jax-ml/jax · error · ValueError

Got {method=}; expected 'auto', 'fft', or 'direct'.

Error message

Got {method=}; expected 'auto', 'fft', or 'direct'.

What it means

The public convolve/correlate 'method' parameter selects the implementation: 'fft' uses fftconvolve, 'direct'/'auto' use the spatial path. Any other string raises this error listing the accepted values.

Source

Thrown at jax/_src/scipy/signal.py:260

    Specifying ``mode = 'same'`` returns a centered convolution the same size
    as the first input:

    >>> jax.scipy.signal.convolve(x, y, mode='same')
    Array([3., 6., 7., 6., 3.], dtype=float32)

    Specifying ``mode = 'valid'`` returns only the portion where the two arrays
    fully overlap:

    >>> jax.scipy.signal.convolve(x, y, mode='valid')
    Array([6., 7., 6.], dtype=float32)
  """
  if method == 'fft':
    return fftconvolve(in1, in2, mode=mode)
  elif method in ['direct', 'auto']:
    return _convolve_nd(in1, in2, mode, precision=precision)
  else:
    raise ValueError(f"Got {method=}; expected 'auto', 'fft', or 'direct'.")


def convolve2d(in1: Array, in2: Array, mode: ModeString = 'full', boundary: str = 'fill',
               fillvalue: float = 0, precision: PrecisionLike = None) -> Array:
  """Convolution of two 2-dimensional arrays.

  JAX implementation of :func:`scipy.signal.convolve2d`.

  Args:
    in1: left-hand input to the convolution. Must have ``in1.ndim == 2``.
    in2: right-hand input to the convolution. Must have ``in2.ndim == 2``.
    mode: controls the size of the output. Available operations are:

      * ``"full"``: (default) output the full convolution of the inputs.
      * ``"same"``: return a centered portion of the ``"full"`` output which
        is the same size as ``in1``.
      * ``"valid"``: return the portion of the ``"full"`` output which do not
        depend on padding at the array edges.

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Set method to exactly 'auto', 'fft', or 'direct'
2. Strip/lowercase external strings before passing: method = method.strip().lower()
3. Default to omitting method (defaults to 'auto') when unsure

Example fix

// before
convolve(x, y, method=os.environ['CONV_METHOD'])
// after
convolve(x, y, method=os.environ['CONV_METHOD'].strip().lower())

METHODS = ('auto', 'fft', 'direct')
assert method in METHODS, f'method must be one of {METHODS}'

def is_valid_method(m: str) -> bool:
    return m in ('auto', 'fft', 'direct')

When it happens

Trigger: jax.scipy.signal.convolve(x, y, method='fft ') (trailing space), method='FFT', or method='auto' misspelled; passing a scipy-style method default copied incorrectly.

Common situations: Config-driven code where method comes from a YAML/env string; case or whitespace mismatches; version drift from scipy APIs that accept different method names.

Related errors

• unsupported mode: {mode}
• For 'valid' mode, One input must be at least as large as the
• mode must be one of ['full', 'same', 'valid']
• in1 and in2 must have the same number of dimensions
• zero-size arrays not supported in convolutions, got shapes {

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/f8372af0bff62b08. Report an issue: GitHub.