fastersom 0.1.0__tar.gz

fastersom-0.1.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Darius Schaub
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

fastersom-0.1.0/PKG-INFO

@@ -0,0 +1,98 @@
+Metadata-Version: 2.4
+Name: fastersom
+Version: 0.1.0
+Summary: Fast batch self-organising maps (SOM) in PyTorch for GPU and CPU
+Keywords: self-organizing map,som,kohonen,clustering,pytorch,gpu,spatial omics
+Author: Darius Schaub
+Author-email: Darius Schaub <schaub.darius@gmail.com>
+License-Expression: Apache-2.0
+License-File: LICENSE
+Classifier: Development Status :: 3 - Alpha
+Classifier: Intended Audience :: Science/Research
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Typing :: Typed
+Requires-Dist: numpy>=1.24
+Requires-Dist: torch>=2.3
+Requires-Python: >=3.11
+Project-URL: Homepage, https://github.com/dschaub95/fastersom
+Project-URL: Issues, https://github.com/dschaub95/fastersom/issues
+Project-URL: Changelog, https://github.com/dschaub95/fastersom/blob/main/CHANGELOG.md
+Description-Content-Type: text/markdown
+
+# fastersom
+
+A small, fast batch self-organising map (SOM) in PyTorch. It implements the same algorithm and defaults as
+[XPySom](https://github.com/Manciukic/xpysom) as configured by [spatiomic](https://github.com/complextissue/spatiomic)
+(`so.dimension.som`): rectangular grid, gaussian neighbourhood, exponential sigma decay, XPySom's seeded initialisation.
+Data stays on the device (CUDA, MPS or CPU) and the per-epoch update uses per-node sums, so an epoch costs one BMU
+search plus a (nodes x nodes) matrix product.
+
+## Install
+
+```bash
+uv add fastersom    # or: pip install fastersom
+```
+
+## Usage
+
+```python
+import numpy as np
+from fastersom import SOM
+
+X = np.load("pixels.npy")  # (n, d) or (..., d); numpy, array-like or torch
+som = SOM(n_rows=30, n_cols=30, n_epochs=10, random_state=0)  # device: cuda > mps > cpu
+som.fit(X)
+som.weights_  # (900, d) node weights, node = row * n_cols + col
+som.labels_  # BMU index per training sample
+labels = som.predict(X)  # BMU index per sample
+dist = som.transform(X)  # (n, 900) distance to every node
+print(som.quantization_error(X), som.topographic_error(X))
+```
+
+Options:
+
+- `metric="cosine"` finds BMUs by cosine similarity. Zero-norm samples have no BMU, get label -1 and are left out of
+  the update and the error measures.
+- `init="sample"` initialises the nodes from random samples instead of XPySom's random unit vectors.
+- `sigma_initial` and `sigma_final` set the neighbourhood width schedule; the defaults reproduce spatiomic.
+- `batch_size` bounds memory (about `batch_size x n_nodes` floats) without changing the algorithm.
+
+`BatchSOM` is an alias for `SOM`.
+
+## Benchmark
+
+480,000 x 34 pixels, 30 x 30 nodes, 10 epochs, fit time only (mean of seeds 0-2):
+
+| implementation               | hardware          | fit time | quantization error |
+| ---------------------------- | ----------------- | -------- | ------------------ |
+| fastersom                    | NVIDIA A100       | 0.085 s  | 0.169              |
+| spatiomic (XPySom, CuPy)     | NVIDIA A100       | 0.46 s   | 0.169              |
+| spatiomic (XPySom, NumPy)    | 256-core CPU      | 40 s     | 0.169              |
+
+## Relation to XPySom and spatiomic
+
+fastersom is a reimplementation in PyTorch of the batch SOM in XPySom (Mancini et al.), which itself follows
+Kohonen's batch map. Starting from the same weights, one epoch matches XPySom to ~1e-7, for both the euclidean and
+the cosine metric. XPySom's learning rate is omitted because it cancels in the batch update. XPySom's `sigma` is
+twice fastersom's `sigma_initial`, because XPySom scales the gaussian width by `std_coeff=0.5`. Only the rectangular
+topology, gaussian neighbourhood and exponential decay are supported.
+
+## Development
+
+```bash
+uv sync                      # installs the dev group: pytest, ruff, pre-commit, twine, xpysom
+uv run pre-commit install    # ruff, uv-lock and file checks on every commit
+uv run pytest
+uv build && uv run twine check --strict dist/*
+```
+
+## License
+
+Apache-2.0

fastersom-0.1.0/README.md

@@ -0,0 +1,70 @@
+# fastersom
+
+A small, fast batch self-organising map (SOM) in PyTorch. It implements the same algorithm and defaults as
+[XPySom](https://github.com/Manciukic/xpysom) as configured by [spatiomic](https://github.com/complextissue/spatiomic)
+(`so.dimension.som`): rectangular grid, gaussian neighbourhood, exponential sigma decay, XPySom's seeded initialisation.
+Data stays on the device (CUDA, MPS or CPU) and the per-epoch update uses per-node sums, so an epoch costs one BMU
+search plus a (nodes x nodes) matrix product.
+
+## Install
+
+```bash
+uv add fastersom    # or: pip install fastersom
+```
+
+## Usage
+
+```python
+import numpy as np
+from fastersom import SOM
+
+X = np.load("pixels.npy")  # (n, d) or (..., d); numpy, array-like or torch
+som = SOM(n_rows=30, n_cols=30, n_epochs=10, random_state=0)  # device: cuda > mps > cpu
+som.fit(X)
+som.weights_  # (900, d) node weights, node = row * n_cols + col
+som.labels_  # BMU index per training sample
+labels = som.predict(X)  # BMU index per sample
+dist = som.transform(X)  # (n, 900) distance to every node
+print(som.quantization_error(X), som.topographic_error(X))
+```
+
+Options:
+
+- `metric="cosine"` finds BMUs by cosine similarity. Zero-norm samples have no BMU, get label -1 and are left out of
+  the update and the error measures.
+- `init="sample"` initialises the nodes from random samples instead of XPySom's random unit vectors.
+- `sigma_initial` and `sigma_final` set the neighbourhood width schedule; the defaults reproduce spatiomic.
+- `batch_size` bounds memory (about `batch_size x n_nodes` floats) without changing the algorithm.
+
+`BatchSOM` is an alias for `SOM`.
+
+## Benchmark
+
+480,000 x 34 pixels, 30 x 30 nodes, 10 epochs, fit time only (mean of seeds 0-2):
+
+| implementation               | hardware          | fit time | quantization error |
+| ---------------------------- | ----------------- | -------- | ------------------ |
+| fastersom                    | NVIDIA A100       | 0.085 s  | 0.169              |
+| spatiomic (XPySom, CuPy)     | NVIDIA A100       | 0.46 s   | 0.169              |
+| spatiomic (XPySom, NumPy)    | 256-core CPU      | 40 s     | 0.169              |
+
+## Relation to XPySom and spatiomic
+
+fastersom is a reimplementation in PyTorch of the batch SOM in XPySom (Mancini et al.), which itself follows
+Kohonen's batch map. Starting from the same weights, one epoch matches XPySom to ~1e-7, for both the euclidean and
+the cosine metric. XPySom's learning rate is omitted because it cancels in the batch update. XPySom's `sigma` is
+twice fastersom's `sigma_initial`, because XPySom scales the gaussian width by `std_coeff=0.5`. Only the rectangular
+topology, gaussian neighbourhood and exponential decay are supported.
+
+## Development
+
+```bash
+uv sync                      # installs the dev group: pytest, ruff, pre-commit, twine, xpysom
+uv run pre-commit install    # ruff, uv-lock and file checks on every commit
+uv run pytest
+uv build && uv run twine check --strict dist/*
+```
+
+## License
+
+Apache-2.0

fastersom-0.1.0/pyproject.toml

@@ -0,0 +1,91 @@
+[project]
+name = "fastersom"
+version = "0.1.0"
+description = "Fast batch self-organising maps (SOM) in PyTorch for GPU and CPU"
+readme = "README.md"
+license = "Apache-2.0"
+license-files = ["LICENSE"]
+requires-python = ">=3.11"
+keywords = [
+    "self-organizing map",
+    "som",
+    "kohonen",
+    "clustering",
+    "pytorch",
+    "gpu",
+    "spatial omics",
+]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Intended Audience :: Science/Research",
+    "Operating System :: OS Independent",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Bio-Informatics",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Typing :: Typed",
+]
+dependencies = [
+    "numpy>=1.24",
+    "torch>=2.3",
+]
+
+[[project.authors]]
+name = "Darius Schaub"
+email = "schaub.darius@gmail.com"
+
+[project.urls]
+Homepage = "https://github.com/dschaub95/fastersom"
+Issues = "https://github.com/dschaub95/fastersom/issues"
+Changelog = "https://github.com/dschaub95/fastersom/blob/main/CHANGELOG.md"
+
+[build-system]
+requires = ["uv_build>=0.13.4,<0.14.0"]
+build-backend = "uv_build"
+
+[dependency-groups]
+dev = [
+    "pre-commit>=4.6.2",
+    "pytest>=9.1.1",
+    "ruff>=0.17.1",
+    "twine>=7.0.0",
+    "xpysom>=1.0.7",
+]
+
+[tool.pytest]
+minversion = "9.0"
+testpaths = ["tests"]
+addopts = [
+    "-ra",
+    "--import-mode=importlib",
+]
+strict = true
+filterwarnings = ["error"]
+
+[tool.ruff]
+line-length = 120
+src = [
+    "src",
+    "tests",
+]
+
+[tool.ruff.lint]
+select = [
+    "E",
+    "W",
+    "F",
+    "I",
+    "B",
+    "UP",
+    "SIM",
+    "C4",
+    "NPY",
+    "PT",
+    "RUF",
+]
+
+[tool.ruff.lint.per-file-ignores]
+"tests/**" = ["C408"]

fastersom-0.1.0/pyproject.toml.orig

@@ -0,0 +1,62 @@
+[project]
+name = "fastersom"
+version = "0.1.0"
+description = "Fast batch self-organising maps (SOM) in PyTorch for GPU and CPU"
+readme = "README.md"
+authors = [{ name = "Darius Schaub", email = "schaub.darius@gmail.com" }]
+license = "Apache-2.0"
+license-files = ["LICENSE"]
+requires-python = ">=3.11"
+keywords = ["self-organizing map", "som", "kohonen", "clustering", "pytorch", "gpu", "spatial omics"]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Intended Audience :: Science/Research",
+    "Operating System :: OS Independent",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Bio-Informatics",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Typing :: Typed",
+]
+dependencies = [
+    "numpy>=1.24",
+    "torch>=2.3",  # first release built against NumPy 2
+]
+
+[project.urls]
+Homepage = "https://github.com/dschaub95/fastersom"
+Issues = "https://github.com/dschaub95/fastersom/issues"
+Changelog = "https://github.com/dschaub95/fastersom/blob/main/CHANGELOG.md"
+
+[build-system]
+requires = ["uv_build>=0.13.4,<0.14.0"]
+build-backend = "uv_build"
+
+[dependency-groups]
+dev = [
+    "pre-commit>=4.6.2",
+    "pytest>=9.1.1",
+    "ruff>=0.17.1",
+    "twine>=7.0.0",
+    "xpysom>=1.0.7",
+]
+
+[tool.pytest]
+minversion = "9.0"
+testpaths = ["tests"]
+addopts = ["-ra", "--import-mode=importlib"]
+strict = true
+filterwarnings = ["error"]
+
+[tool.ruff]
+line-length = 120
+src = ["src", "tests"]
+
+[tool.ruff.lint]
+select = ["E", "W", "F", "I", "B", "UP", "SIM", "C4", "NPY", "PT", "RUF"]
+
+[tool.ruff.lint.per-file-ignores]
+"tests/**" = ["C408"]

fastersom-0.1.0/src/fastersom/__init__.py

@@ -0,0 +1,8 @@
+"""Fast batch self-organising maps in PyTorch."""
+
+from importlib.metadata import version
+
+from ._som import SOM, BatchSOM
+
+__all__ = ["SOM", "BatchSOM"]
+__version__ = version("fastersom")

fastersom-0.1.0/src/fastersom/_som.py

@@ -0,0 +1,258 @@
+"""Batch self-organising map (Kohonen's batch map) in PyTorch.
+
+Each epoch assigns every sample x_i to its best matching unit c(x_i) and then sets every node to the
+neighbourhood-weighted mean of the samples,
+
+    w_k = sum_i h(c(x_i), k) x_i / sum_i h(c(x_i), k),    h(j, k) = exp(-||r_j - r_k||^2 / (2 sigma(t)^2)),
+
+keeping w_k if the denominator is 0. r_j are the node coordinates on the rectangular grid and sigma decays
+exponentially, sigma(t) = sigma_initial * (sigma_final / sigma_initial) ** (t / n_epochs).
+
+Grouping samples by BMU gives the same update from per-node sums: with S_j the sum and n_j the count of samples whose
+BMU is j, numerator = H @ S and denominator = H @ n, H the (nodes x nodes) neighbourhood matrix. This costs
+O(nodes^2) per epoch instead of O(n_samples * nodes).
+
+References:
+    T. Kohonen, Self-Organizing Maps, 3rd ed., Springer, 2001.
+    J. Vesanto et al., SOM Toolbox for Matlab 5, Helsinki University of Technology, 2000.
+"""
+
+from math import ceil, isfinite
+
+import numpy as np
+import torch
+
+_METRICS = ("euclidean", "cosine")
+_INITS = ("random", "sample")
+
+
+def _is_int(value, minimum: int) -> bool:
+    return isinstance(value, int | np.integer) and not isinstance(value, bool) and value >= minimum
+
+
+def _default_device() -> torch.device:
+    if torch.cuda.is_available():
+        return torch.device("cuda")
+    if torch.backends.mps.is_available():
+        return torch.device("mps")
+    return torch.device("cpu")
+
+
+class SOM:
+    """Batch self-organising map on a rectangular grid with a gaussian neighbourhood.
+
+    The defaults reproduce the maps of spatiomic's ``so.dimension.som``.
+
+    Args:
+        n_rows: Number of grid rows.
+        n_cols: Number of grid columns.
+        n_epochs: Number of epochs; each makes one weight update from all samples.
+        sigma_initial: Initial gaussian neighbourhood width in grid units. Defaults to
+            ceil(min(n_rows, n_cols) / 2) / 2.
+        sigma_final: Width the exponential schedule decays towards (reached at t = n_epochs), > 0. Defaults to
+            0.1 * sigma_initial.
+        init: ``"random"`` draws weights uniformly from [-1, 1]^d with numpy's ``RandomState(random_state)`` and
+            normalises them to unit length; ``"sample"`` uses randomly chosen samples (non-zero ones for cosine).
+        metric: ``"euclidean"`` or ``"cosine"``. Under cosine, zero-norm samples have no BMU: they get label -1, are
+            excluded from the update and the error measures, and have cosine distance 1 to every node.
+        random_state: Seed for the initialisation.
+        device: Torch device. Defaults to cuda, then mps, then cpu.
+        batch_size: Samples processed at once in BMU search, aggregation, predict and transform. Affects memory
+            (about batch_size x n_nodes floats) and speed only. Each epoch's update is the same up to float
+            summation order, but over several epochs such rounding differences (as between devices) can flip
+            near-tied BMUs, so maps agree in quality rather than bit for bit. Defaults to 65536 on GPU and 8192
+            on CPU.
+
+    Attributes:
+        weights_: Node weights, shape (n_rows * n_cols, n_features), row-major (node = row * n_cols + col).
+        labels_: BMU of every training sample under the final weights.
+        n_features_in_: Number of features seen in fit.
+    """
+
+    def __init__(
+        self,
+        n_rows: int = 30,
+        n_cols: int = 30,
+        n_epochs: int = 10,
+        sigma_initial: float | None = None,
+        sigma_final: float | None = None,
+        init: str = "random",
+        metric: str = "euclidean",
+        random_state: int | None = None,
+        device: str | torch.device | None = None,
+        batch_size: int | None = None,
+    ):
+        self.n_rows = n_rows
+        self.n_cols = n_cols
+        self.n_epochs = n_epochs
+        self.sigma_initial = sigma_initial
+        self.sigma_final = sigma_final
+        self.init = init
+        self.metric = metric
+        self.random_state = random_state
+        self.device = device
+        self.batch_size = batch_size
+
+    def _setup(self) -> None:
+        if self.metric not in _METRICS:
+            raise ValueError(f"metric must be one of {_METRICS}, got {self.metric!r}")
+        if self.init not in _INITS:
+            raise ValueError(f"init must be one of {_INITS}, got {self.init!r}")
+        for name in ("n_rows", "n_cols", "n_epochs"):
+            if not _is_int(getattr(self, name), 1):
+                raise ValueError(f"{name} must be an integer >= 1, got {getattr(self, name)!r}")
+        if self.batch_size is not None and not _is_int(self.batch_size, 1):
+            raise ValueError(f"batch_size must be None or an integer >= 1, got {self.batch_size!r}")
+        self._device = torch.device(self.device) if self.device is not None else _default_device()
+        self._batch_size = self.batch_size or (8192 if self._device.type == "cpu" else 1 << 16)
+        self._sigma0 = ceil(min(self.n_rows, self.n_cols) / 2) / 2 if self.sigma_initial is None else self.sigma_initial
+        self._sigma1 = 0.1 * self._sigma0 if self.sigma_final is None else self.sigma_final
+        if not all(isfinite(sigma) and sigma > 0 for sigma in (self._sigma0, self._sigma1)):
+            raise ValueError("sigma_initial and sigma_final must be finite and > 0")
+        self._coords = torch.cartesian_prod(torch.arange(self.n_rows), torch.arange(self.n_cols))
+        self._coords = self._coords.to(self._device, torch.float32)
+        self._grid_sq_dist = torch.cdist(self._coords, self._coords) ** 2
+
+    def _check_X(self, X) -> torch.Tensor:
+        data = torch.as_tensor(
+            X.detach() if isinstance(X, torch.Tensor) else np.ascontiguousarray(X),
+            dtype=torch.float32,
+            device=self._device,
+        )
+        if data.ndim < 2 or data.numel() == 0:
+            raise ValueError(f"X must have shape (n_samples, n_features) or (..., n_features), got {tuple(data.shape)}")
+        data = data.reshape(-1, data.shape[-1])
+        if hasattr(self, "n_features_in_") and data.shape[1] != self.n_features_in_:
+            raise ValueError(f"X has {data.shape[1]} features, but SOM was fitted with {self.n_features_in_}")
+        if not torch.isfinite(data).all():
+            raise ValueError("X contains NaN or infinite values")
+        return data
+
+    def _check_fitted(self) -> None:
+        if not hasattr(self, "weights_"):
+            raise RuntimeError("This SOM instance is not fitted yet, call fit() first")
+
+    def _init_weights(self, data: torch.Tensor) -> torch.Tensor:
+        rng = np.random.RandomState(self.random_state)
+        n_nodes = self.n_rows * self.n_cols
+        if self.init == "random":
+            weights = rng.rand(self.n_rows, self.n_cols, data.shape[1]) * 2 - 1
+            weights /= np.linalg.norm(weights, axis=-1, keepdims=True)
+            return torch.as_tensor(weights.reshape(n_nodes, -1), dtype=torch.float32, device=self._device)
+        candidates = torch.arange(len(data), device=self._device)
+        if self.metric == "cosine":
+            candidates = candidates[data.norm(dim=1) > 0]
+        if len(candidates) == 0:
+            raise ValueError("init='sample' with metric='cosine' needs at least one non-zero sample")
+        picks = rng.choice(len(candidates), n_nodes, replace=len(candidates) < n_nodes)
+        return data[candidates[torch.as_tensor(picks, device=self._device)]].clone()
+
+    def _scores(self, batch: torch.Tensor, weights: torch.Tensor) -> torch.Tensor:
+        """Per-node score to minimise: ||w||^2 - 2 x.w (euclidean, ||x||^2 dropped) or -x.w / ||w|| (cosine)."""
+        if self.metric == "euclidean":
+            return torch.addmm((weights**2).sum(1), batch, weights.T, alpha=-2)
+        return -(batch @ (weights / weights.norm(dim=1, keepdim=True).clamp_min(1e-12)).T)
+
+    def _bmus(self, data: torch.Tensor, weights: torch.Tensor) -> torch.Tensor:
+        """BMU per sample; -1 for zero-norm samples under the cosine metric."""
+        bmus = torch.cat([self._scores(batch, weights).argmin(1) for batch in data.split(self._batch_size)])
+        if self.metric == "cosine":
+            bmus[data.norm(dim=1) == 0] = -1
+        return bmus
+
+    def fit(self, X, y=None) -> "SOM":
+        """Fits the map from a fresh initialisation.
+
+        Args:
+            X: Samples, shape (n_samples, n_features) or (..., n_features); numpy array, array-like or torch tensor.
+                Any real values are fine; data is copied to the device once and stays there.
+            y: Ignored.
+
+        Returns:
+            self
+        """
+        self._setup()
+        for attr in ("weights_", "labels_", "n_features_in_"):
+            self.__dict__.pop(attr, None)
+        data = self._check_X(X)
+        weights = self._init_weights(data)
+        n_nodes = weights.shape[0]
+        for t in range(self.n_epochs):
+            sigma = self._sigma0 * (self._sigma1 / self._sigma0) ** (t / self.n_epochs)
+            sums, counts = torch.zeros_like(weights), weights.new_zeros(n_nodes)
+            for batch in data.split(self._batch_size):
+                bmus = self._scores(batch, weights).argmin(1)
+                if self.metric == "cosine":  # zero-norm samples have no BMU and do not contribute
+                    keep = batch.norm(dim=1) > 0
+                    batch, bmus = batch[keep], bmus[keep]
+                sums.index_add_(0, bmus, batch)
+                counts += torch.bincount(bmus, minlength=n_nodes)
+            h = torch.exp(-self._grid_sq_dist / (2 * sigma**2))
+            numerator, denominator = h @ sums, h @ counts
+            weights = torch.where(denominator[:, None] > 0, numerator / denominator[:, None], weights)
+        self._weights = weights
+        self.weights_ = weights.cpu().numpy()
+        self.n_features_in_ = data.shape[1]
+        self.labels_ = self._bmus(data, weights).cpu().numpy()
+        return self
+
+    def predict(self, X) -> np.ndarray:
+        """Returns the BMU index (int64) per sample; -1 for zero-norm samples under the cosine metric."""
+        self._check_fitted()
+        return self._bmus(self._check_X(X), self._weights).cpu().numpy()
+
+    def fit_predict(self, X, y=None) -> np.ndarray:
+        """Fits the map and returns labels_."""
+        return self.fit(X).labels_
+
+    def _distances(self, batch: torch.Tensor) -> torch.Tensor:
+        if self.metric == "euclidean":  # exact, unlike the BMU search, so large offsets do not cancel
+            return torch.cdist(batch, self._weights, compute_mode="donot_use_mm_for_euclid_dist")
+        cos = -self._scores(batch, self._weights) / batch.norm(dim=1, keepdim=True)
+        return torch.where(torch.isfinite(cos), (1 - cos).clamp(0, 2), 1.0)
+
+    def transform(self, X) -> np.ndarray:
+        """Returns the distance of every sample to every node, shape (n_samples, n_nodes).
+
+        Euclidean or cosine distances (1 - cosine similarity) depending on the metric. The output is a dense
+        n_samples x n_nodes float32 array; for large inputs call it on slices or use predict.
+        """
+        self._check_fitted()
+        data = self._check_X(X)
+        return torch.cat([self._distances(batch) for batch in data.split(self._batch_size)]).cpu().numpy()
+
+    def fit_transform(self, X, y=None) -> np.ndarray:
+        """Fits the map and returns transform(X)."""
+        return self.fit(X).transform(X)
+
+    def _top2(self, X) -> tuple[torch.Tensor, torch.Tensor]:
+        """Distance to the BMU and indices of the two best nodes, for every sample that has a BMU."""
+        self._check_fitted()
+        data = self._check_X(X)
+        if self.metric == "cosine":
+            data = data[data.norm(dim=1) > 0]
+        dists, nodes = [], []
+        for batch in data.split(self._batch_size):
+            best = self._scores(batch, self._weights).topk(min(2, len(self._weights)), dim=1, largest=False)
+            if self.metric == "euclidean":
+                dists.append((batch - self._weights[best.indices[:, 0]]).norm(dim=1))
+            else:
+                dists.append((1 + best.values[:, 0] / batch.norm(dim=1)).clamp(0, 2))
+            nodes.append(best.indices)
+        return torch.cat(dists), torch.cat(nodes)
+
+    def quantization_error(self, X) -> float:
+        """Mean distance (in the chosen metric) between each sample and its BMU."""
+        dist, _ = self._top2(X)
+        return float(dist.cpu().double().mean())  # on the CPU because MPS has no float64
+
+    def topographic_error(self, X) -> float:
+        """Fraction of samples whose first and second BMU are not adjacent on the grid (8-neighbourhood)."""
+        _, nodes = self._top2(X)
+        if nodes.shape[1] < 2:
+            return 0.0
+        offset = (self._coords[nodes[:, 0]] - self._coords[nodes[:, 1]]).abs().amax(1)
+        return float((offset > 1).cpu().double().mean())
+
+
+BatchSOM = SOM