Closed OverLordGoldDragon closed 3 years ago
:exclamation: No coverage uploaded for pull request base (
master@1ac7ed2
). Click here to learn what that means. The diff coverage isn/a
.
@@ Coverage Diff @@
## master #32 +/- ##
=========================================
Coverage ? 92.77%
=========================================
Files ? 17
Lines ? 3377
Branches ? 0
=========================================
Hits ? 3133
Misses ? 244
Partials ? 0
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 1ac7ed2...4cdcee2. Read the comment docs.
Dramatically speed up compute and reduce memory usage
FEATURES:
cwt, stft, ssq_cwt, ssq_stft
); see Performance guidessqueezepy.FFT
, supporting single- & multi-threaded CPU execution, and GPU execution, optionally viapyfftw
dtype='float32'
and'float64'
support forcwt, stft, ssq_cwt, ssq_stft, Wavelet
Wavelet.Psih(scale=, N=)
will store the computed wavelet(s) and, if subsequent calls have identicalscale
andN
, will return it directly without recomputing (significant speedup).BREAKING:
ftz
downsample
: 3 -> 4 inutils.cwt_utils.make_scales
EPS
deprecated in favor ofEPS32
&EPS64
for respective precisionsssq_cwt(flipud=True)
default now returnsTx = np.flipud(Tx)
relative to previous versionsssq_cwt
&ssq_stft
number of variables returned now depend onget_w, get_dWx
parameters; see docstringsdtype
defaults to'float32'
(can change viaconfigs.ini
); neithercwt
norstft
, for most applications, require extreme precision like filters do, so defaults should prioritize computeFIXES:
ssq_stft
would still defaultn_fft = len(x)
; defaulter line removed, delegated tostft
.ssqueezing
: improperly handled return ofinfer_scaletype
ssqueezing
:_get_center_frequency
computed atN
instead ofp2up(N)
withpadtype != None
MISC:
configs.ini
: added new configurable defaultsssqueezing.ssqueeze
: addedpadtype
arg (see FIXES)ssqueezing.ssqueeze
&ssq_cwt
: addedfind_closest_parallel
arg (see its docstring)utils.cwt_utils
:find_downsampling_scale
added argumentN
visuals.imshow
: removed default'interpolation' = 'none'
# Arguments:
docstring toWavelet
cwt
significantly sped up: 1) perWavelet
reuse; 2) rid ofifftshift
and*pn
(they undo each other); 3) eliminated redundant allocation invectorized
NOTES: