Enhance _cwt.py by introducing a configurable hop size parameter #804

phandangthoai · 2025-05-26T10:47:24Z

Generating a scalogram from the full output of the Continuous Wavelet Transform (CWT) entails high computational cost while providing limited performance gains in acoustic recognition models based on deep learning. Therefore, this update proposes reducing the output size during the intermediate processing stage—rather than after CWT generation—to improve computational efficiency of CWT. This pull request reflects the research findings presented in the following paper.

Phan, D. T., Huynh, T. A., Pham, V. T., Tran, C. M., Mai, V. T., & Tran, N. Q. (2025). Optimal Scalogram for Computational Complexity Reduction in Acoustic Recognition Using Deep Learning. arXiv preprint arXiv:2505.13017.

Generating a scalogram from the full output of the Continuous Wavelet Transform (CWT) entails high computational cost while providing limited performance gains in acoustic recognition models based on deep learning. Therefore, this update proposes reducing the output size during the intermediate processing stage—rather than after CWT generation—to improve computational efficiency of CWT. This pull request reflects the research findings presented in the following paper. Phan, D. T., Huynh, T. A., Pham, V. T., Tran, C. M., Mai, V. T., & Tran, N. Q. (2025). Optimal Scalogram for Computational Complexity Reduction in Acoustic Recognition Using Deep Learning. arXiv preprint arXiv:2505.13017.

rgommers

Thanks @phandangthoai. A few quick review comments inline. This change also needs tests, and it would be good to demonstrate what the gain of this is with a small benchmark.

pywt/_cwt.py

phandangthoai · 2025-05-27T18:27:53Z

Hello @rgommers,
Thanks for your quick reply with the guidelines. I am in preparation to response to your review comments accordingly.

Update _cwt.py

phandangthoai · 2025-07-16T13:52:30Z

Hello @rgommers,
Please see my responses below. I appreciate your feedback, which helps improve my pull request.

I have updated the code, and all tests have passed successfully.
The proposed method (optimal scalogram) is validated and benchmarked against the traditional approach (normal scalogram) using the MIMII dataset. The results indicate that a minor reduction in model performance leads to a substantial improvement in computational efficiency—achieving a speedup of nearly eight times, as illustrated in the figure below.

rgommers

Thanks @phandangthoai. A few inline comments. In addtion, could you please:

Share a benchmark script that shows the performance improvement
Add a References section in the docstring to the paper in numpydoc format

rgommers · 2025-07-17T08:03:13Z

pywt/_cwt.py

@@ -24,9 +24,9 @@ def next_fast_len(n):
    return 2**ceil(np.log2(n))


-def cwt(data, scales, wavelet, sampling_period=1., method='conv', axis=-1):
+def cwt(data, scales, wavelet, hop_size=1, sampling_period=1., method='conv', axis=-1):


Can you please move the new keyword to the end of the signature? It's not backwards-compatible to add a new keyword in the middle. That would break calls like cwt(data, scales, wavelet, 2.5) (2.5 was the sampling period).

Also, please add it as *, hop_size=1 (the * means keyword-only).

The examples and tests then need to be updated for that change.

New key word hop_size is move to the end, and defined as keyword-only.

The examples and tests are revised accordingly and passed.

pywt/_cwt.py

Move hop_size to the end

Add References to the research paper.

Revise the notes

phandangthoai · 2025-07-19T10:13:52Z

Hello @rgommers ,

Thank you for your valuable feedback. I have updated the pull request in accordance with your comments.

The script for generating the optimal scalogram and the script for generating the standard scalogram have been included for reference.

A References section citing the relevant paper has been added to the function docstring, following the numpydoc format.

Please let me know if any further revisions are required.

Thank you again for your time and guidance.

Best regards,
Dang Thoai Phan

phandangthoai · 2025-08-01T13:05:27Z

Hello @rgommers ,

I have addressed the trailing whitespace issue reported by the style check tool in my latest commit. However, it appears that the pre-commit style check did not re-trigger automatically. Could you kindly re-run it or let me know if there is anything further I need to do on my end?

Best regards,
Dang Thoai Phan

rgommers · 2025-08-03T20:22:48Z

This accumulated a merge conflict, could you please merge with main to address that?

I've triggered the CI to run. The added tests only use hop_size=1, is that realistic enough? Seems like you'd want to use something higher as well.

Fix trailing whitespaces error

rgommers · 2025-08-04T10:25:25Z

The merge conflict is still unresolved, so I won't retrigger CI now.

If you want to find a trivial typo or one word change somewhere that I can merge immediately in a separate PR, then CI jobs won't need manual approval from me after that.

Test for hop_size greater than 1.

Resolve conflicts when merging to main

…-optimal-scalogram

…alogram 2 merge main into optimal scalogram

Test for hop size greater than 1

phandangthoai · 2025-08-05T07:55:43Z

Hello @rgommers ,

I have completed the following updates to the PR:

Resolved merge conflicts
Updated the test case to include a hop size greater than 1
Fixed trailing whitespace issues

Could you please review the changes and proceed with the merge into main?
Thank you for your time and support.

Best regards,
Dang Thoai Phan

Fix doctest error

phandangthoai · 2025-08-05T10:07:25Z

Hello @rgommers,

All checks have passed, and I am thrilled to be at the stage of having my first official PR ready for merge.
I would greatly appreciate it if you could review the changes and proceed with the merge at your convenience.

Thank you for your time and support.

Best regards,
Dang Thoai Phan

phandangthoai · 2025-08-23T08:26:26Z

Hello @rgommers,

I hope you are doing well. I would appreciate it if you could take some time to review my pull request and merge it into the main branch if there are no further issues.
Thank you for your time and support.

Best regards,
Dang Thoai Phan

rgommers

Thanks @phandangthoai, added some more comments.

rgommers · 2025-08-28T04:23:05Z

pywt/_cwt.py

-def cwt(data, scales, wavelet, sampling_period=1., method='conv', axis=-1,
-        *, precision=12):
+def cwt(data, scales, wavelet, sampling_period=1., method='conv', axis=-1, precision=12,
+        *, hop_size=1):


This is still not quite right. You changed precision from a keyword-only argument to one that can also be positional.

rgommers · 2025-08-28T04:23:36Z

pywt/_cwt.py

@@ -65,6 +66,14 @@ def cwt(data, scales, wavelet, sampling_period=1., method='conv', axis=-1,
        compute a bit slower. Too low will distort coefficients and their
        norms, with a zipper-like effect. The default is 12, it's recommended
        to use >=12.
+    hop_size : int


int, optional as the type specifier

rgommers · 2025-08-28T04:29:28Z

pywt/_cwt.py

@@ -89,7 +104,7 @@ def cwt(data, scales, wavelet, sampling_period=1., method='conv', axis=-1,
    >>> x = np.exp(np.linspace(0, 2, 512))
    >>> y = np.cos(2*np.pi*x)  # exponential chirp
    >>> scales = np.logspace(np.log10(1), np.log10(128), 128)
-    >>> coef, freqs = pywt.cwt(y, scales, 'gaus1')
+    >>> coef, freqs = pywt.cwt(y, scales, 'gaus1', hop_size=16)


It'd be better not to change the existing examples, but rather to add a new one at the end of the Examples section. And above it, explain what hopsize actually does: subsample, so the computation is faster, for limited reduction in accuracy. Two images below each other, the first one with the full accuracy and the second one with reduced accuracy, allow the reader to then actually see the effect of using hopsize.

rgommers · 2025-08-28T04:31:05Z

pywt/tests/test_cwt_wavelets.py

@@ -401,6 +402,7 @@ def test_cwt_batch(axis, method):
    dtype = np.float64
    time, sst = pywt.data.nino()
    n_batch = 8
+    hop_size = 1


This change does nothing, as do the hopsize=hopsize additions below it. Can you drop those changes and rely on the default value?

rgommers · 2025-08-28T04:33:38Z

pywt/tests/test_cwt_wavelets.py

@@ -379,9 +379,10 @@ def test_cwt_complex(dtype, tol, method):
    dt = time[1] - time[0]
    wavelet = 'cmor1.5-1.0'
    scales = np.arange(1, 32)
+    hop_size = 128


Surely the right way to test that hopsize actually does something reasonable is to use cwt(...) on some data, use cwt(..., hopsize=4) (or whatever value makes sense) on that same data, and then assert that (a) the output array has the expected 4x smaller size, and that the output values or some measure of accuracy of it are within some tolerance value?

rgommers requested changes May 26, 2025

View reviewed changes

pywt/_cwt.py Outdated Show resolved Hide resolved

pywt/_cwt.py Outdated Show resolved Hide resolved

phandangthoai added 15 commits June 1, 2025 12:39

Update document for hop_size in _cwt.py

9a8bf1a

Update _cwt.py to fix downsampled_length error

ad41521

Update _cwt.py with hop_size as integer

6b7d7f1

Fix fft lib for _cwt

3c9289b

Validation for hop_size in _cwt.py

0a53e69

Add hop_size to test cases of _cwt

6776f98

Update test_cwt_wavelets.py

82a4ef0

Fix shape of output if data in n-dim

81cd1bb

Update _cwt.py

Update _cwt.py with reshape the coef

dc93c83

Update _cwt.py to fix coef shape

a8a9aee

Update _cwt.py for fixing coef size

f607aa0

Update _cwt.py

5cda18e

Update _cwt.py

6f4e57a

Update _cwt.py with correct sampling

099f895

Update _cwt.py

7282d9f

phandangthoai requested a review from rgommers July 16, 2025 13:54

rgommers reviewed Jul 17, 2025

View reviewed changes

phandangthoai added 4 commits July 19, 2025 10:05

Update _cwt.py

20838a7

Move hop_size to the end

Update test_cwt_wavelets.py

d986733

Move hop_size to the end

Update _cwt.py

5eed6aa

Add References to the research paper.

Update _cwt.py

8189d35

Revise the notes

phandangthoai requested a review from rgommers July 19, 2025 10:14

rgommers added the enhancement label Jul 29, 2025

Fix trailing whitespace to pass pre-commit hook

85482d4

Update _cwt.py

bd5134b

Fix trailing whitespaces error

phandangthoai added 15 commits August 4, 2025 12:26

Update test_cwt_wavelets.py

ce87862

Test for hop_size greater than 1.

Update _cwt.py

fc90fec

Resolve conflicts when merging to main

Merge branch 'phandangthoai-optimal-scalogram' into 2-merge-main-into…

20d2af4

…-optimal-scalogram

Merge pull request #1 from phandangthoai/2-merge-main-into-optimal-sc…

f6fa074

…alogram 2 merge main into optimal scalogram

Update test_cwt_wavelets.py

84c3fcb

Update test_cwt_wavelets.py

f4231a1

Update test_cwt_wavelets.py

fe2a9f3

Test for hop size greater than 1

Update test_cwt_wavelets.py

fef6593

Test for hop size greater than 1

Update test_cwt_wavelets.py

f5640ae

Update test_cwt_wavelets.py

368f272

Update test_cwt_wavelets.py

9c2b414

Update test_cwt_wavelets.py

00fb8d4

Update test_cwt_wavelets.py

c7da227

Update test_cwt_wavelets.py

79acb20

Update test_cwt_wavelets.py

8996c80

phandangthoai mentioned this pull request Aug 5, 2025

MAINT: a tiny edit rewording a single code code comment #818

Merged

Update _cwt.py

746d569

Fix doctest error

rgommers reviewed Aug 28, 2025

View reviewed changes

Update _cwt.py

0f0ffc6

Uh oh!

Enhance _cwt.py by introducing a configurable hop size parameter #804

Are you sure you want to change the base?

Enhance _cwt.py by introducing a configurable hop size parameter #804

Uh oh!

Conversation

phandangthoai commented May 26, 2025

Uh oh!

rgommers left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

phandangthoai commented May 27, 2025

Uh oh!

phandangthoai commented Jul 16, 2025

Uh oh!

rgommers left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

phandangthoai commented Jul 19, 2025

Uh oh!

phandangthoai commented Aug 1, 2025

Uh oh!

rgommers commented Aug 3, 2025

Uh oh!

rgommers commented Aug 4, 2025

Uh oh!

phandangthoai commented Aug 5, 2025

Uh oh!

phandangthoai commented Aug 5, 2025

Uh oh!

phandangthoai commented Aug 23, 2025

Uh oh!

rgommers left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!