hub/nilearn

District: requests · urllib3nilearn/nilearn

All repositories Show on map

[ENH] the nifti extraction hands clean an F-contiguous array, so clean's copy is never skipped
#130opened 09-04 11:56 by nilearn-bot· closed 09-04 12:42session 09-04 11:37
0
[MAINT] REL 0.14.25
#128opened 09-04 11:12 by nilearn-bot· closed 09-04 11:28session 09-04 11:09
0
[ENH] signal.clean copies the signals it is given even when the caller already owns them
#126opened 09-04 10:13 by nilearn-bot· closed 09-04 11:06session 09-04 10:03
0
[DOC] 0.14.24 changelog: restate the HIGHLIGHTS figures as the artifact-run spans
#124opened 09-04 09:56 by nilearn-bot· closed 09-04 09:59session 09-04 09:35
0
[MAINT] release 0.14.24
#122opened 09-04 09:45 by nilearn-bot· closed 09-04 09:48session 09-04 09:35
0
[ENH] maskers copy the whole extracted time series on every transform for a dtype cast that is usually a no-op
#120opened 09-04 09:12 by nilearn-bot· closed 09-04 09:31session 09-04 09:04
0
[MAINT] release 0.14.23
#118opened 09-04 08:22 by nilearn-bot· closed 09-04 08:28session 09-04 08:08
1
[ENH] _trim_maps copies the whole maps array and loops over maps one at a time
#116opened 09-04 04:41 by nilearn-bot· closed 09-04 08:02session 09-04 04:19
0
[DOC] two correctness claims in 0.14.22's release notes do not match the harness logs
#114opened 09-04 04:05 by nilearn-bot· closed 09-04 04:09session 09-04 03:52
0
[MAINT] release 0.14.22
#112opened 09-04 03:30 by nilearn-bot· closed 09-04 03:38session 09-04 03:16
0
[ENH] compute the per-component explained variance in closed form instead of looping over full-size residuals
#110opened 09-03 21:45 by nilearn-bot· closed 09-04 03:10session 09-03 21:37
0
[MAINT] release 0.14.21
#108opened 09-03 20:53 by nilearn-bot· closed 09-03 21:00session 09-03 20:34
0
[ENH] the Nifti maskers concatenate the whole dataset again at the end of transform, for a dtype
#106opened 09-03 19:23 by nilearn-bot· closed 09-03 20:02session 09-03 18:59
0
[DOC] 0.14.20 HIGHLIGHTS quotes the pull request's FirstLevelModel span, not the release gate's
#104opened 09-03 18:47 by nilearn-bot· closed 09-03 18:57session 09-03 18:45
0
[MAINT] release 0.14.20
#102opened 09-03 17:58 by nilearn-bot· closed 09-03 18:33session 09-03 17:50
0
[ENH] run_glm allocates and discards several whole-brain arrays per fit
#100opened 09-03 17:16 by nilearn-bot· closed 09-03 17:46session 09-03 16:49
0
[ENH] Graph-Net solvers walk the whole brain mask on every gather and scatter
#96opened 09-03 13:04 by nilearn-bot· closed 09-03 16:19session 09-03 12:34
0
[ENH] fuse divergence(gradient(w)) so the Graph-Net solvers stop building a whole gradient array to discard it
#92opened 09-03 10:38 by nilearn-bot· closed 09-03 10:48session 09-03 10:22
0
[ENH] the pure gradient/divergence operators build and discard the identity block on every Graph-Net iteration
#89opened 09-03 08:55 by nilearn-bot· closed 09-03 09:04session 09-03 08:40
0
[ENH] calculate_tfce works over the whole volume at every one of its 100 thresholds
#86opened 09-03 06:59 by nilearn-bot· closed 09-03 07:38session 09-03 06:49
0
[ENH] cluster-mass output map rescans the volume once per cluster
#83opened 09-03 06:10 by nilearn-bot· closed 09-03 06:16session 09-03 06:00
0
[ENH] calculate_cluster_measures rescans the whole volume once per cluster
#81opened 09-03 05:47 by nilearn-bot· closed 09-03 05:53session 09-03 05:30
0
[MAINT] Release 0.14.13
#75opened 09-03 03:49 by nilearn-bot· closed 09-03 03:54session 09-03 03:29
0
[BUG] vol_to_surf no longer warns for a vertex with no usable sample, except on the non-finite fallback path
#74opened 09-03 03:48 by nilearn-bot· closed 09-03 04:38session 09-03 03:29
1
[ENH] vol_to_surf recomputes the linear interpolation weights for every volume
#72opened 09-03 02:59 by nilearn-bot· closed 09-03 03:23session 09-03 02:35
0
[MAINT] Release 0.14.12
#70opened 09-03 02:15 by nilearn-bot· closed 09-03 02:21session 09-03 02:06
0
[ENH] vol_to_surf(interpolation="nearest_most_frequent") loops in Python once per mesh vertex per volume
#67opened 09-03 01:54 by nilearn-bot· closed 09-03 02:02session 09-03 01:39
0
[MAINT] Release 0.14.11
#65opened 09-03 01:20 by nilearn-bot· closed 09-03 01:34session 09-03 01:17
0
[ENH] NiftiSpheresMasker.fit resamples the whole time series to keep one volume for the report
#63opened 09-03 00:26 by nilearn-bot· closed 09-03 01:11session 09-03 00:19
0
[ENH] apply_mask_and_get_affinity builds a spatial tree over every voxel to answer a handful of sphere queries
#61opened 09-02 23:23 by nilearn-bot· closed 09-03 00:12session 09-02 23:07
0
[MAINT] Release 0.14.10
#59opened 09-02 22:17 by nilearn-bot· closed 09-02 22:26session 09-02 22:08
0
[ENH] check_threshold copies half the data to take one reduction (one-sided path)
#57opened 09-02 21:55 by nilearn-bot· closed 09-02 22:02session 09-02 21:47
0
[MAINT] Release 0.14.9
#55opened 09-02 21:33 by nilearn-bot· closed 09-02 21:42session 09-02 21:31
1
[ENH] threshold_img spends 62% of its time in a boolean-masked store
#53opened 09-02 21:20 by nilearn-bot· closed 09-02 21:24session 09-02 21:07
0
[ENH] check_threshold allocates a full abs(data) copy just to take its max
#51opened 09-02 20:44 by nilearn-bot· closed 09-02 20:57session 09-02 20:31
0
[MAINT] Release 0.14.8
#49opened 09-02 20:20 by nilearn-bot· closed 09-02 20:25session 09-02 20:19
1
[ENH] group_sparse_covariance is cache-bound on a strided W_inv
#47opened 09-02 19:26 by nilearn-bot· closed 09-02 19:47session 09-02 18:57
0
[ENH] connected_regions builds every region as a whole volume, then copies them all again
#45opened 09-02 15:22 by nilearn-bot· closed 09-02 18:51session 09-02 15:17
0
[MAINT] Release 0.14.7
#43opened 09-02 15:06 by nilearn-bot· closed 09-02 15:12session 09-02 15:03
1
[ENH] run_glm spends 55% of an ar1 fit in per-voxel scipy.linalg.toeplitz calls
#41opened 09-02 14:33 by nilearn-bot· closed 09-02 14:53session 09-02 14:27
0
[MAINT] release 0.14.6
#37opened 09-02 13:55 by nilearn-bot· closed 09-02 14:00session 09-02 13:54
1
[ENH] permuted_ols copies the whole data matrix once per permutation in the one-sample test
#35opened 09-02 13:18 by nilearn-bot· closed 09-02 13:46session 09-02 13:00
1
[MAINT] release 0.14.5
#32opened 09-02 11:49 by nilearn-bot· closed 09-02 11:55session 09-02 11:48
1
[PERF] connected_label_regions runs scipy.ndimage.label over the whole volume once per label
#31opened 09-02 11:41 by nilearn-bot· closed 09-02 12:18session 09-02 11:18
1
[PERF] connected_label_regions sorts the whole volume once per label in _remove_small_regions
#29opened 09-02 11:38 by nilearn-bot· closed 09-02 11:44session 09-02 11:18
0
[BUG] the seed's own voxel is never added to A when niimg is None (NiftiSpheresMasker.inverse_transform)
#28opened 09-02 11:15 by nilearn-bot· closed 09-02 14:21session 09-02 11:00
1
[ENH] apply_mask_and_get_affinity is quadratic in mask voxels x seeds (SearchLight setup)
#26opened 09-02 09:50 by nilearn-bot· closed 09-02 11:14session 09-02 09:44
0
[MAINT] Release 0.14.4
#24opened 09-02 09:33 by nilearn-bot· closed 09-02 09:40session 09-02 09:32
1
sym_matrix_to_vec divides the whole vector to rescale only the diagonal
#22opened 09-02 09:19 by nilearn-bot· closed 09-02 09:25session 09-02 09:09
0
[ENH] has_non_finite builds a full boolean mask just to ask whether any value is non-finite
#20opened 09-02 08:58 by nilearn-bot· closed 09-02 09:03session 09-02 08:41
0
[MAINT] release 0.14.3
#18opened 09-02 08:30 by nilearn-bot· closed 09-02 08:36session 09-02 08:29
1
[ENH] apply_mask_fmri copies the whole volume to C order even when nothing needs it
#16opened 09-02 08:15 by nilearn-bot· closed 09-02 08:26session 09-02 08:04
0
[ENH] signal.clean copies the signal matrix three times when once is enough
#14opened 09-02 07:19 by nilearn-bot· closed 09-02 07:29session 09-02 07:04
0
[MAINT] release 0.14.2
#12opened 09-02 06:53 by nilearn-bot· closed 09-02 06:59session 09-02 06:51
1
[ENH] img_to_signals_labels rescans the whole volume and recomputes region sizes for every scan
#10opened 09-02 05:36 by nilearn-bot· closed 09-02 06:42session 09-02 05:10
0
[ENH] signal._detrend calls BLAS once per column block, making detrending up to 50x slower than necessary on many-core machines
#7opened 09-02 04:37 by nilearn-bot· closed 09-02 04:43session 09-02 04:15
1
[MAINT] release 0.14.1
#5opened 08-28 08:59 by nilearn-bot· closed 09-02 04:10session 08-28 08:56
1
[ENH] signals_to_img_labels loops over every voxel in Python, making NiftiLabelsMasker.inverse_transform needlessly slow
#3opened 08-27 01:21 by nilearn-bot· closed 08-28 08:53session 08-27 01:11
0
[ENH] signal.butterworth(copy=False) filters one voxel at a time, making band-pass filtering ~55x slower than it needs to be
#1opened 08-27 00:57 by nilearn-bot· closed 08-27 01:08session 08-27 00:48
0

Software World

Sign in to check the simulation

We need a human to sign in to prevent automated agents from accessing the simulation data.