Do not attempt to broadcast when global option arithmetic_broadcast=False (#8784)

* increase plot size

* added old tests

* Keep relevant test

* what's new

* PR comment

* remove unnecessary (?) check

* unnecessary line removal

* removal of variable reassignment to avoid type issue

* Update xarray/core/variable.py

Co-authored-by: Deepak Cherian <[email protected]>

* Update xarray/core/variable.py

Co-authored-by: Deepak Cherian <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Update tests

* what's new

* Update doc/whats-new.rst

---------

Co-authored-by: Deepak Cherian <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: 3 people

doc/whats-new.rst

@@ -23,6 +23,9 @@ v2024.03.0 (unreleased)
 New Features
 ~~~~~~~~~~~~
 
+- Do not broadcast in arithmetic operations when global option ``arithmetic_broadcast=False``
+  (:issue:`6806`, :pull:`8784`).
+  By `Etienne Schalk <https://github.com/etienneschalk>`_ and `Deepak Cherian <https://github.com/dcherian>`_.
 - Add the ``.oindex`` property to Explicitly Indexed Arrays for orthogonal indexing functionality. (:issue:`8238`, :pull:`8750`)
   By `Anderson Banihirwe <https://github.com/andersy005>`_.
 

xarray/core/options.py

@@ -34,6 +34,7 @@
     ]
 
     class T_Options(TypedDict):
+        arithmetic_broadcast: bool
         arithmetic_join: Literal["inner", "outer", "left", "right", "exact"]
         cmap_divergent: str | Colormap
         cmap_sequential: str | Colormap
@@ -59,6 +60,7 @@ class T_Options(TypedDict):
 
 
 OPTIONS: T_Options = {
+    "arithmetic_broadcast": True,
     "arithmetic_join": "inner",
     "cmap_divergent": "RdBu_r",
     "cmap_sequential": "viridis",
@@ -92,6 +94,7 @@ def _positive_integer(value: int) -> bool:
 
 
 _VALIDATORS = {
+    "arithmetic_broadcast": lambda value: isinstance(value, bool),
     "arithmetic_join": _JOIN_OPTIONS.__contains__,
     "display_max_rows": _positive_integer,
     "display_values_threshold": _positive_integer,

xarray/core/variable.py

@@ -2871,6 +2871,16 @@ def broadcast_variables(*variables: Variable) -> tuple[Variable, ...]:
 
 
 def _broadcast_compat_data(self, other):
+    if not OPTIONS["arithmetic_broadcast"]:
+        if (isinstance(other, Variable) and self.dims != other.dims) or (
+            is_duck_array(other) and self.ndim != other.ndim
+        ):
+            raise ValueError(
+                "Broadcasting is necessary but automatic broadcasting is disabled via "
+                "global option `'arithmetic_broadcast'`. "
+                "Use `xr.set_options(arithmetic_broadcast=True)` to enable automatic broadcasting."
+            )
+
     if all(hasattr(other, attr) for attr in ["dims", "data", "shape", "encoding"]):
         # `other` satisfies the necessary Variable API for broadcast_variables
         new_self, new_other = _broadcast_compat_variables(self, other)

xarray/tests/__init__.py

@@ -89,6 +89,13 @@ def _importorskip(
 has_pynio, requires_pynio = _importorskip("Nio")
 has_cftime, requires_cftime = _importorskip("cftime")
 has_dask, requires_dask = _importorskip("dask")
+with warnings.catch_warnings():
+    warnings.filterwarnings(
+        "ignore",
+        message="The current Dask DataFrame implementation is deprecated.",
+        category=DeprecationWarning,
+    )
+    has_dask_expr, requires_dask_expr = _importorskip("dask_expr")
 has_bottleneck, requires_bottleneck = _importorskip("bottleneck")
 has_rasterio, requires_rasterio = _importorskip("rasterio")
 has_zarr, requires_zarr = _importorskip("zarr")

xarray/tests/test_dataarray.py

@@ -51,6 +51,7 @@
     requires_bottleneck,
     requires_cupy,
     requires_dask,
+    requires_dask_expr,
     requires_iris,
     requires_numexpr,
     requires_pint,
@@ -3203,6 +3204,42 @@ def test_align_str_dtype(self) -> None:
         assert_identical(expected_b, actual_b)
         assert expected_b.x.dtype == actual_b.x.dtype
 
+    def test_broadcast_on_vs_off_global_option_different_dims(self) -> None:
+        xda_1 = xr.DataArray([1], dims="x1")
+        xda_2 = xr.DataArray([1], dims="x2")
+
+        with xr.set_options(arithmetic_broadcast=True):
+            expected_xda = xr.DataArray([[1.0]], dims=("x1", "x2"))
+            actual_xda = xda_1 / xda_2
+            assert_identical(actual_xda, expected_xda)
+
+        with xr.set_options(arithmetic_broadcast=False):
+            with pytest.raises(
+                ValueError,
+                match=re.escape(
+                    "Broadcasting is necessary but automatic broadcasting is disabled via "
+                    "global option `'arithmetic_broadcast'`. "
+                    "Use `xr.set_options(arithmetic_broadcast=True)` to enable automatic broadcasting."
+                ),
+            ):
+                xda_1 / xda_2
+
+    @pytest.mark.parametrize("arithmetic_broadcast", [True, False])
+    def test_broadcast_on_vs_off_global_option_same_dims(
+        self, arithmetic_broadcast: bool
+    ) -> None:
+        # Ensure that no error is raised when arithmetic broadcasting is disabled,
+        # when broadcasting is not needed. The two DataArrays have the same
+        # dimensions of the same size.
+        xda_1 = xr.DataArray([1], dims="x")
+        xda_2 = xr.DataArray([1], dims="x")
+        expected_xda = xr.DataArray([2.0], dims=("x",))
+
+        with xr.set_options(arithmetic_broadcast=arithmetic_broadcast):
+            assert_identical(xda_1 + xda_2, expected_xda)
+            assert_identical(xda_1 + np.array([1.0]), expected_xda)
+            assert_identical(np.array([1.0]) + xda_1, expected_xda)
+
     def test_broadcast_arrays(self) -> None:
         x = DataArray([1, 2], coords=[("a", [-1, -2])], name="x")
         y = DataArray([1, 2], coords=[("b", [3, 4])], name="y")
@@ -3381,6 +3418,7 @@ def test_to_dataframe_0length(self) -> None:
         assert len(actual) == 0
         assert_array_equal(actual.index.names, list("ABC"))
 
+    @requires_dask_expr
     @requires_dask
     def test_to_dask_dataframe(self) -> None:
         arr_np = np.arange(3 * 4).reshape(3, 4)