numpy · Tontonio3 · Mar 27, 2025 · Mar 27, 2025 · Mar 27, 2025 · Mar 27, 2025
diff --git a/doc/release/upcoming_changes/28589.improvement.rst b/doc/release/upcoming_changes/28589.improvement.rst
@@ -0,0 +1,7 @@
+Improved error handling in `np.quantile`
+----------------------------------------
+`np.quantile` now raises errors if:
+
+* All weights are zero
+* At least one weight is `np.nan`
+* At least one weight is `np.inf`
diff --git a/numpy/lib/_function_base_impl.py b/numpy/lib/_function_base_impl.py
@@ -4535,9 +4535,11 @@ def quantile(a,
         if axis is not None:
             axis = _nx.normalize_axis_tuple(axis, a.ndim, argname="axis")
         weights = _weights_are_valid(weights=weights, a=a, axis=axis)
-        if np.any(weights < 0):
-            raise ValueError("Weights must be non-negative.")
-
+        if np.any(weights <= 0):
+            if np.any(weights < 0):
+                raise ValueError("Weights must be non-negative.")
+            elif np.all(weights == 0):
+                raise ValueError("At least one weight must be non-zero.")
     return _quantile_unchecked(
         a, q, axis, out, overwrite_input, method, keepdims, weights)
 
@@ -4892,6 +4894,10 @@ def _quantile(
         # We use the weights to calculate the empirical cumulative
         # distribution function cdf
         cdf = weights.cumsum(axis=0, dtype=np.float64)
+        if np.any(np.isinf(cdf[-1])):
+            raise ValueError("Weights must be finite.")
+        elif np.any(np.isnan(cdf[-1])):
+            raise ValueError("At least one weight is nan.")
         cdf /= cdf[-1, ...]  # normalization to 1
         # Search index i such that
         #   sum(weights[j], j=0..i-1) < quantile <= sum(weights[j], j=0..i)

diff --git a/numpy/lib/tests/test_function_base.py b/numpy/lib/tests/test_function_base.py
@@ -4142,6 +4142,31 @@ def test_closest_observation(self):
         assert_equal(4, np.quantile(arr[0:9], q, method=m))
         assert_equal(5, np.quantile(arr, q, method=m))
 
+    @pytest.mark.parametrize(["err_msg", "weight"],
+                             [("Weights must be finite.", [1, np.inf, 1, 1]),
+                              ("Weights must be non-negative.", [1, -np.inf, 1, 1]),
+                              ("Weights must be finite.", [1, np.inf, 1, np.inf]),
+                              ("At least one weight must be non-zero.", np.zeros(4))])
+    @pytest.mark.parametrize("dty", ["f8", "O"])
+    def test_inf_zeroes_err(self, err_msg, weight, dty):
+
+        m = "inverted_cdf"
+        q = 0.5
+        arr = [1, 2, 3, 4]
+        wgts = np.array(weight, dtype=dty)
+        with pytest.raises(ValueError, match=err_msg):
+            a = np.quantile(arr, q, weights=wgts, method=m)
+
+    @pytest.mark.parametrize("weight", [[1, np.nan, 1, 1], [1, np.nan, np.nan, 1]])
+    @pytest.mark.parametrize(["err", "dty"], [(ValueError, "f8"), ((RuntimeWarning, ValueError), "O")])
+    def test_nan_err(self, err, dty, weight):
+
+        m = "inverted_cdf"
+        q = 0.5
+        arr = [1, 2, 3, 4]
+        wgts = np.array(weight, dtype=dty)
+        with pytest.raises(err):
+            a = np.quantile(arr, q, weights=wgts, method=m)
 
 class TestLerp:
     @hypothesis.given(t0=st.floats(allow_nan=False, allow_infinity=False,