jax-ml/jax · error · ValueError

size must be a positive integer; got {size=}

Error message

size must be a positive integer; got {size=}

What it means

Raised by the Nonzero HiJAX primitive when the static size argument is a negative integer. size declares how many nonzero entries to return (with padding), so it must be non-negative. Symbolic dimensions skip this check, but concrete negative sizes fail immediately.

Source

Thrown at jax/_src/numpy/hijax.py:201

  """HiJAX primitive for nonzero."""

  size: int
  axes: tuple[int, ...]
  out_dtype: np.dtype

  def __init__(
      self,
      a_aval: core.ShapedArray,
      *fill_value_avals: core.ShapedArray,
      size: int,
      axes: tuple[int, ...],
      out_dtype: np.dtype):
    if core.is_symbolic_dim(size):
      pass
    else:
      size = operator.index(size)
      if size < 0:
        raise ValueError(f"size must be a positive integer; got {size=}")
    if not dtypes.issubdtype(out_dtype, np.integer):
      raise ValueError(f"out_dtype must be integer typed; got {out_dtype=}")
    if not all(0 <= ax < a_aval.ndim for ax in axes):
      raise ValueError(f"axes out of range for array with {a_aval.ndim} dimensions:  {axes=}")
    if len(axes) != len(set(axes)):
      raise ValueError(f"duplicate axes are not allowed: {axes=}")
    if fill_value_avals and len(fill_value_avals) != len(axes):
      raise ValueError(f"Expected {len(axes)} fill values, got {len(fill_value_avals)}")
    if any(fv.dtype != out_dtype for fv in fill_value_avals):
      raise ValueError(f"Expected fill values to have dtype {out_dtype}, got {fill_value_avals}")
    batch_shape = tuple(
        s for i, s in enumerate(a_aval.shape) if i not in axes
    )
    for fv_aval in fill_value_avals:
      try:
        broadcasted = lax.broadcast_shapes(fv_aval.shape, batch_shape)
      except ValueError as e:
        raise ValueError(

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Clamp the computed size: size = max(0, computed_size)
2. Pass an explicit non-negative upper bound on the number of nonzeros

Example fix

# before
idx = nonzero(a, size=len(a) - k)  # negative when k > len(a)
# after
idx = nonzero(a, size=max(0, len(a) - k))

size = operator.index(size)
assert size >= 0, size

def valid_size(n) -> bool:
    return isinstance(n, int) and n >= 0

When it happens

Trigger: Calling jax.numpy nonzero(..., size=-1) or passing a computed size that went negative (e.g. size = n_true - margin without clamping).

Common situations: Estimating size from a data-dependent count that can be negative in edge batches; off-by-one arithmetic on an upper bound.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

• out_dtype must be integer typed; got {out_dtype=}
• axes out of range for array with {a_aval.ndim} dimensions:
• duplicate axes are not allowed: {axes=}
• Expected {len(axes)} fill values, got {len(fill_value_avals)
• fill_value tuple must have length equal to number of axes ({

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/9d575c0b3854e665. Report an issue: GitHub.