jax-ml/jax · error · ValueError

series_order must be <= 30.

Error message

series_order must be <= 30.

What it means

log_ndtr's series_order is capped at 30 because the precomputed coefficient tables and the asymptotic expansion only cover 30 terms; more terms bring no accuracy benefit and would need larger tables. Values above 30 raise ValueError.

Source

Thrown at jax/_src/scipy/special.py:1670

  Args:
    x: an array of type `float32`, `float64`.
    series_order: Positive Python integer. Maximum depth to
      evaluate the asymptotic expansion. This is the `N` above.

  Returns:
    an array with `dtype=x.dtype`.

  Raises:
    TypeError: if `x.dtype` is not handled.
    TypeError: if `series_order` is a not Python `integer.`
    ValueError:  if `series_order` is not in `[0, 30]`.
  """
  if not isinstance(series_order, int):
    raise TypeError("series_order must be a Python integer.")
  if series_order < 0:
    raise ValueError("series_order must be non-negative.")
  if series_order > 30:
    raise ValueError("series_order must be <= 30.")

  x_arr = jnp.asarray(x)
  dtype = lax.dtype(x_arr)

  if dtype == np.float64:
    lower_segment: np.ndarray = _LOGNDTR_FLOAT64_LOWER
    upper_segment: np.ndarray = _LOGNDTR_FLOAT64_UPPER
  elif dtype == np.float32:
    lower_segment = _LOGNDTR_FLOAT32_LOWER
    upper_segment = _LOGNDTR_FLOAT32_UPPER
  else:
    raise TypeError(f"x.dtype={np.dtype(dtype)} is not supported.")

  # The basic idea here was ported from:
  #   https://root.cern.ch/doc/v608/SpecFuncCephesInv_8cxx_source.html
  # We copy the main idea, with a few changes
  # * For x >> 1, and X ~ Normal(0, 1),
  #     Log[P[X < x]] = Log[1 - P[X < -x]] approx -P[X < -x],

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Clamp to 30: series_order = min(30, int(order)) — the default of 3 is usually already sufficient for float32/64 accuracy
2. Read the docstring's documented [0, 30] range and pick a modest value (3–18)
3. If more tail accuracy is truly needed, use log_ndtr(x.astype(jnp.float64), order) instead of raising order

Example fix

// before
jax.scipy.special.log_ndtr(x, series_order=50)
// after
jax.scipy.special.log_ndtr(x, series_order=min(30, 50))

series_order = min(30, max(0, int(series_order)))

def order_in_range(o):
    return isinstance(o, int) and 0 <= o <= 30

When it happens

Trigger: Calling log_ndtr(x, series_order=31) or higher, often from a sweep or an accuracy-tuning loop that assumes more is better.

Common situations: Accuracy tuning where users crank the order; copying order values from other libraries' settings; automated hyperparameter search exceeding the bound.

Related errors

• series_order must be non-negative.
• series_order must be a Python integer.
• x.dtype={np.dtype(dtype)} is not supported.
• z must be a 1D array.
• `compute_on`'s compute_type argument must be a string.

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/f61d6f392ab18eef. Report an issue: GitHub.