Merge pull request #30 from nschloe/tensordot

tensordot

.pre-commit-config.yaml

@@ -5,7 +5,7 @@ repos:
       - id: isort
 
   - repo: https://github.com/psf/black
-    rev: 21.10b0
+    rev: 22.1.0
     hooks:
       - id: black
         language_version: python3

LICENSE

@@ -1,4 +1,4 @@
-Copyright 2021 Nico Schlömer
+Copyright 2021-2022 Nico Schlömer
 
 Redistribution and use in source and binary forms, with or without modification, are
 permitted provided that the following conditions are met:

justfile

@@ -1,21 +1,12 @@
-version := `python3 -c "from configparser import ConfigParser; p = ConfigParser(); p.read('setup.cfg'); print(p['metadata']['version'])"`
-name := `python3 -c "from configparser import ConfigParser; p = ConfigParser(); p.read('setup.cfg'); print(p['metadata']['name'])"`
-
+version := `python3 -c "from src.npx.__about__ import __version__; print(__version__)"`
 
 default:
 	@echo "\"just publish\"?"
 
-tag:
-	@if [ "$(git rev-parse --abbrev-ref HEAD)" != "main" ]; then exit 1; fi
-	curl -H "Authorization: token `cat ~/.github-access-token`" -d '{"tag_name": "v{{version}}"}' https://api.github.com/repos/nschloe/{{name}}/releases
-
-upload: clean
+publish:
 	@if [ "$(git rev-parse --abbrev-ref HEAD)" != "main" ]; then exit 1; fi
-	# https://stackoverflow.com/a/58756491/353337
-	python3 -m build --sdist --wheel .
-	twine upload dist/*
-
-publish: tag upload
+	gh release create "v{{version}}"
+	flit publish
 
 clean:
 	@find . | grep -E "(__pycache__|\.pyc|\.pyo$)" | xargs rm -rf

pyproject.toml

@@ -1,3 +1,35 @@
 [build-system]
-requires = ["setuptools>=42", "wheel"]
-build-backend = "setuptools.build_meta"
+requires = ["flit_core >=3.2,<4"]
+build-backend = "flit_core.buildapi"
+
+[tool.isort]
+profile = "black"
+
+[project]
+name = "npx"
+authors = [{name = "Nico Schlömer", email = "nico.schloemer@gmail.com"}]
+description = "Some useful extensions for NumPy"
+readme = "README.md"
+license = {file = "LICENSE"}
+classifiers = [
+  "Development Status :: 4 - Beta",
+  "Intended Audience :: Science/Research",
+  "License :: OSI Approved :: BSD License",
+  "Operating System :: OS Independent",
+  "Programming Language :: Python",
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3.7",
+  "Programming Language :: Python :: 3.8",
+  "Programming Language :: Python :: 3.9",
+  "Programming Language :: Python :: 3.10",
+  "Topic :: Scientific/Engineering",
+  "Topic :: Utilities",
+]
+dynamic = ["version"]
+requires-python = ">=3.7"
+dependencies = ["numpy >= 1.20.0"]
+
+[project.urls]
+Code = "https://github.com/nschloe/npx"
+Issues = "https://github.com/nschloe/npx/issues"
+Funding = "https://github.com/sponsors/nschloe"

src/npx/__about__.py

@@ -0,0 +1 @@
+__version__ = "0.1.0"

src/npx/__init__.py

@@ -1,9 +1,11 @@
+from .__about__ import __version__
 from ._isin import isin_rows
 from ._main import add_at, dot, outer, solve, subtract_at, sum_at
 from ._mean import mean
 from ._unique import unique, unique_rows
 
 __all__ = [
+    "__version__",
     "dot",
     "outer",
     "solve",

src/npx/_main.py

@@ -15,9 +15,7 @@ def dot(a: ArrayLike, b: ArrayLike) -> np.ndarray:
     """Take arrays `a` and `b` and form the dot product between the last axis of `a` and
     the first of `b`.
     """
-    a = np.asarray(a)
-    b = np.asarray(b)
-    return np.dot(a, b.reshape(b.shape[0], -1)).reshape(a.shape[:-1] + b.shape[1:])
+    return np.tensordot(a, b, 1)
 
 
 def outer(a: ArrayLike, b: ArrayLike) -> np.ndarray:

src/npx/_mean.py

@@ -11,6 +11,10 @@ def _logsumexp(x: ArrayLike):
 
 
 def mean(x: ArrayLike, p: float = 1) -> np.ndarray:
+    """Generalized mean.
+
+    See <https://github.com/numpy/numpy/issues/19341> for the numpy issue.
+    """
     x = np.asarray(x)
 
     n = len(x)
@@ -24,15 +28,17 @@ def mean(x: ArrayLike, p: float = 1) -> np.ndarray:
         if np.any(x < 0.0):
             raise ValueError("p=0 only works with nonnegative x.")
         return np.prod(np.power(x, 1 / n))
+        # alternative:
+        # return np.exp(np.mean(np.log(x)))
     elif p == np.inf:
         return np.max(np.abs(x))
 
-    if not isinstance(p, int) and np.any(x < 0.0):
-        raise ValueError("Non-integer p only work with nonnegative x.")
-
     if np.all(x > 0.0):
         # logsumexp trick to avoid overflow for large p
         # only works for positive x though
         return np.exp((_logsumexp(p * np.log(x)) - np.log(n)) / p)
-    else:
-        return (np.sum(x ** p) / n) ** (1.0 / p)
+
+    if not isinstance(p, (int, np.integer)):
+        raise ValueError(f"Non-integer p (={p}) only work with nonnegative x.")
+
+    return (np.sum(x**p) / n) ** (1.0 / p)

tests/speedtest.py

@@ -37,6 +37,6 @@ def npx_sum_at(data):
 b = perfplot.bench(
     setup=setup,
     kernels=[np_add_at, npx_add_at, npx_sum_at],
-    n_range=[2 ** k for k in range(23)],
+    n_range=[2**k for k in range(23)],
 )
 b.save("perf-add-at.svg")