This is an automated email from the ASF dual-hosted git repository.
zhengruifeng pushed a commit to branch branch-4.x
in repository https://gitbox.apache.org/repos/asf/spark.git
The following commit(s) were added to refs/heads/branch-4.x by this push:
new 39b7d17a733e [SPARK-58465][PS] Use native expressions for NumPy
reciprocal of floating-point columns
39b7d17a733e is described below
commit 39b7d17a733e6377fd6312e2d9a6c62529994965
Author: Ruifeng Zheng <[email protected]>
AuthorDate: Tue Aug 4 11:12:34 2026 +0800
[SPARK-58465][PS] Use native expressions for NumPy reciprocal of
floating-point columns
### What changes were proposed in this pull request?
Use native Spark SQL expressions for NumPy reciprocal on pandas-on-Spark
columns with Spark `float` or `double` types. The expressions preserve nulls,
NaN, infinity, and signed zero. Inputs with other types continue to use the
existing pandas UDF.
### Why are the changes needed?
Native expressions avoid the Python worker boundary for floating-point
reciprocal while retaining the established behavior for input types with
NumPy-specific semantics.
### Does this PR introduce _any_ user-facing change?
No. The behavior and result type remain unchanged.
### How was this patch tested?
- Added reciprocal coverage for negative zero in
`NumPyCompatTests.test_np_math_functions`.
- Passed `NumPyCompatTests.test_np_math_functions` and
`NumPyCompatTests.test_np_spark_compat_series`.
- Passed Ruff format and lint checks for
`python/pyspark/pandas/numpy_compat.py`.
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Codex (GPT-5)
Closes #57668 from zhengruifeng/pandas-native-reciprocal-dev2.
Authored-by: Ruifeng Zheng <[email protected]>
Signed-off-by: Ruifeng Zheng <[email protected]>
(cherry picked from commit 67a9b0dc89af0be7e391860e5bb0aae12248b30a)
Signed-off-by: Ruifeng Zheng <[email protected]>
---
python/pyspark/pandas/numpy_compat.py | 12 ++++++++++--
python/pyspark/pandas/tests/test_numpy_compat.py | 4 ++++
2 files changed, 14 insertions(+), 2 deletions(-)
diff --git a/python/pyspark/pandas/numpy_compat.py
b/python/pyspark/pandas/numpy_compat.py
index 1653121d3124..2a39fd65d933 100644
--- a/python/pyspark/pandas/numpy_compat.py
+++ b/python/pyspark/pandas/numpy_compat.py
@@ -66,8 +66,16 @@ unary_np_spark_mappings = {
"positive": F.positive,
"rad2deg": F.degrees,
"radians": F.radians,
- "reciprocal": pandas_udf( # type: ignore[call-overload]
- lambda s: np.reciprocal(s), DoubleType()
+ "reciprocal": lambda c: F.when(
+ F.typeof(c).isin("float", "double"),
+ F.when(c.isNull(), c.cast("double"))
+ .when(
+ c == 0,
+ F.when(c.cast("string") == "-0.0",
F.lit(float("-inf"))).otherwise(F.lit(float("inf"))),
+ )
+ .otherwise(F.lit(1.0) / c),
+ ).otherwise(
+ pandas_udf(np.reciprocal, DoubleType())(c) # type:
ignore[call-overload]
),
"rint": lambda c: F.rint(c.cast("double")),
"sign": F.signum,
diff --git a/python/pyspark/pandas/tests/test_numpy_compat.py
b/python/pyspark/pandas/tests/test_numpy_compat.py
index 2530d6b04a80..719e6e0114b2 100644
--- a/python/pyspark/pandas/tests/test_numpy_compat.py
+++ b/python/pyspark/pandas/tests/test_numpy_compat.py
@@ -105,6 +105,10 @@ class NumPyCompatTestsMixin:
(np.positive, [-np.inf, -64.0, -2.0, 0.0, 2.0, 64.0, np.inf,
np.nan]),
(np.rad2deg, [-np.inf, -64.0, -np.pi, 0.0, np.pi, 64.0, np.inf,
np.nan]),
(np.rint, [-np.inf, -2.5, -1.5, -0.5, -0.0, 0.0, 0.5, 1.5, 2.5,
np.inf, np.nan]),
+ (
+ np.reciprocal,
+ [-np.inf, -64.0, -2.0, -1.0, -0.0, 0.0, 1.0, 2.0, 64.0,
np.inf, np.nan],
+ ),
(np.sign, [-np.inf, -64.0, -2.0, -0.0, 0.0, 2.0, 64.0, np.inf,
np.nan]),
(np.sinh, [-np.inf, -64.0, -2.0, 0.0, 2.0, 64.0, np.inf, np.nan]),
(np.square, [-np.inf, -64.0, -2.0, 0.0, 2.0, 64.0, np.inf,
np.nan]),
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]