Promote Python-scalar primals to jnp arrays at jet() entry

Several primitive Taylor rules (notably _log1p_taylor) accessed .ndim on
the user-supplied primal, which fails when the caller passes a Python
float. Convert primals (and series) to jnp arrays once in jet() so every
rule can rely on array attributes.

Adds parametrized regression test test_python_float_primal across all
supported test functions.

Also: update author to Cambridge Yang only.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: thisiscam

README.md

@@ -280,7 +280,7 @@ If you use `jet-array` in academic work, please cite the repository:
 ```bibtex
 @software{jet_array,
   title  = {jet-array: array-form Taylor-mode automatic differentiation in JAX},
-  author = {Yang, Cambridge and Li, Dongdong},
+  author = {Yang, Cambridge},
   year   = {2026},
   url    = {https://github.com/thisiscam/jet-array},
 }

jet_array/__init__.py

@@ -206,6 +206,11 @@ def jet(fun, primals, series, effective_order=None, **_):
             if not treedef_is_leaf(treedef):
                 raise ValueError(f"term {j} for argument {i} is not an array")
 
+    # Promote Python scalars to jnp arrays so that primitive rules can rely
+    # on .ndim/.shape/.dtype.
+    primals = tuple(jnp.asarray(p) for p in primals)
+    series = tuple(jnp.asarray(s) for s in series)
+
     @lu.transformation_with_aux2
     def flatten_fun_output(f, store, *args):
         ans = f(*args)

pyproject.toml

@@ -8,7 +8,6 @@ license = "Apache-2.0"
 license-files = ["LICENSE"]
 authors = [
     { name = "Cambridge Yang", email = "camyang@csail.mit.edu" },
-    { name = "Dongdong Li", email = "dli.stats@gmail.com" },
 ]
 keywords = ["jax", "automatic-differentiation", "taylor-mode", "jet", "higher-order-derivatives"]
 classifiers = [

tests/test_against_jax_jet.py

@@ -151,3 +151,26 @@ def test_identity():
     np.testing.assert_allclose(p, 0.7)
     np.testing.assert_allclose(s[0], 1.0)
     np.testing.assert_allclose(s[1:], jnp.zeros(4), atol=1e-7)
+
+
+@pytest.mark.parametrize("name,fn", TEST_FUNCTION_PARAMS)
+def test_python_float_primal(name, fn):
+    """Python float primals are promoted to jnp arrays automatically.
+
+    Regression test: rules like _log1p_taylor previously called .ndim on the
+    primal, which failed for Python floats.
+    """
+    if name == "log1p":
+        x0 = 0.3
+    else:
+        x0 = 0.5
+    order = 5
+    series_in = [1.0] + [0.0] * (order - 1)
+    primal, series = jet(fn, (x0,), (series_in,))
+    # Compare against the jnp-array path.
+    series_in_arr = jnp.zeros(order, dtype=jnp.float64).at[0].set(1.0)
+    primal_ref, series_ref = jet(
+        fn, (jnp.asarray(x0, dtype=jnp.float64),), (series_in_arr,)
+    )
+    np.testing.assert_allclose(primal, primal_ref, rtol=RTOL, atol=ATOL)
+    np.testing.assert_allclose(series, series_ref, rtol=RTOL, atol=ATOL)