Merge pull request jax-ml#24400 from jakevdp:subtract-ufunc

PiperOrigin-RevId: 688190106

Author: Google-ML-Automation

jax/_src/basearray.pyi

@@ -284,6 +284,8 @@ class _IndexUpdateRef:
           mode: str | None = None, fill_value: StaticScalar | None = None) -> Array: ...
   def add(self, values: Any, indices_are_sorted: bool = False,
           unique_indices: bool = False, mode: str | None = None) -> Array: ...
+  def subtract(self, values: Any, *, indices_are_sorted: bool = False,
+               unique_indices: bool = False, mode: str | None = None) -> Array: ...
   def mul(self, values: Any, indices_are_sorted: bool = False,
           unique_indices: bool = False, mode: str | None = None) -> Array: ...
   def multiply(self, values: Any, indices_are_sorted: bool = False,

jax/_src/numpy/lax_numpy.py

@@ -1787,7 +1787,7 @@ def diff(a: ArrayLike, n: int = 1, axis: int = -1,
   slice1_tuple = tuple(slice1)
   slice2_tuple = tuple(slice2)
 
-  op = ufuncs.not_equal if arr.dtype == np.bool_ else ufuncs.subtract
+  op = operator.not_equal if arr.dtype == np.bool_ else operator.sub
   for _ in range(n):
     arr = op(arr[slice1_tuple], arr[slice2_tuple])
 

jax/_src/numpy/ufuncs.py

@@ -1432,11 +1432,37 @@ def not_equal(x: ArrayLike, y: ArrayLike, /) -> Array:
   """
   return lax.ne(*promote_args("not_equal", x, y))
 
-@implements(np.subtract, module='numpy')
+
 @partial(jit, inline=True)
-def subtract(x: ArrayLike, y: ArrayLike, /) -> Array:
+def _subtract(x: ArrayLike, y: ArrayLike, /) -> Array:
+  """Subtract two arrays element-wise.
+
+  JAX implementation of :obj:`numpy.subtract`. This is a universal function,
+  and supports the additional APIs described at :class:`jax.numpy.ufunc`.
+  This function provides the implementation of the ``-`` operator for
+  JAX arrays.
+
+  Args:
+    x, y: arrays to subtract. Must be broadcastable to a common shape.
+
+  Returns:
+    Array containing the result of the element-wise subtraction.
+
+  Examples:
+    Calling ``subtract`` explicitly:
+
+    >>> x = jnp.arange(4)
+    >>> jnp.subtract(x, 10)
+    Array([-10,  -9,  -8,  -7], dtype=int32)
+
+    Calling ``subtract`` via the ``-`` operator:
+
+    >>> x - 10
+    Array([-10,  -9,  -8,  -7], dtype=int32)
+  """
   return lax.sub(*promote_args("subtract", x, y))
 
+
 @implements(np.arctan2, module='numpy')
 @partial(jit, inline=True)
 def arctan2(x1: ArrayLike, x2: ArrayLike, /) -> Array:
@@ -3604,6 +3630,9 @@ def _add_at(a: Array, indices: Any, b: ArrayLike):
     return a.at[indices].add(b).astype(bool)
   return a.at[indices].add(b)
 
+def _subtract_at(a: Array, indices: Any, b: ArrayLike):
+  return a.at[indices].subtract(b)
+
 def _multiply_at(a: Array, indices: Any, b: ArrayLike):
   if a.dtype == bool:
     a = a.astype('int32')
@@ -3628,3 +3657,4 @@ def _multiply_at(a: Array, indices: Any, b: ArrayLike):
 logical_or = ufunc(_logical_or, name="logical_or", nin=2, nout=1, identity=False, call=_logical_or, reduce=_logical_or_reduce)
 logical_xor = ufunc(_logical_xor, name="logical_xor", nin=2, nout=1, identity=False, call=_logical_xor)
 negative = ufunc(_negative, name="negative", nin=1, nout=1, call=_negative)
+subtract = ufunc(_subtract, name="subtract", nin=2, nout=1, call=_subtract, at=_subtract_at)

jax/numpy/__init__.pyi

@@ -829,7 +829,7 @@ def stack(
 def std(a: ArrayLike, axis: _Axis = ..., dtype: DTypeLike = ...,
         out: None = ..., ddof: int = ..., keepdims: builtins.bool = ..., *,
         where: ArrayLike | None = ..., correction: int | float | None = ...) -> Array: ...
-def subtract(x: ArrayLike, y: ArrayLike, /) -> Array: ...
+subtract: BinaryUfunc
 def sum(
     a: ArrayLike,
     axis: _Axis = ...,

tests/lax_numpy_ufuncs_test.py

@@ -250,6 +250,9 @@ def test_binary_ufunc_reduce(self, name, shape, axis, dtype):
     jnp_fun = getattr(jnp, name)
     np_fun = getattr(np, name)
 
+    if jnp_fun.identity is None and axis is None and len(shape) > 1:
+      self.skipTest("Multiple-axis reduction over non-reorderable ufunc.")
+
     jnp_fun_reduce = partial(jnp_fun.reduce, axis=axis)
     np_fun_reduce = partial(np_fun.reduce, axis=axis)