.keep +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/notebook/fit_and_batch_matching.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/notebook/ibl_dandi_example.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/setup.cfg +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/__main__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/cli.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/clustering_features.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/density.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/forward_backward.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/kmeans.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/clustering/merge.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/detect/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/detect/detect.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/comparison.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/hybrid_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/sim_template_tools.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/simkit.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/evaluate/simlib.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/localize/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/localize/localize_torch.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/main.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/grab.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/compressed_upsampled.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/drifty.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/matching_base.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/pairwise.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/matching_util/pairwise_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/peel_lib.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/reduction_template.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/shaver.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/peel/threshold.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/default_temporal_kernel.npy +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/detect_np1.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/detect_np2.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/glasbey1024.npz +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/offset_trained_denoiser.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/offset_trained_denoiser_added_temp.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/offset_trained_denoiser_added_temp_boost.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/original_single_chan_denoiser.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/pretrained/single_chan_denoiser.pt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/get_templates.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/postprocess_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/realignment.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/template_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/templates.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/templates/templib.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/all_transformers.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/amortized_localization.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/amplitudes.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/decollider.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/denoising_scorer.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/enforce_decrease.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/fixed_prop.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/interp.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/localize.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/matched_filter_net.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/matching_denoiser.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/pipeline.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/single_channel_denoiser.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/supervised_denoiser.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/temporal_pca.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/transform_base.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/vq_matched_filter.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/transform/whiten.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/cli_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/interpolation_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/job_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/logging_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/main_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/more_operators.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/motion.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/nn_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/noise_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/peel_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/py_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/residual_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/sparse_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/spikeio.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/spiketorch.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/testing_util/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/testing_util/matching_debug_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/testing_util/mixture_testing_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/torch_optimization_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/util/torch_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/__init__.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/analysis_plots.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/colors.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/gt.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/layout.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/metrics_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/recanim.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/scatterplots.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/sorting.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/unit_comparison.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/versus.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/vismain.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort/vis/waveforms.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/SOURCES.txt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/dependency_links.txt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/entry_points.txt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/src/dartsort.egg-info/top_level.txt +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/conftest.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_agglomerate.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_alignment.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_cluster_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_clustering_features.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_config.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_dartsort.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_data_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_density.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_detect.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_drifty_matching.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_enforce_decrease.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_grab_and_featurize.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_hybridkit.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_kmeans.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_linalg.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_localize_torch.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_matching.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_mixture_classifier.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_modeling.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_spiketorch.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_subtract.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_templates.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_threshold.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_torch_optimization_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_transform.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/tests/test_waveform_util.py +0 -0
- {dartsort-0.5.28 → dartsort-0.5.30}/zensical.toml +0 -0
|
@@ -19,14 +19,14 @@ jobs:
|
|
|
19
19
|
|
|
20
20
|
steps:
|
|
21
21
|
- name: Checkout
|
|
22
|
-
uses: actions/checkout@
|
|
22
|
+
uses: actions/checkout@v7
|
|
23
23
|
with:
|
|
24
24
|
path: dartsort
|
|
25
25
|
|
|
26
26
|
- name: Fetch tags and describe
|
|
27
27
|
run: pushd dartsort/ && git fetch --prune --unshallow --tags && git describe && popd
|
|
28
28
|
|
|
29
|
-
- uses: actions/setup-python@
|
|
29
|
+
- uses: actions/setup-python@v7
|
|
30
30
|
with:
|
|
31
31
|
python-version: "3.x"
|
|
32
32
|
cache: "pip" # caching pip dependencies
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: dartsort
|
|
3
|
-
Version: 0.5.
|
|
3
|
+
Version: 0.5.30
|
|
4
4
|
Summary: DARTsort
|
|
5
5
|
Author-email: Charlie Windolf <ciw2048@gmail.com>
|
|
6
6
|
Project-URL: Homepage, https://github.com/cwindolf/dartsort
|
|
@@ -8,12 +8,11 @@ Project-URL: Bug Tracker, https://github.com/cwindolf/dartsort/issues
|
|
|
8
8
|
Classifier: Programming Language :: Python :: 3
|
|
9
9
|
Classifier: License :: OSI Approved :: MIT License
|
|
10
10
|
Classifier: Operating System :: OS Independent
|
|
11
|
-
Requires-Python: >=3.
|
|
11
|
+
Requires-Python: >=3.12
|
|
12
12
|
Description-Content-Type: text/markdown
|
|
13
|
-
Requires-Dist: dredge-ephys
|
|
13
|
+
Requires-Dist: dredge-ephys>=0.4.0
|
|
14
14
|
Requires-Dist: h5py
|
|
15
15
|
Requires-Dist: jaxtyping
|
|
16
|
-
Requires-Dist: kdepy
|
|
17
16
|
Requires-Dist: linear_operator
|
|
18
17
|
Requires-Dist: numba
|
|
19
18
|
Requires-Dist: numpy>=1.20; python_version < "3.13"
|
|
@@ -37,6 +36,7 @@ Provides-Extra: test
|
|
|
37
36
|
Requires-Dist: pytest; extra == "test"
|
|
38
37
|
Requires-Dist: pytest-cov; extra == "test"
|
|
39
38
|
Provides-Extra: vis
|
|
39
|
+
Requires-Dist: kdepy; extra == "vis"
|
|
40
40
|
Requires-Dist: matplotlib; extra == "vis"
|
|
41
41
|
Requires-Dist: seaborn; extra == "vis"
|
|
42
42
|
Requires-Dist: matplotlib-venn; extra == "vis"
|
|
@@ -8,17 +8,16 @@ dynamic = ["version"]
|
|
|
8
8
|
authors = [{ name = "Charlie Windolf", email = "ciw2048@gmail.com" }]
|
|
9
9
|
description = "DARTsort"
|
|
10
10
|
readme = "README.md"
|
|
11
|
-
requires-python = ">=3.
|
|
11
|
+
requires-python = ">=3.12"
|
|
12
12
|
classifiers = [
|
|
13
13
|
"Programming Language :: Python :: 3",
|
|
14
14
|
"License :: OSI Approved :: MIT License",
|
|
15
15
|
"Operating System :: OS Independent",
|
|
16
16
|
]
|
|
17
17
|
dependencies = [
|
|
18
|
-
"dredge-ephys",
|
|
18
|
+
"dredge-ephys>=0.4.0",
|
|
19
19
|
"h5py",
|
|
20
20
|
"jaxtyping",
|
|
21
|
-
"kdepy",
|
|
22
21
|
"linear_operator",
|
|
23
22
|
"numba",
|
|
24
23
|
"numpy>=1.20;python_version<'3.13'",
|
|
@@ -40,7 +39,7 @@ dependencies = [
|
|
|
40
39
|
[project.optional-dependencies]
|
|
41
40
|
doc = ["zensical", "mkdocstrings-python"]
|
|
42
41
|
test = ["pytest", "pytest-cov"]
|
|
43
|
-
vis = ["matplotlib", "seaborn", "matplotlib-venn"]
|
|
42
|
+
vis = ["kdepy", "matplotlib", "seaborn", "matplotlib-venn"]
|
|
44
43
|
|
|
45
44
|
[project.urls]
|
|
46
45
|
"Homepage" = "https://github.com/cwindolf/dartsort"
|
|
@@ -6,9 +6,15 @@ from typing import cast
|
|
|
6
6
|
import numba
|
|
7
7
|
import numpy as np
|
|
8
8
|
import torch
|
|
9
|
-
from KDEpy import FFTKDE
|
|
10
9
|
from spikeinterface.core import BaseRecording
|
|
11
10
|
|
|
11
|
+
try:
|
|
12
|
+
import KDEpy
|
|
13
|
+
|
|
14
|
+
HAVE_KDEPY = True
|
|
15
|
+
except ImportError:
|
|
16
|
+
HAVE_KDEPY = False
|
|
17
|
+
|
|
12
18
|
from ..templates.template_util import shared_basis_compress_templates
|
|
13
19
|
from ..templates.templates import TemplateData
|
|
14
20
|
from ..util.data_util import (
|
|
@@ -479,6 +485,8 @@ def qda(
|
|
|
479
485
|
) -> QDAResult:
|
|
480
486
|
from ..util.data_util import get_gmm_scores
|
|
481
487
|
|
|
488
|
+
assert HAVE_KDEPY or not bimodality
|
|
489
|
+
|
|
482
490
|
# reconstruct scores from sorting attached data (exclude train_ix?)
|
|
483
491
|
gscores = get_gmm_scores(sorting)
|
|
484
492
|
glabels = sorting.labels
|
|
@@ -597,6 +605,8 @@ def _qda_job(ij):
|
|
|
597
605
|
assert np.isclose(binc[bc], 0.0)
|
|
598
606
|
assert binc.shape[0] == 2 * bc + 1
|
|
599
607
|
|
|
608
|
+
from KDEpy import FFTKDE
|
|
609
|
+
|
|
600
610
|
try:
|
|
601
611
|
kde = FFTKDE(bw="ISJ").fit(dll)
|
|
602
612
|
except ValueError as e:
|
|
@@ -807,7 +817,7 @@ def _check_soft_assign_invariants(
|
|
|
807
817
|
maxdiff = -np.inf
|
|
808
818
|
n_neginf_viol = 0
|
|
809
819
|
|
|
810
|
-
for s in numba.prange(cand.shape[0]):
|
|
820
|
+
for s in numba.prange(cand.shape[0]):
|
|
811
821
|
for j in range(n_cand):
|
|
812
822
|
if cand[s, j] < 0:
|
|
813
823
|
nbye += 1
|
|
@@ -827,7 +837,7 @@ def _assign_labels(
|
|
|
827
837
|
n_changed = 0
|
|
828
838
|
n_labeled = 0
|
|
829
839
|
|
|
830
|
-
for s in numba.prange(cand.shape[0]):
|
|
840
|
+
for s in numba.prange(cand.shape[0]):
|
|
831
841
|
if labels[s] >= 0:
|
|
832
842
|
n_labeled += 1
|
|
833
843
|
if labels[s] != cand[s, 0]:
|
|
@@ -846,7 +856,7 @@ def _combine_loop(
|
|
|
846
856
|
):
|
|
847
857
|
n_cand = cand.shape[1]
|
|
848
858
|
|
|
849
|
-
for s in numba.prange(cand.shape[0]):
|
|
859
|
+
for s in numba.prange(cand.shape[0]):
|
|
850
860
|
spike_cand = cand[s]
|
|
851
861
|
for j in range(n_cand - 1):
|
|
852
862
|
spike_candj = spike_cand[j]
|
|
@@ -121,7 +121,7 @@ def merge_group_shifts(
|
|
|
121
121
|
|
|
122
122
|
@numba.njit(nogil=True, parallel=True)
|
|
123
123
|
def apply_time_shifts(times: np.ndarray, labels: np.ndarray, unit_shifts: np.ndarray):
|
|
124
|
-
for i in numba.prange(times.shape[0]):
|
|
124
|
+
for i in numba.prange(times.shape[0]):
|
|
125
125
|
label = labels[i]
|
|
126
126
|
if label >= 0:
|
|
127
127
|
times[i] -= unit_shifts[label]
|
|
@@ -867,7 +867,7 @@ def _violation_count_matrix(
|
|
|
867
867
|
n = times.shape[0]
|
|
868
868
|
|
|
869
869
|
# parallelize over chunks
|
|
870
|
-
for c in numba.prange(starts.shape[0] - 1):
|
|
870
|
+
for c in numba.prange(starts.shape[0] - 1):
|
|
871
871
|
out = counts[c] # my thread's output buffer
|
|
872
872
|
|
|
873
873
|
for i in range(starts[c], starts[c + 1]):
|
|
@@ -904,7 +904,7 @@ def _violation_weight_matrix(
|
|
|
904
904
|
window_length = weights.shape[0] - 1
|
|
905
905
|
|
|
906
906
|
# parallelize over chunks
|
|
907
|
-
for c in numba.prange(starts.shape[0] - 1):
|
|
907
|
+
for c in numba.prange(starts.shape[0] - 1):
|
|
908
908
|
out = sums[c] # my thread's output buffer
|
|
909
909
|
|
|
910
910
|
for i in range(starts[c], starts[c + 1]):
|
|
@@ -838,6 +838,7 @@ class TMMRefinement(Refinement):
|
|
|
838
838
|
save_step_labels_format=self.labels_fmt,
|
|
839
839
|
save_step_labels_dir=self.save_labels_dir,
|
|
840
840
|
save_cfg=self.save_cfg,
|
|
841
|
+
seed=self.refinement_cfg.sampling_cfg.seed,
|
|
841
842
|
skip_final_assign_and_return_mix_data=skip_final_assign_and_return_mix_data,
|
|
842
843
|
)
|
|
843
844
|
gc.collect()
|
|
@@ -7598,7 +7598,7 @@ def _update_lut_mean_batch(
|
|
|
7598
7598
|
|
|
7599
7599
|
# constplogdet. add in the signal-rank-0-only terms.
|
|
7600
7600
|
lut_params.constplogdet[i0:i1] = neighb_cov.nobs[nn].mul_(LOG_2PI) # type: ignore # ty: ignore[x]
|
|
7601
|
-
lut_params.constplogdet[i0:i1] += neighb_cov.b.logdet[nn]
|
|
7601
|
+
lut_params.constplogdet[i0:i1] += neighb_cov.b.logdet[nn] # ty: ignore[not-subscriptable]
|
|
7602
7602
|
if pnoid:
|
|
7603
7603
|
assert lut_params.constplogdet[i0:i1].isfinite().all() # type: ignore # ty: ignore[x]
|
|
7604
7604
|
|
|
@@ -7658,7 +7658,7 @@ def _update_lut_ppca_batch(
|
|
|
7658
7658
|
assert lut_params.Tpad is not None
|
|
7659
7659
|
lut_params.Tpad[i0:i1, :, :-1] = T # type: ignore # ty: ignore[x]
|
|
7660
7660
|
cap_logdet = L.diagonal(dim1=-2, dim2=-1).log().sum(dim=1).mul_(2.0)
|
|
7661
|
-
lut_params.constplogdet[i0:i1] += cap_logdet
|
|
7661
|
+
lut_params.constplogdet[i0:i1] += cap_logdet # ty: ignore[not-subscriptable]
|
|
7662
7662
|
if pnoid:
|
|
7663
7663
|
assert lut_params.b.constplogdet[i0:i1].isfinite().all()
|
|
7664
7664
|
|
|
@@ -4,12 +4,17 @@ from typing import cast
|
|
|
4
4
|
import numba
|
|
5
5
|
import numpy as np
|
|
6
6
|
import torch
|
|
7
|
-
from KDEpy import FFTKDE
|
|
8
|
-
from KDEpy.bw_selection import improved_sheather_jones
|
|
9
7
|
from scipy.spatial import KDTree
|
|
10
8
|
from scipy.stats import norm
|
|
11
9
|
from spikeinterface.core.baserecording import BaseRecording
|
|
12
10
|
|
|
11
|
+
try:
|
|
12
|
+
import KDEpy
|
|
13
|
+
|
|
14
|
+
HAVE_KDEPY = True
|
|
15
|
+
except ImportError:
|
|
16
|
+
HAVE_KDEPY = False
|
|
17
|
+
|
|
13
18
|
from dartsort.clustering.mixture import Scores
|
|
14
19
|
|
|
15
20
|
from ..util import data_util, spiketorch
|
|
@@ -392,7 +397,9 @@ def collision_cleaning_error_filter(
|
|
|
392
397
|
assert flag_scores.excess_rate is not None
|
|
393
398
|
flag_bad |= flag_scores.excess_rate > refinement_cfg.cc_flag_excess_rate
|
|
394
399
|
|
|
395
|
-
logger.dartsortdebug(
|
|
400
|
+
logger.dartsortdebug(
|
|
401
|
+
f"CC flag criterion dropped {flag_bad.sum()} / {nu0} units."
|
|
402
|
+
)
|
|
396
403
|
keep_mask &= np.logical_not(flag_bad)
|
|
397
404
|
|
|
398
405
|
bad_ids = np.flatnonzero(np.logical_not(keep_mask))
|
|
@@ -646,6 +653,7 @@ def gmm_isolation_scores(
|
|
|
646
653
|
kde_rhs=50.0,
|
|
647
654
|
kde_dx: float = 0.5,
|
|
648
655
|
) -> GMMIsolationScores:
|
|
656
|
+
assert HAVE_KDEPY
|
|
649
657
|
if unit_ids is None:
|
|
650
658
|
unit_ids = np.arange(scores.candidates[:, 0].max().item() + 1)
|
|
651
659
|
assert unit_ids is not None
|
|
@@ -761,6 +769,9 @@ def _iso_job(unit_id) -> tuple[float, np.ndarray | None, np.ndarray | None]:
|
|
|
761
769
|
# fit kde. will symmetrize around 0 to avoid boundary issues (since lr>=0)
|
|
762
770
|
amax = in_lr.abs().amax().item()
|
|
763
771
|
lr = in_lr.double().numpy(force=True)
|
|
772
|
+
from KDEpy import FFTKDE
|
|
773
|
+
from KDEpy.bw_selection import improved_sheather_jones
|
|
774
|
+
|
|
764
775
|
try:
|
|
765
776
|
# select bandwidth based on positive part
|
|
766
777
|
bw = improved_sheather_jones(lr[:, None])
|
|
@@ -10,6 +10,7 @@ from .util.internal_config import (
|
|
|
10
10
|
KmeansppSelection,
|
|
11
11
|
KmeansppStopping,
|
|
12
12
|
MixtureStep,
|
|
13
|
+
MotionBatchingMode,
|
|
13
14
|
PreprocessingStrategy,
|
|
14
15
|
RealignStrategy,
|
|
15
16
|
TemplateSVDMethod,
|
|
@@ -250,6 +251,10 @@ class DARTsortUserConfig:
|
|
|
250
251
|
max_dist_from_median_um: float = 250.0
|
|
251
252
|
"""Motion bins farther than this from the local median will be replaced by interpolation."""
|
|
252
253
|
median_neighborhood_bins: int = 51
|
|
254
|
+
motion_batching_mode: MotionBatchingMode = "full"
|
|
255
|
+
"""'full' or 'online'. Online helps with long recordings."""
|
|
256
|
+
motion_chunk_len_s: Annotated[float, Field(gt=0)] = 3000.0
|
|
257
|
+
"""Chunk length in seconds for online motion tracking"""
|
|
253
258
|
|
|
254
259
|
# -- pipeline control parameters for specific use cases
|
|
255
260
|
fit_matching_models_only: bool = False
|
|
@@ -369,8 +374,8 @@ class DeveloperConfig(DARTsortUserConfig):
|
|
|
369
374
|
kmeanspp_patience: int = 21
|
|
370
375
|
kmeanspp_greedy_proposals: int = 1
|
|
371
376
|
kmeanspp_neighb_overlap: float | None = None
|
|
372
|
-
kmeanspp_selection: KmeansppSelection = "
|
|
373
|
-
kmeanspp_stopping: KmeansppStopping = "
|
|
377
|
+
kmeanspp_selection: KmeansppSelection = "marginal"
|
|
378
|
+
kmeanspp_stopping: KmeansppStopping = "dpmeanspp"
|
|
374
379
|
|
|
375
380
|
# filters
|
|
376
381
|
gmm_isolation_threshold: float | None = None
|
|
@@ -401,7 +406,7 @@ class DeveloperConfig(DARTsortUserConfig):
|
|
|
401
406
|
gmm_n_candidates: int = 5
|
|
402
407
|
gmm_n_search: int | None = 3
|
|
403
408
|
gmm_val_proportion: Annotated[float, Field(gt=0)] = 0.5
|
|
404
|
-
gmm_batch_size: int =
|
|
409
|
+
gmm_batch_size: int = 4096
|
|
405
410
|
initial_basis_shrinkage: float = 1.0
|
|
406
411
|
prior_pseudocount: float = 0.0
|
|
407
412
|
cov_kind: str = "factorizednoise"
|
|
@@ -16,7 +16,7 @@ import spikeinterface.core as sc
|
|
|
16
16
|
import torch
|
|
17
17
|
from sklearn.decomposition import PCA
|
|
18
18
|
|
|
19
|
-
from ..clustering.agglomerate import QDAResult, qda, template_distances
|
|
19
|
+
from ..clustering.agglomerate import HAVE_KDEPY, QDAResult, qda, template_distances
|
|
20
20
|
from ..templates import TemplateData
|
|
21
21
|
from ..util import job_util, logging_util
|
|
22
22
|
from ..util.data_util import (
|
|
@@ -193,6 +193,7 @@ class DARTsortAnalysis:
|
|
|
193
193
|
|
|
194
194
|
if (
|
|
195
195
|
allow_qda
|
|
196
|
+
and HAVE_KDEPY
|
|
196
197
|
and template_data is not None
|
|
197
198
|
and hasattr(sorting, "gmm_candidates")
|
|
198
199
|
):
|
|
@@ -135,7 +135,7 @@ def make_hybrid_recording(
|
|
|
135
135
|
assert target_recording.get_num_segments() == 1
|
|
136
136
|
assert injected_sorting.get_num_segments() == 1
|
|
137
137
|
if hasattr(injected_sorting, "get_last_spike_frame"):
|
|
138
|
-
assert injected_sorting.get_last_spike_frame() < target_recording.get_num_frames()
|
|
138
|
+
assert injected_sorting.get_last_spike_frame() < target_recording.get_num_frames()
|
|
139
139
|
assert abs(templates.sampling_frequency - target_sampling_frequency) < 10
|
|
140
140
|
|
|
141
141
|
if folder is not None and not overwrite:
|
|
@@ -71,7 +71,7 @@ def get_background_recording(
|
|
|
71
71
|
spatial_std, spatial_vt = rbf_kernel_sqrt(
|
|
72
72
|
geom, bandwidth=noise_spatial_kernel_bandwidth
|
|
73
73
|
)
|
|
74
|
-
noise_temporal_kernel = noise_temporal_kernel.astype(spatial_vt.dtype) + 0j
|
|
74
|
+
noise_temporal_kernel = noise_temporal_kernel.astype(spatial_vt.dtype) + 0j
|
|
75
75
|
noise = StationaryFactorizedNoise(
|
|
76
76
|
spatial_std=spatial_std,
|
|
77
77
|
vt_spatial=spatial_vt,
|
|
@@ -119,7 +119,7 @@ class WhiteNoiseRecording(BaseRecording):
|
|
|
119
119
|
channel_ids = np.asarray(channel_ids)
|
|
120
120
|
assert channel_ids.shape == (n_channels,)
|
|
121
121
|
assert n_segments == 1
|
|
122
|
-
super().__init__(sampling_frequency, channel_ids=channel_ids, dtype=dtype)
|
|
122
|
+
super().__init__(sampling_frequency, channel_ids=channel_ids, dtype=dtype)
|
|
123
123
|
self._serializability["json"] = False
|
|
124
124
|
self._serializability["pickle"] = False
|
|
125
125
|
segment = WhiteNoiseRecordingSegment(
|
|
@@ -21,7 +21,6 @@ def localize_waveforms(
|
|
|
21
21
|
radius=None,
|
|
22
22
|
n_channels_subset=None,
|
|
23
23
|
):
|
|
24
|
-
C, dim = geom.shape
|
|
25
24
|
amp_vecs = ptp(waveforms, 1)
|
|
26
25
|
if main_channels is None:
|
|
27
26
|
main_channels = amp_vecs.argmax(1)
|
|
@@ -287,7 +286,7 @@ def point_source_mse(locs, amp_vecs, channels, channel_index, geom):
|
|
|
287
286
|
torch.asarray(amp_vecs).nan_to_num(),
|
|
288
287
|
channel_mask,
|
|
289
288
|
*torch.asarray(locs).T,
|
|
290
|
-
torch.asarray(dxz).nan_to_num(),
|
|
289
|
+
torch.asarray(dxz).nan_to_num(), # ty: ignore[too-many-positional-arguments]
|
|
291
290
|
)
|
|
292
291
|
alpha = alpha.numpy()
|
|
293
292
|
else:
|
|
@@ -907,17 +907,10 @@ class BasePeeler(BModule):
|
|
|
907
907
|
def sample_chunks(
|
|
908
908
|
self, chunk_starts_samples: np.ndarray, n_chunks: int, ordered: bool = False
|
|
909
909
|
) -> np.ndarray:
|
|
910
|
-
|
|
911
|
-
|
|
912
|
-
chunk_starts_samples =
|
|
913
|
-
|
|
914
|
-
)
|
|
915
|
-
elif self.fit_sampling_cfg.chunk_sampling == "kmeanspp":
|
|
916
|
-
_ix = _kmeansppix(len(chunk_starts_samples), n_chunks, self.random_seed)
|
|
917
|
-
chunk_starts_samples = chunk_starts_samples[_ix]
|
|
918
|
-
else:
|
|
919
|
-
panic(self.fit_sampling_cfg.chunk_sampling)
|
|
920
|
-
|
|
910
|
+
rg = np.random.default_rng(self.random_seed)
|
|
911
|
+
chunk_starts_samples = rg.choice(
|
|
912
|
+
chunk_starts_samples, size=n_chunks, replace=False
|
|
913
|
+
)
|
|
921
914
|
if ordered:
|
|
922
915
|
chunk_starts_samples.sort()
|
|
923
916
|
|
|
@@ -1354,31 +1347,3 @@ def _peeler_process_job(chunk_start_samples__n_resid_snips):
|
|
|
1354
1347
|
skip_features=_peeler_process_context.ctx.skip_features,
|
|
1355
1348
|
to_cpu=_peeler_process_context.ctx.to_cpu,
|
|
1356
1349
|
)
|
|
1357
|
-
|
|
1358
|
-
|
|
1359
|
-
def _kmeansppix(n: int, k: int | None = None, seed=0):
|
|
1360
|
-
rg = np.random.default_rng(seed)
|
|
1361
|
-
k = k if k is not None else n
|
|
1362
|
-
assert 0 < k <= n
|
|
1363
|
-
|
|
1364
|
-
xx = np.arange(float(n))
|
|
1365
|
-
xx -= np.mean(xx)
|
|
1366
|
-
xx /= np.std(xx)
|
|
1367
|
-
|
|
1368
|
-
inds = np.full((k,), n + 1, dtype=np.int64)
|
|
1369
|
-
inds[0] = rg.integers(n)
|
|
1370
|
-
|
|
1371
|
-
dist = xx - xx[inds[0]]
|
|
1372
|
-
dist *= dist
|
|
1373
|
-
|
|
1374
|
-
scratch = np.empty_like(dist)
|
|
1375
|
-
|
|
1376
|
-
for j in range(1, k):
|
|
1377
|
-
p = np.divide(dist, dist.sum(), out=scratch)
|
|
1378
|
-
inds[j] = rg.choice(n, p=p)
|
|
1379
|
-
dj = np.subtract(xx, xx[inds[j]], out=scratch)
|
|
1380
|
-
dj *= dj
|
|
1381
|
-
np.minimum(dist, dj, out=dist)
|
|
1382
|
-
|
|
1383
|
-
assert np.unique(inds).shape == (k,)
|
|
1384
|
-
return inds
|
|
@@ -43,7 +43,6 @@ from .subtract_util import (
|
|
|
43
43
|
ChunkSubtracter,
|
|
44
44
|
make_peak_proposer,
|
|
45
45
|
proposal_filters_from_pipeline,
|
|
46
|
-
proposal_fit_featurization_cfg,
|
|
47
46
|
with_proposal_filters,
|
|
48
47
|
)
|
|
49
48
|
|
|
@@ -64,7 +63,6 @@ class SubtractionPeeler(BasePeeler):
|
|
|
64
63
|
p: SubtractionConfig = default_subtraction_cfg,
|
|
65
64
|
waveform_cfg: WaveformConfig = default_waveform_cfg,
|
|
66
65
|
fit_sampling_cfg: FitSamplingConfig = default_peeling_fit_sampling_cfg,
|
|
67
|
-
featurization_cfg: FeaturizationConfig | None = None,
|
|
68
66
|
save_iteration=False,
|
|
69
67
|
save_residnorm_decrease=False,
|
|
70
68
|
save_collidedness=False,
|
|
@@ -94,8 +92,6 @@ class SubtractionPeeler(BasePeeler):
|
|
|
94
92
|
self.save_iteration = save_iteration
|
|
95
93
|
self.save_residnorm_decrease = save_residnorm_decrease
|
|
96
94
|
self.save_collidedness = save_collidedness
|
|
97
|
-
# non-voltage proposal may need this config
|
|
98
|
-
self.featurization_cfg = featurization_cfg
|
|
99
95
|
self.dedup_batch_size = self.nearest_batch_length()
|
|
100
96
|
if self.p.whiten:
|
|
101
97
|
self.threshold = self.p.threshold_before_whitening
|
|
@@ -388,7 +384,6 @@ class SubtractionPeeler(BasePeeler):
|
|
|
388
384
|
p=subtraction_cfg,
|
|
389
385
|
waveform_cfg=waveform_cfg,
|
|
390
386
|
fit_sampling_cfg=sampling_cfg,
|
|
391
|
-
featurization_cfg=featurization_cfg,
|
|
392
387
|
save_iteration=subtraction_cfg.save_iteration,
|
|
393
388
|
save_residnorm_decrease=subtraction_cfg.save_residnorm_decrease,
|
|
394
389
|
save_collidedness=save_collidedness,
|
|
@@ -628,10 +623,8 @@ class SubtractionPeeler(BasePeeler):
|
|
|
628
623
|
|
|
629
624
|
def _threshold_to_fit(self, tmp_dir, fit_pipeline, computation_cfg):
|
|
630
625
|
threshold_cfg = ThresholdingConfig(
|
|
631
|
-
detection_proposal=
|
|
632
|
-
proposal_filters=self.p.proposal_filters,
|
|
626
|
+
detection_proposal="voltage",
|
|
633
627
|
voltage_threshold=self.p.voltage_threshold,
|
|
634
|
-
proposal_threshold=self.p.proposal_threshold,
|
|
635
628
|
spatial_dedup_radius_um=(
|
|
636
629
|
self.p.first_denoiser_spatial_dedup_radius or self.p.subtract_radius_um
|
|
637
630
|
),
|
|
@@ -659,10 +652,15 @@ class SubtractionPeeler(BasePeeler):
|
|
|
659
652
|
waveform_cfg=self.waveform_cfg,
|
|
660
653
|
channel_index=self.b.sub_channel_index,
|
|
661
654
|
threshold_cfg=threshold_cfg,
|
|
662
|
-
featurization_cfg=
|
|
663
|
-
|
|
664
|
-
|
|
665
|
-
|
|
655
|
+
featurization_cfg=FeaturizationConfig(
|
|
656
|
+
do_tpca_denoise=False,
|
|
657
|
+
do_enforce_decrease="no",
|
|
658
|
+
save_input_voltages=False,
|
|
659
|
+
save_input_waveforms=True,
|
|
660
|
+
save_input_tpca_projs=False,
|
|
661
|
+
save_amplitudes=False,
|
|
662
|
+
do_localization=False,
|
|
663
|
+
input_waveforms_name="",
|
|
666
664
|
),
|
|
667
665
|
sampling_cfg=sampling_cfg,
|
|
668
666
|
max_waveforms_fit=self.p.first_denoiser_max_waveforms_fit,
|
|
@@ -708,34 +708,6 @@ def with_proposal_filters(
|
|
|
708
708
|
panic(proposal)
|
|
709
709
|
|
|
710
710
|
|
|
711
|
-
def proposal_fit_featurization_cfg(
|
|
712
|
-
featurization_cfg: FeaturizationConfig | None,
|
|
713
|
-
proposal: DetectionProposal,
|
|
714
|
-
n_filters: int,
|
|
715
|
-
) -> FeaturizationConfig:
|
|
716
|
-
if featurization_cfg is None and proposal != "voltage":
|
|
717
|
-
raise ValueError(
|
|
718
|
-
f"{proposal=} needs the featurization config its filters are fit with."
|
|
719
|
-
)
|
|
720
|
-
if featurization_cfg is None:
|
|
721
|
-
featurization_cfg = FeaturizationConfig()
|
|
722
|
-
waveforms_only = FeaturizationConfig(
|
|
723
|
-
do_tpca_denoise=False,
|
|
724
|
-
do_enforce_decrease="no",
|
|
725
|
-
save_input_voltages=False,
|
|
726
|
-
save_input_waveforms=True,
|
|
727
|
-
save_input_tpca_projs=False,
|
|
728
|
-
save_amplitudes=False,
|
|
729
|
-
do_localization=False,
|
|
730
|
-
input_waveforms_name="",
|
|
731
|
-
tpca_rank=featurization_cfg.tpca_rank,
|
|
732
|
-
tpca_fit_radius=featurization_cfg.tpca_fit_radius,
|
|
733
|
-
tpca_max_waveforms=featurization_cfg.tpca_max_waveforms,
|
|
734
|
-
input_tpca_waveform_cfg=featurization_cfg.input_tpca_waveform_cfg,
|
|
735
|
-
)
|
|
736
|
-
return with_proposal_filters(waveforms_only, proposal, n_filters)
|
|
737
|
-
|
|
738
|
-
|
|
739
711
|
def proposal_filters_from_pipeline(
|
|
740
712
|
pipeline: "WaveformPipeline", proposal: DetectionProposal, n_filters: int
|
|
741
713
|
) -> tuple[Tensor, int]:
|
|
@@ -353,7 +353,7 @@ def get_noise_h5(
|
|
|
353
353
|
|
|
354
354
|
@numba.njit(parallel=True)
|
|
355
355
|
def take_along_dim2_masked(arr, out, channels, mask_val):
|
|
356
|
-
for i in numba.prange(arr.shape[0]):
|
|
356
|
+
for i in numba.prange(arr.shape[0]):
|
|
357
357
|
wf = arr[i]
|
|
358
358
|
cc = channels[i]
|
|
359
359
|
for t in range(arr.shape[1]):
|
|
@@ -159,24 +159,9 @@ class TruncatedMixtureModelTransformer(BaseWaveformFeaturizer):
|
|
|
159
159
|
cii, cjj = np.nonzero(self.channel_index_np < len(self.channel_index_np))
|
|
160
160
|
self.channel_index_valid_inds = cii, cjj
|
|
161
161
|
|
|
162
|
-
# precompute drifting neighborhood lookup tables
|
|
163
162
|
if self.motion.drifting:
|
|
164
|
-
# in this case, prebake a lookup table
|
|
165
|
-
# TODO just consider relevant time bins here?
|
|
166
|
-
# or chunk this somehow?
|
|
167
|
-
tbs = self.motion.time_bins_s
|
|
168
|
-
tcm_t, nid_t = neighborhood_mapping_at_time(
|
|
169
|
-
motion=self.motion,
|
|
170
|
-
t_s=torch.from_numpy(tbs),
|
|
171
|
-
neighborhoods=self.b.neighborhoods,
|
|
172
|
-
channel_index=self.channel_index_np,
|
|
173
|
-
workers=self.workers,
|
|
174
|
-
static_neighbs=self.static_neighbs,
|
|
175
|
-
channel_index_valid_inds=self.channel_index_valid_inds,
|
|
176
|
-
)
|
|
177
163
|
nid_map = None
|
|
178
164
|
else:
|
|
179
|
-
tcm_t = nid_t = tbs = None
|
|
180
165
|
# in this case, channel index is a superset of neighborhoods and
|
|
181
166
|
# we prebake the mapping from chans to neighb ids
|
|
182
167
|
ci_eq_neighb = _outer_all_equal(
|
|
@@ -193,13 +178,6 @@ class TruncatedMixtureModelTransformer(BaseWaveformFeaturizer):
|
|
|
193
178
|
)
|
|
194
179
|
nid_map[_chans] = _chan_nids
|
|
195
180
|
self.register_buffer_or_none("neighborhood_ids_map", nid_map)
|
|
196
|
-
self.drifting_time_bins_s = tbs
|
|
197
|
-
self.register_buffer_or_none(
|
|
198
|
-
"drifting_target_channels_map_t", tcm_t, persistent=False
|
|
199
|
-
)
|
|
200
|
-
self.register_buffer_or_none(
|
|
201
|
-
"drifting_neighborhood_ids_map_t", nid_t, persistent=False
|
|
202
|
-
)
|
|
203
181
|
|
|
204
182
|
def fit(
|
|
205
183
|
self,
|
|
@@ -360,10 +338,21 @@ class TruncatedMixtureModelTransformer(BaseWaveformFeaturizer):
|
|
|
360
338
|
# drifting channel mapping
|
|
361
339
|
assert self.motion is not None
|
|
362
340
|
if self.motion.drifting:
|
|
363
|
-
|
|
364
|
-
|
|
365
|
-
|
|
366
|
-
|
|
341
|
+
time_bins_s = self.motion.time_bins_s
|
|
342
|
+
t_bin_s = time_bins_s[find_nearest(time_bins_s, chunk_center_s)]
|
|
343
|
+
depths_um = self.motion.geom[:, 1]
|
|
344
|
+
_, n_pitches_shift = self.motion.pitch_shifts(
|
|
345
|
+
times_s=np.full_like(depths_um, t_bin_s), depths_um=depths_um
|
|
346
|
+
)
|
|
347
|
+
target_channels_map, neighborhood_ids_map = neighborhood_mapping_at_shift(
|
|
348
|
+
self.motion,
|
|
349
|
+
n_pitches_shift=n_pitches_shift,
|
|
350
|
+
channel_index=self.channel_index_np,
|
|
351
|
+
neighborhoods=self.b.neighborhoods,
|
|
352
|
+
static_neighbs=self.static_neighbs,
|
|
353
|
+
channel_index_valid_inds=self.channel_index_valid_inds,
|
|
354
|
+
workers=self.workers,
|
|
355
|
+
)
|
|
367
356
|
else:
|
|
368
357
|
target_channels_map = self.b.channel_index
|
|
369
358
|
neighborhood_ids_map = self.b.neighborhood_ids_map
|
|
@@ -411,59 +400,37 @@ class TruncatedMixtureModelTransformer(BaseWaveformFeaturizer):
|
|
|
411
400
|
return res
|
|
412
401
|
|
|
413
402
|
|
|
414
|
-
def
|
|
403
|
+
def neighborhood_mapping_at_shift(
|
|
415
404
|
motion: MotionInfo,
|
|
416
405
|
*,
|
|
417
|
-
|
|
406
|
+
n_pitches_shift: np.ndarray,
|
|
418
407
|
channel_index: np.ndarray,
|
|
419
408
|
neighborhoods: torch.Tensor,
|
|
420
|
-
workers: int = 4,
|
|
421
|
-
shift_mode="round",
|
|
422
409
|
static_neighbs: np.ndarray,
|
|
423
410
|
channel_index_valid_inds: tuple[np.ndarray, np.ndarray],
|
|
424
|
-
|
|
425
|
-
|
|
426
|
-
|
|
427
|
-
|
|
428
|
-
|
|
429
|
-
|
|
430
|
-
|
|
431
|
-
|
|
432
|
-
|
|
411
|
+
workers: int = 4,
|
|
412
|
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
|
413
|
+
"""Map channels to registered target channels and neighborhood ids"""
|
|
414
|
+
shifted_neighbs = static_neighbs.copy()
|
|
415
|
+
shifted_neighbs[:, :, 1] += n_pitches_shift[:, None] * motion.pitch
|
|
416
|
+
_, umatch = motion.rgeom_kdt.query(
|
|
417
|
+
shifted_neighbs[channel_index_valid_inds],
|
|
418
|
+
distance_upper_bound=motion.min_dist,
|
|
419
|
+
workers=workers,
|
|
433
420
|
)
|
|
421
|
+
index = np.full(channel_index.shape, motion.rgeom.shape[0])
|
|
422
|
+
index[channel_index_valid_inds] = umatch
|
|
423
|
+
index = torch.asarray(index, device=neighborhoods.device)
|
|
424
|
+
assert index.shape[1] == neighborhoods.shape[1]
|
|
425
|
+
|
|
426
|
+
mapping = _outer_all_equal(index, neighborhoods)
|
|
427
|
+
_chans, _chan_nids = mapping.nonzero(as_tuple=True)
|
|
428
|
+
nids = torch.full(
|
|
429
|
+
(channel_index.shape[0],), neighborhoods.shape[0], device=neighborhoods.device
|
|
430
|
+
)
|
|
431
|
+
nids[_chans] = _chan_nids
|
|
434
432
|
|
|
435
|
-
|
|
436
|
-
probe_disp = -motion.disp_at_s(
|
|
437
|
-
times_s=t, depths_um=motion.geom[:, 1], grid=True
|
|
438
|
-
)
|
|
439
|
-
if shift_mode == "floor":
|
|
440
|
-
n_pitches_shift = (probe_disp / motion.pitch).astype(np.int32)
|
|
441
|
-
elif shift_mode == "round":
|
|
442
|
-
n_pitches_shift = np.round(probe_disp / motion.pitch).astype(np.int32)
|
|
443
|
-
else:
|
|
444
|
-
panic(shift_mode)
|
|
445
|
-
shift = n_pitches_shift * motion.pitch
|
|
446
|
-
shifted_neighbs = static_neighbs.copy()
|
|
447
|
-
shifted_neighbs[:, :, 1] += shift
|
|
448
|
-
_, umatch = motion.rgeom_kdt.query(
|
|
449
|
-
shifted_neighbs[channel_index_valid_inds],
|
|
450
|
-
distance_upper_bound=motion.min_dist,
|
|
451
|
-
workers=workers,
|
|
452
|
-
)
|
|
453
|
-
index = np.full(channel_index.shape, motion.rgeom.shape[0])
|
|
454
|
-
index[channel_index_valid_inds] = umatch
|
|
455
|
-
index = torch.asarray(index, device=neighborhoods.device)
|
|
456
|
-
assert index.shape[1] == neighborhoods.shape[1]
|
|
457
|
-
|
|
458
|
-
index_at_time[j] = index
|
|
459
|
-
|
|
460
|
-
mapping = _outer_all_equal(index, neighborhoods)
|
|
461
|
-
_chans, _chan_nids = mapping.nonzero(as_tuple=True)
|
|
462
|
-
nids_at_time[j, _chans] = _chan_nids
|
|
463
|
-
|
|
464
|
-
assert (index_at_time >= 0).all()
|
|
465
|
-
|
|
466
|
-
return index_at_time, nids_at_time
|
|
433
|
+
return index, nids
|
|
467
434
|
|
|
468
435
|
|
|
469
436
|
@torch_compile
|