jax-ml/jax · error · ValueError

betainc does not support complex-valued inputs.

Error message

betainc does not support complex-valued inputs.

What it means

jax.scipy.special.betainc (the regularized incomplete beta function) explicitly rejects complex-valued inputs. After promoting arguments to inexact dtypes, it checks whether x is a complex subtype and raises ValueError because the underlying lax.betainc primitive is only defined for real numbers.

Source

Thrown at jax/_src/scipy/special.py:461

     \mathrm{betainc}(a, b, x) = \frac{1}{B(a, b)}\int_0^x t^{a-1}(1-t)^{b-1}\mathrm{d}t

  where :math:`B(a, b)` is the :func:`~jax.scipy.special.beta` function.

  Args:
    a: arraylike, real-valued. Parameter *a* of the beta distribution.
    b: arraylike, real-valued. Parameter *b* of the beta distribution.
    x: arraylike, real-valued. Upper limit of the integration.

  Returns:
    array containing values of the betainc function

  See Also:
    - :func:`jax.scipy.special.beta`
    - :func:`jax.scipy.special.betaln`
  """
  a, b, x = promote_args_inexact("betainc", a, b, x)
  if dtypes.issubdtype(x.dtype, np.complexfloating):
    raise ValueError("betainc does not support complex-valued inputs.")
  return lax.betainc(a, b, x)


def digamma(x: ArrayLike) -> Array:
  r"""The digamma function

  JAX implementation of :obj:`scipy.special.digamma`.

  .. math::

     \mathrm{digamma}(x) = \psi(x) = \frac{\mathrm{d}}{\mathrm{d}x}\log \Gamma(x)

  where :math:`\Gamma(x)` is the :func:`~jax.scipy.special.gamma` function.

  Args:
    x: arraylike, real-valued.

  Returns:

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

Convert x (and a, b) to real before calling: jnp.real(x) or x.real if the imaginary part is known to be zero
Check x.dtype with np.issubdtype(x.dtype, np.complexfloating) and branch to a real-only path
If you truly need complex incomplete beta, implement it yourself or use mpmath outside of JAX

Example fix

// before
jax.scipy.special.betainc(a, b, z)  # z is complex
// after
jax.scipy.special.betainc(a, b, jnp.real(z))

Defensive patterns

Strategy: type-guard

Validate before calling

x = jnp.asarray(x)
if np.issubdtype(x.dtype, np.complexfloating):
    raise TypeError('betainc requires real x; got ' + str(x.dtype))

Type guard

def is_real(x) -> bool:
    return not np.issubdtype(jnp.dtype(x), np.complexfloating)

Prevention

Keep statistical CDF inputs real by design; add dtype assertions at pipeline entry
Log jnp.result_type(a, b, x) before special-function calls to catch promotion surprises

When it happens

Trigger: Calling jax.scipy.special.betainc(a, b, x) where any argument (after promotion) yields a complex dtype, e.g. x = 1+2j or mixing a complex scalar with real arrays so promote_args_inexact upgrades the result to complex128.

Common situations: Porting SciPy code that operates on complex spectra; accidentally passing complex tensors from a signal-processing or quantum pipeline into a statistical CDF; dtype promotion surprises where one complex operand makes the whole call complex.

Related errors

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/9b8542407e95869a. Report an issue: GitHub.