pynumdiff 0.2.4__tar.gz → 0.3__tar.gz

pynumdiff-0.3/.github/workflows/draft-pdf.yml

@@ -0,0 +1,19 @@
+on: [push]
+
+jobs:
+  paper:
+    runs-on: ubuntu-latest
+    name: Paper Draft
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v6
+      - name: Build draft PDF
+        uses: openjournals/openjournals-draft-action@master
+        with:
+          journal: joss
+          paper-path: paper/paper.md
+      - name: Upload
+        uses: actions/upload-artifact@v7
+        with:
+          name: paper
+          path: paper/paper.pdf

pynumdiff-0.3/CITATION.cff

@@ -0,0 +1,70 @@
+cff-version: 1.2.0
+message: "If you use this software, please cite the article under preferred-citation."
+title: PyNumDiff
+authors:
+  - family-names: Komarov
+    given-names: Pavel
+    orcid: "https://orcid.org/0009-0007-7482-2807"
+  - family-names: van Breugel
+    given-names: Floris
+    orcid: "https://orcid.org/0000-0001-6538-7179"
+  - family-names: Protogerou
+    given-names: Maria
+  - family-names: Kutz
+    given-names: J. Nathan
+    orcid: "https://orcid.org/0000-0002-6004-2275"
+version: 0.2.4
+date-released: 2026-08-29
+preferred-citation:
+  type: article
+  title: "PyNumDiff: Practical Numerical Differentiation for Noisy Data"
+  authors:
+    - family-names: Komarov
+      given-names: Pavel
+      orcid: "https://orcid.org/0009-0007-7482-2807"
+    - family-names: van Breugel
+      given-names: Floris
+      orcid: "https://orcid.org/0000-0001-6538-7179"
+    - family-names: Protogerou
+      given-names: Maria
+    - family-names: Kutz
+      given-names: J. Nathan
+      orcid: "https://orcid.org/0000-0002-6004-2275"
+  journal: Journal of Open Source Software
+  year: 2026
+  doi: 10.21105/joss.11172
+  status: submitted
+  notes: "Describes the package's second generation, through the v0.3 release series. Under review; the DOI is reserved and resolves on acceptance."
+references:
+  - type: article
+    title: "PyNumDiff: A Python package for numerical differentiation of noisy time-series data"
+    authors:
+      - family-names: van Breugel
+        given-names: Floris
+      - family-names: Liu
+        given-names: Yuying
+      - family-names: Brunton
+        given-names: Bingni W.
+      - family-names: Kutz
+        given-names: J. Nathan
+    journal: Journal of Open Source Software
+    year: 2022
+    volume: 7
+    number: 71
+    pages: 4078
+    doi: 10.21105/joss.04078
+    version: 0.1.2.4
+    notes: "The first-generation article, describing the package through version 0.1.x."
+  - type: article
+    title: A Taxonomy of Numerical Differentiation Methods
+    authors:
+      - family-names: Komarov
+        given-names: Pavel
+      - family-names: van Breugel
+        given-names: Floris
+      - family-names: Kutz
+        given-names: J. Nathan
+    year: 2025
+    status: submitted
+    url: "https://arxiv.org/abs/2512.09090"
+    notes: "Companion paper. Benchmarks the methods implemented here and provides their theoretical grounding."

{pynumdiff-0.2.4 → pynumdiff-0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pynumdiff
-Version: 0.2.4
+Version: 0.3
 Summary: pynumdiff: numerical derivatives in python
 Maintainer-email: Floris van Breugel <fvanbreugel@unr.edu>, Pavel Komarov <pvlkmrv@uw.edu>, Yuying Liu <yliu814@uw.edu>
 License-Expression: MIT
@@ -38,61 +38,62 @@ Dynamic: license-file
 
 # PyNumDiff
 
-Python methods for numerical differentiation of noisy data, including multi-objective optimization routines for automated parameter selection.
+Python methods for numerical differentiation and smoothing of noisy data, including automated hyperparameter selection.
 
 <p align="center">
   <a href="https://pynumdiff.readthedocs.io/master/">
-    <img alt="Python for Numerical Differentiation of noisy time series data" src="https://raw.githubusercontent.com/florisvb/PyNumDiff/master/logo.png" width="300" height="200" />
+    <img alt="Python for Numerical Differentiation of noisy time series data" src="https://raw.githubusercontent.com/florisvb/PyNumDiff/master/logo.png" width="300" />
   </a>
 </p>
 
 <p align="center">
-    <img src='https://github.com/florisvb/pynumdiff/actions/workflows/test.yml/badge.svg'/>
+    <a href="https://github.com/florisvb/PyNumDiff/actions">
+        <img src='https://github.com/florisvb/pynumdiff/actions/workflows/test.yml/badge.svg'/></a>
     <a href='https://pynumdiff.readthedocs.io/master/'>
         <img src='https://app.readthedocs.org/projects/pynumdiff/badge/?version=master' alt='Documentation Status' /></a>
     <a href='https://coveralls.io/github/florisvb/PyNumDiff?branch=master'>
         <img src='https://coveralls.io/repos/github/florisvb/PyNumDiff/badge.svg?branch=master' alt='Coverage Status' /></a>
     <a href="https://badge.fury.io/py/pynumdiff">
         <img src="https://badge.fury.io/py/pynumdiff.svg" alt="PyPI"></a>
-    <!--a href="https://doi.org/10.5281/zenodo.6374098">
-        <img src="https://zenodo.org/badge/DOI/10.5281/zenodo.6374098.svg" alt="DOI"></a-->
-    <a href="https://joss.theoj.org/papers/102257ee4b0142bf49bc18d7c810e9d5">
-        <img src="https://joss.theoj.org/papers/102257ee4b0142bf49bc18d7c810e9d5/status.svg"></a>
+    <a href="https://joss.theoj.org/papers/3b87180224e98705b4312cc35aaa615b">
+        <img src="https://joss.theoj.org/papers/3b87180224e98705b4312cc35aaa615b/status.svg"></a>
+    <a href="https://github.com/florisvb/PyNumDiff/blob/master/LICENSE.txt">
+        <img src="https://img.shields.io/badge/License-MIT-blue.svg"></a>
 </p>
 
 ## Introduction
 
-PyNumDiff is a Python package that implements many methods for computing numerical derivatives and smooth estimates from noisy data, which can be a critical step in developing dynamic models or designing control. There are seven different families of methods in this repository:
+PyNumDiff is a Python package that implements many methods for computing numerical derivatives and smooth estimates from noisy data, often a critical step in developing dynamic models or designing control. There are seven different families of methods in this repository:
 
 1. prefiltering followed by finite difference calculation
 2. iterated finite differencing
-3. polynomial fit methods
-4. basis function fit methods
+3. polynomial fits
+4. basis function fits
 5. total variation regularization of a finite difference derivative
 6. generalized Kalman smoothing
-7. local approximation with linear model
+7. local approximation with data-driven linear dynamics models
 
-All are ultimately smoothing with similar runtime and accuracy, but some have flexibility advantages over others, summarized in the table under [Usage](#usage) below. For further details and comparison, see section 7 of our [Taxonomy Paper](https://arxiv.org/abs/2512.09090).
+All ultimately regularize based on a smoothness prior, but the underlying models vary to suit particular assumptions: linear dynamics, frequency dependence, multiscale structure, piecewise polynomials, outliers, wrapping domain, etc. Runtimes scale linearly with data length, except for one which uses the FFT (but is still much faster than approaches using a convex solver). Accuracies tend to be broadly similar on generic data, although matching assumptions to the data generator can occasionally edge out competitors (e.g. linear dynamics for oscillators). Some models have flexibility advantages over others, like the ability to handle irregular data spacing, collated in the table under [Usage](#usage) below. For further details and comparison, see section 7 of our [Taxonomy Paper](https://arxiv.org/abs/2512.09090).
 
-All methods have hyperparameters, described in the [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). We take a principled approach and propose a multi-objective optimization framework for choosing settings that minimize a loss function that balances faithfulness to data with smoothness of the derivative estimate. For more details, refer to [this paper](https://doi.org/10.1109/ACCESS.2020.3034077).
+All methods have hyperparameters, described in the [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). We use a principled multi-objective optimization framework for choosing settings that minimize a loss function that balances faithfulness to data with smoothness of the derivative estimate. For more details, refer to [this paper](https://doi.org/10.1109/ACCESS.2020.3034077). Hyperparameter optimization runtime is primarily governed by search space dimension, which varies between 2 and 5 across methods.
+
+![Three simulated signals and their derivatives, estimated by six of the seven method families, with hyperparameters chosen by `pynumdiff.optimize`.](paper/methods_comparison.png)
 
 ## Installing
 
-Dependencies are listed in [pyproject.toml](https://github.com/florisvb/PyNumDiff/blob/master/pyproject.toml). They include the usual suspects like `numpy` and `scipy`, plus `pywavelets` for `waveletdiff`, `tqdm` for the optimizer, and `cvxpy` for `robustdiff` and `tvrdiff`.
+Dependencies are listed in [pyproject.toml](https://github.com/florisvb/PyNumDiff/blob/master/pyproject.toml). They include the usual suspects like `numpy` and `scipy`, plus `pywavelets` for `waveletdiff`, `tqdm` for monitoring optimization, and `cvxpy` for `tvrdiff`, `robustdiff`, and `lineardiff`.
 
 The code is compatible with >=Python 3.11. Install from PyPI with `pip install pynumdiff`, from source with `pip install git+https://github.com/florisvb/PyNumDiff`, or from local download with `pip install .`. Call `pip install pynumdiff[advanced]` to automatically install optional dependencies from the advanced list, like [CVXPY](https://www.cvxpy.org).
 
 ## Usage
 
 For more details, read our [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). The basic pattern of all differentiation methods is:
-
 ```python
-somethingdiff(x, dt, **kwargs)
+x_hat, dxdt_hat = somethingdiff(x, dt, **kwargs)
 ```
+where `x` is data, `dt` is a step size, and keyword arguments are hyperparameters which control behavior. Methods marked as able to handle multidimensional data have an `axis` argument to select which dimension of a block to differentiate along, and those supporting variable step size rename the second parameter `dt_or_t`, which accepts either a constant step size or an array of sample locations. Here is a summary of all major methods, indicating which situations they support:
 
-where `x` is data, `dt` is a step size, and various keyword arguments control the behavior. Methods marked multidimensional take an `axis` argument selecting which dimension of a block to differentiate along, and those supporting variable step size rename the second parameter `dt_or_t`, which accepts either a constant step size or an array of sample locations. Handing a method data it doesn't support raises a `ValueError` explaining why.
-
-| Method | Multidim | Variable step | Missing data | Outliers | Circular domain | Needs CVXPY |
+| Method | Multidim data | Variable step | Missing data | Outliers | Circular domain | Needs CVXPY |
 | --- | :-: | :-: | :-: | :-: | :-: | :-: |
 | `kerneldiff` | ✓ | | | | | |
 | `butterdiff` | ✓ | | | | | |
@@ -106,32 +107,23 @@ where `x` is data, `dt` is a step size, and various keyword arguments control th
 | `tvrdiff` | ✓ | | | ✓ | | ✓ |
 | `rtsdiff` | ✓ | ✓ | ✓ | | ✓ | |
 | `robustdiff` | ✓ | ✓ | ✓ | ✓ | | ✓ |
-| `lineardiff` | ✓ | | | | | ✓ |
-
-There is also presently a swathe of deprecated methods. Don't use them, but if you do you'll just get warnings telling you how to use whichever new-and-improved version. There are also a few minor methods kept for general interest (`iterative_velocity` and `smooth_acceleration`) but in practice dominated by or redundant with others from the table.
+| `lineardiff` | ✓ | ✓ | ✓ | | | ✓ |
 
-You can set the hyperparameters manually with a construction like:
-```python
-from pynumdiff.submodule import method
-
-x_hat, dxdt_hat = method(x, dt, param1=val1, param2=val2, ...)     
-```
+There are also a couple minor methods kept for general interest (`iterative_velocity` and `smooth_acceleration`) but in practice dominated by or redundant with others from the table.
 
-Or you can find hyperparameter settings by calling the multi-objective optimization algorithm from the `optimize` module:
+You can set the hyperparameters manually, or you can find hyperparameter settings by calling the multi-objective optimization algorithm from the `optimize` module:
 ```python
 from pynumdiff.optimize import optimize
 
-# estimate cutoff_frequency by (a) counting the number of true peaks per second in the data or (b) look at power spectra and choose cutoff
-tvgamma = np.exp(-1.6*np.log(cutoff_frequency) -0.71*np.log(dt) - 5.1) # see https://ieeexplore.ieee.org/abstract/document/9241009
-
-params, val = optimize(somethingdiff, x, dt, tvgamma=tvgamma, # smoothness hyperparameter which defaults to None if dxdt_truth given
-            dxdt_truth=None, # give ground truth data if available, in which case tvgamma goes unused
-            search_space_updates={'param1':[vals], 'param2':[vals], ...})
+# estimate bandlimit by (a) counting the number of true peaks per second in the data or (b) look at the power spectrum
+params, val = optimize(somethingdiff, x, dt, bandlimit=bandlimit, # smoothness hyper-parameter which defaults to None if dxdt_truth given
+        dxdt_truth=None, # give ground truth data if available, in which case bandlimit goes unused
+        search_space_updates={'param1':[vals], 'param2':{vals}, ...})
 
 print('Optimal parameters: ', params)
 x_hat, dxdt_hat = somethingdiff(x, dt, **params)
 ```
-`tvgamma` governs the smoothness targeted by the optimization procedure, with larger values yielding smoother derivatives. Its value is dependent upon sampling rate and frequency content of the underlying signal, and it is universal across methods, making it possible to compare results post optimization. A default search space is used to initialize and perform optimiation, defined at the top of `optimize.py`, with overwrites from `search_space_updates`. Be aware the optimization is a fairly heavy process.
+`bandlimit` governs the smoothness targeted by the optimization procedure, with smaller values yielding smoother derivatives. Its value is dependent upon frequency content of the underlying signal, and it is universal across methods, making it possible to compare results post optimization. A default search space is used to initialize and limit optimization, defined at the top of `optimize.py`, with overwrites passable via `search_space_updates`. Be aware optimization can be a fairly heavy process for some methods.
 
 ### Notebook examples
 
@@ -166,6 +158,21 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
 
 ### PyNumDiff python package:
 
+The second-generation article, describing the package through the v0.3 release series.
+
+    @article{PyNumDiff2026,
+      doi = {10.21105/joss.11172},
+      url = {https://doi.org/10.21105/joss.11172},
+      year = {2026},
+      publisher = {The Open Journal},
+      author = {Pavel Komarov and Floris van Breugel and Maria Protogerou and J. Nathan Kutz},
+      title = {PyNumDiff: Practical Numerical Differentiation for Noisy Data},
+      journal = {Journal of Open Source Software},
+      note = {In review}
+    }
+
+The first-generation article, describing the package through v0.1.x:
+
     @article{PyNumDiff2022,
       doi = {10.21105/joss.04078},
       url = {https://doi.org/10.21105/joss.04078},
@@ -181,7 +188,7 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
 
 ### Collection of numerical differentiation methods:
 
-    @misc{komarov2025taxonomynumericaldifferentiationmethods,
+    @misc{komarov2025taxonomy,
       title={A Taxonomy of Numerical Differentiation Methods},
       author={Pavel Komarov and Floris van Breugel and J. Nathan Kutz},
       year={2025},
@@ -191,15 +198,14 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
       url={https://arxiv.org/abs/2512.09090}
     }
 
-
 ### Optimization algorithm:
 
     @article{ParamOptimizationDerivatives2020, 
-    doi={10.1109/ACCESS.2020.3034077}
-    author={F. {van Breugel} and J. {Nathan Kutz} and B. W. {Brunton}}, 
-    journal={IEEE Access}, 
-    title={Numerical differentiation of noisy data: A unifying multi-objective optimization framework}, 
-    year={2020}
+      doi={10.1109/ACCESS.2020.3034077}
+      author={F. {van Breugel} and J. {Nathan Kutz} and B. W. {Brunton}},
+      journal={IEEE Access},
+      title={Numerical differentiation of noisy data: A unifying multi-objective optimization framework},
+      year={2020}
     }
 
 ## Running the tests
@@ -210,7 +216,6 @@ Run tests locally by navigating to the repo in a terminal and calling
 ```bash
 > pytest -s
 ```
-
 Add the flag `--plot` to see plots of the methods against test functions. Add the flag `--bounds` to print $\log$ error bounds (useful when changing method behavior).
 
 ## License

{pynumdiff-0.2.4 → pynumdiff-0.3}/README.md

@@ -1,60 +1,61 @@
 # PyNumDiff
 
-Python methods for numerical differentiation of noisy data, including multi-objective optimization routines for automated parameter selection.
+Python methods for numerical differentiation and smoothing of noisy data, including automated hyperparameter selection.
 
 <p align="center">
   <a href="https://pynumdiff.readthedocs.io/master/">
-    <img alt="Python for Numerical Differentiation of noisy time series data" src="https://raw.githubusercontent.com/florisvb/PyNumDiff/master/logo.png" width="300" height="200" />
+    <img alt="Python for Numerical Differentiation of noisy time series data" src="https://raw.githubusercontent.com/florisvb/PyNumDiff/master/logo.png" width="300" />
   </a>
 </p>
 
 <p align="center">
-    <img src='https://github.com/florisvb/pynumdiff/actions/workflows/test.yml/badge.svg'/>
+    <a href="https://github.com/florisvb/PyNumDiff/actions">
+        <img src='https://github.com/florisvb/pynumdiff/actions/workflows/test.yml/badge.svg'/></a>
     <a href='https://pynumdiff.readthedocs.io/master/'>
         <img src='https://app.readthedocs.org/projects/pynumdiff/badge/?version=master' alt='Documentation Status' /></a>
     <a href='https://coveralls.io/github/florisvb/PyNumDiff?branch=master'>
         <img src='https://coveralls.io/repos/github/florisvb/PyNumDiff/badge.svg?branch=master' alt='Coverage Status' /></a>
     <a href="https://badge.fury.io/py/pynumdiff">
         <img src="https://badge.fury.io/py/pynumdiff.svg" alt="PyPI"></a>
-    <!--a href="https://doi.org/10.5281/zenodo.6374098">
-        <img src="https://zenodo.org/badge/DOI/10.5281/zenodo.6374098.svg" alt="DOI"></a-->
-    <a href="https://joss.theoj.org/papers/102257ee4b0142bf49bc18d7c810e9d5">
-        <img src="https://joss.theoj.org/papers/102257ee4b0142bf49bc18d7c810e9d5/status.svg"></a>
+    <a href="https://joss.theoj.org/papers/3b87180224e98705b4312cc35aaa615b">
+        <img src="https://joss.theoj.org/papers/3b87180224e98705b4312cc35aaa615b/status.svg"></a>
+    <a href="https://github.com/florisvb/PyNumDiff/blob/master/LICENSE.txt">
+        <img src="https://img.shields.io/badge/License-MIT-blue.svg"></a>
 </p>
 
 ## Introduction
 
-PyNumDiff is a Python package that implements many methods for computing numerical derivatives and smooth estimates from noisy data, which can be a critical step in developing dynamic models or designing control. There are seven different families of methods in this repository:
+PyNumDiff is a Python package that implements many methods for computing numerical derivatives and smooth estimates from noisy data, often a critical step in developing dynamic models or designing control. There are seven different families of methods in this repository:
 
 1. prefiltering followed by finite difference calculation
 2. iterated finite differencing
-3. polynomial fit methods
-4. basis function fit methods
+3. polynomial fits
+4. basis function fits
 5. total variation regularization of a finite difference derivative
 6. generalized Kalman smoothing
-7. local approximation with linear model
+7. local approximation with data-driven linear dynamics models
 
-All are ultimately smoothing with similar runtime and accuracy, but some have flexibility advantages over others, summarized in the table under [Usage](#usage) below. For further details and comparison, see section 7 of our [Taxonomy Paper](https://arxiv.org/abs/2512.09090).
+All ultimately regularize based on a smoothness prior, but the underlying models vary to suit particular assumptions: linear dynamics, frequency dependence, multiscale structure, piecewise polynomials, outliers, wrapping domain, etc. Runtimes scale linearly with data length, except for one which uses the FFT (but is still much faster than approaches using a convex solver). Accuracies tend to be broadly similar on generic data, although matching assumptions to the data generator can occasionally edge out competitors (e.g. linear dynamics for oscillators). Some models have flexibility advantages over others, like the ability to handle irregular data spacing, collated in the table under [Usage](#usage) below. For further details and comparison, see section 7 of our [Taxonomy Paper](https://arxiv.org/abs/2512.09090).
 
-All methods have hyperparameters, described in the [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). We take a principled approach and propose a multi-objective optimization framework for choosing settings that minimize a loss function that balances faithfulness to data with smoothness of the derivative estimate. For more details, refer to [this paper](https://doi.org/10.1109/ACCESS.2020.3034077).
+All methods have hyperparameters, described in the [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). We use a principled multi-objective optimization framework for choosing settings that minimize a loss function that balances faithfulness to data with smoothness of the derivative estimate. For more details, refer to [this paper](https://doi.org/10.1109/ACCESS.2020.3034077). Hyperparameter optimization runtime is primarily governed by search space dimension, which varies between 2 and 5 across methods.
+
+![Three simulated signals and their derivatives, estimated by six of the seven method families, with hyperparameters chosen by `pynumdiff.optimize`.](paper/methods_comparison.png)
 
 ## Installing
 
-Dependencies are listed in [pyproject.toml](https://github.com/florisvb/PyNumDiff/blob/master/pyproject.toml). They include the usual suspects like `numpy` and `scipy`, plus `pywavelets` for `waveletdiff`, `tqdm` for the optimizer, and `cvxpy` for `robustdiff` and `tvrdiff`.
+Dependencies are listed in [pyproject.toml](https://github.com/florisvb/PyNumDiff/blob/master/pyproject.toml). They include the usual suspects like `numpy` and `scipy`, plus `pywavelets` for `waveletdiff`, `tqdm` for monitoring optimization, and `cvxpy` for `tvrdiff`, `robustdiff`, and `lineardiff`.
 
 The code is compatible with >=Python 3.11. Install from PyPI with `pip install pynumdiff`, from source with `pip install git+https://github.com/florisvb/PyNumDiff`, or from local download with `pip install .`. Call `pip install pynumdiff[advanced]` to automatically install optional dependencies from the advanced list, like [CVXPY](https://www.cvxpy.org).
 
 ## Usage
 
 For more details, read our [Sphinx documentation](https://pynumdiff.readthedocs.io/master/). The basic pattern of all differentiation methods is:
-
 ```python
-somethingdiff(x, dt, **kwargs)
+x_hat, dxdt_hat = somethingdiff(x, dt, **kwargs)
 ```
+where `x` is data, `dt` is a step size, and keyword arguments are hyperparameters which control behavior. Methods marked as able to handle multidimensional data have an `axis` argument to select which dimension of a block to differentiate along, and those supporting variable step size rename the second parameter `dt_or_t`, which accepts either a constant step size or an array of sample locations. Here is a summary of all major methods, indicating which situations they support:
 
-where `x` is data, `dt` is a step size, and various keyword arguments control the behavior. Methods marked multidimensional take an `axis` argument selecting which dimension of a block to differentiate along, and those supporting variable step size rename the second parameter `dt_or_t`, which accepts either a constant step size or an array of sample locations. Handing a method data it doesn't support raises a `ValueError` explaining why.
-
-| Method | Multidim | Variable step | Missing data | Outliers | Circular domain | Needs CVXPY |
+| Method | Multidim data | Variable step | Missing data | Outliers | Circular domain | Needs CVXPY |
 | --- | :-: | :-: | :-: | :-: | :-: | :-: |
 | `kerneldiff` | ✓ | | | | | |
 | `butterdiff` | ✓ | | | | | |
@@ -68,32 +69,23 @@ where `x` is data, `dt` is a step size, and various keyword arguments control th
 | `tvrdiff` | ✓ | | | ✓ | | ✓ |
 | `rtsdiff` | ✓ | ✓ | ✓ | | ✓ | |
 | `robustdiff` | ✓ | ✓ | ✓ | ✓ | | ✓ |
-| `lineardiff` | ✓ | | | | | ✓ |
-
-There is also presently a swathe of deprecated methods. Don't use them, but if you do you'll just get warnings telling you how to use whichever new-and-improved version. There are also a few minor methods kept for general interest (`iterative_velocity` and `smooth_acceleration`) but in practice dominated by or redundant with others from the table.
+| `lineardiff` | ✓ | ✓ | ✓ | | | ✓ |
 
-You can set the hyperparameters manually with a construction like:
-```python
-from pynumdiff.submodule import method
-
-x_hat, dxdt_hat = method(x, dt, param1=val1, param2=val2, ...)     
-```
+There are also a couple minor methods kept for general interest (`iterative_velocity` and `smooth_acceleration`) but in practice dominated by or redundant with others from the table.
 
-Or you can find hyperparameter settings by calling the multi-objective optimization algorithm from the `optimize` module:
+You can set the hyperparameters manually, or you can find hyperparameter settings by calling the multi-objective optimization algorithm from the `optimize` module:
 ```python
 from pynumdiff.optimize import optimize
 
-# estimate cutoff_frequency by (a) counting the number of true peaks per second in the data or (b) look at power spectra and choose cutoff
-tvgamma = np.exp(-1.6*np.log(cutoff_frequency) -0.71*np.log(dt) - 5.1) # see https://ieeexplore.ieee.org/abstract/document/9241009
-
-params, val = optimize(somethingdiff, x, dt, tvgamma=tvgamma, # smoothness hyperparameter which defaults to None if dxdt_truth given
-            dxdt_truth=None, # give ground truth data if available, in which case tvgamma goes unused
-            search_space_updates={'param1':[vals], 'param2':[vals], ...})
+# estimate bandlimit by (a) counting the number of true peaks per second in the data or (b) look at the power spectrum
+params, val = optimize(somethingdiff, x, dt, bandlimit=bandlimit, # smoothness hyper-parameter which defaults to None if dxdt_truth given
+        dxdt_truth=None, # give ground truth data if available, in which case bandlimit goes unused
+        search_space_updates={'param1':[vals], 'param2':{vals}, ...})
 
 print('Optimal parameters: ', params)
 x_hat, dxdt_hat = somethingdiff(x, dt, **params)
 ```
-`tvgamma` governs the smoothness targeted by the optimization procedure, with larger values yielding smoother derivatives. Its value is dependent upon sampling rate and frequency content of the underlying signal, and it is universal across methods, making it possible to compare results post optimization. A default search space is used to initialize and perform optimiation, defined at the top of `optimize.py`, with overwrites from `search_space_updates`. Be aware the optimization is a fairly heavy process.
+`bandlimit` governs the smoothness targeted by the optimization procedure, with smaller values yielding smoother derivatives. Its value is dependent upon frequency content of the underlying signal, and it is universal across methods, making it possible to compare results post optimization. A default search space is used to initialize and limit optimization, defined at the top of `optimize.py`, with overwrites passable via `search_space_updates`. Be aware optimization can be a fairly heavy process for some methods.
 
 ### Notebook examples
 
@@ -128,6 +120,21 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
 
 ### PyNumDiff python package:
 
+The second-generation article, describing the package through the v0.3 release series.
+
+    @article{PyNumDiff2026,
+      doi = {10.21105/joss.11172},
+      url = {https://doi.org/10.21105/joss.11172},
+      year = {2026},
+      publisher = {The Open Journal},
+      author = {Pavel Komarov and Floris van Breugel and Maria Protogerou and J. Nathan Kutz},
+      title = {PyNumDiff: Practical Numerical Differentiation for Noisy Data},
+      journal = {Journal of Open Source Software},
+      note = {In review}
+    }
+
+The first-generation article, describing the package through v0.1.x:
+
     @article{PyNumDiff2022,
       doi = {10.21105/joss.04078},
       url = {https://doi.org/10.21105/joss.04078},
@@ -143,7 +150,7 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
 
 ### Collection of numerical differentiation methods:
 
-    @misc{komarov2025taxonomynumericaldifferentiationmethods,
+    @misc{komarov2025taxonomy,
       title={A Taxonomy of Numerical Differentiation Methods},
       author={Pavel Komarov and Floris van Breugel and J. Nathan Kutz},
       year={2025},
@@ -153,15 +160,14 @@ See CITATION.cff file, but here are some possible BibTeX entries for convenience
       url={https://arxiv.org/abs/2512.09090}
     }
 
-
 ### Optimization algorithm:
 
     @article{ParamOptimizationDerivatives2020, 
-    doi={10.1109/ACCESS.2020.3034077}
-    author={F. {van Breugel} and J. {Nathan Kutz} and B. W. {Brunton}}, 
-    journal={IEEE Access}, 
-    title={Numerical differentiation of noisy data: A unifying multi-objective optimization framework}, 
-    year={2020}
+      doi={10.1109/ACCESS.2020.3034077}
+      author={F. {van Breugel} and J. {Nathan Kutz} and B. W. {Brunton}},
+      journal={IEEE Access},
+      title={Numerical differentiation of noisy data: A unifying multi-objective optimization framework},
+      year={2020}
     }
 
 ## Running the tests
@@ -172,7 +178,6 @@ Run tests locally by navigating to the repo in a terminal and calling
 ```bash
 > pytest -s
 ```
-
 Add the flag `--plot` to see plots of the methods against test functions. Add the flag `--bounds` to print $\log$ error bounds (useful when changing method behavior).
 
 ## License

pynumdiff-0.3/docs/source/finite_difference.rst

@@ -0,0 +1,7 @@
+finite_difference
+=================
+
+.. automodule:: pynumdiff.finite_difference
+    :no-members:
+
+.. autofunction:: pynumdiff.finite_difference.finitediff

{pynumdiff-0.2.4 → pynumdiff-0.3}/docs/source/kalman_smooth.rst

@@ -6,9 +6,6 @@ kalman_smooth
 
 .. autofunction:: pynumdiff.kalman_smooth.rtsdiff
 .. autofunction:: pynumdiff.kalman_smooth.robustdiff
-.. autofunction:: pynumdiff.kalman_smooth.constant_velocity
-.. autofunction:: pynumdiff.kalman_smooth.constant_acceleration
-.. autofunction:: pynumdiff.kalman_smooth.constant_jerk
 .. autofunction:: pynumdiff.kalman_smooth.kalman_filter
 .. autofunction:: pynumdiff.kalman_smooth.rts_smooth
 .. autofunction:: pynumdiff.kalman_smooth.convex_smooth

pynumdiff-0.3/docs/source/smooth_finite_difference.rst

@@ -0,0 +1,8 @@
+smooth_finite_difference
+========================
+
+.. automodule:: pynumdiff.smooth_finite_difference
+    :no-members:
+
+.. autofunction:: pynumdiff.smooth_finite_difference.kerneldiff
+.. autofunction:: pynumdiff.smooth_finite_difference.butterdiff

{pynumdiff-0.2.4 → pynumdiff-0.3}/docs/source/total_variation_regularization.rst

@@ -5,8 +5,5 @@ total_variation_regularization
    :no-members:
 
 .. autofunction:: pynumdiff.total_variation_regularization.tvrdiff
-.. autofunction:: pynumdiff.total_variation_regularization.velocity
-.. autofunction:: pynumdiff.total_variation_regularization.acceleration
-.. autofunction:: pynumdiff.total_variation_regularization.jerk
 .. autofunction:: pynumdiff.total_variation_regularization.iterative_velocity
 .. autofunction:: pynumdiff.total_variation_regularization.smooth_acceleration