n should be a positive integer or None

Package:

tensorflow

158813

Exception Class:

ValueError

Raise code

def _validate_dct_arguments(input_tensor, dct_type, n, axis, norm):
  """Checks that DCT/IDCT arguments are compatible and well formed."""
  if axis != -1:
    raise NotImplementedError("axis must be -1. Got: %s" % axis)
  if n is not None and n < 1:
    raise ValueError("n should be a positive integer or None")
  if dct_type not in (1, 2, 3, 4):
    raise ValueError("Types I, II, III and IV (I)DCT are supported.")
  if dct_type == 1:
    if norm == "ortho":
      raise ValueError("Normalization is not supported for the Type-I DCT.")
    if input_tensor.shape[-1] is not None and input_tensor.shape[-1] < 2:
      raise ValueError(

Links to the raise (1)

https://github.com/tensorflow/tensorflow/blob/7acd515ec218b414d5b16e6710268ac03d9f5421/tensorflow/python/ops/signal/dct_ops.py#L37

Ways to fix

tf.signal.dct computes the 1D [Discrete Cosine Transform (DCT)][dct] of `input.`

Usage:

result = tf.signal.dct(input,
                       type=2, 
                       n=None, 
                       axis=-1, 
                       norm=None, 
                       name=None)

The parameter n here is the length of the transform and it should be a positive integer. If a negative value is given it causes an error.

Reproducing the error:

pipenv install tensorflow

from tensorflow.python.ops.signal import dct_ops
signals = np.random.rand(4,3)
n = -3 # this value cause the error 
tf_dct = dct_ops.dct(signals, n=n,norm="ortho")
print(tf_dct)

The error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-27-13c3895d48f9> in <module>
      3 signals = np.random.rand(4,3)
      4 n = -3
----> 5 tf_dct = dct_ops.dct(signals, n=n,norm="ortho")

~/my_env_project/lib/python3.8/site-packages/tensorflow/python/util/dispatch.py in wrapper(*args, **kwargs)
    204     """Call target, and fall back on dispatchers if there is a TypeError."""
    205     try:
--> 206       return target(*args, **kwargs)
    207     except (TypeError, ValueError):
    208       # Note: convert_to_eager_tensor currently raises a ValueError, not a

~/my_env_project/lib/python3.8/site-packages/tensorflow/python/ops/signal/dct_ops.py in dct(input, type, n, axis, norm, name)
     97   [dct]: https://en.wikipedia.org/wiki/Discrete_cosine_transform
     98   """
---> 99   _validate_dct_arguments(input, type, n, axis, norm)
    100   with _ops.name_scope(name, "dct", [input]):
    101     input = _ops.convert_to_tensor(input)

~/my_env_project/lib/python3.8/site-packages/tensorflow/python/ops/signal/dct_ops.py in _validate_dct_arguments(input_tensor, dct_type, n, axis, norm)
     35     raise NotImplementedError("axis must be -1. Got: %s" % axis)
     36   if n is not None and n < 1:
---> 37     raise ValueError("n should be a positive integer or None")
     38   if dct_type not in (1, 2, 3, 4):
     39     raise ValueError("Types I, II, III and IV (I)DCT are supported.")

ValueError: n should be a positive integer or None

Fixed version of the code:

from tensorflow.python.ops.signal import dct_ops
signals = np.random.rand(4,3)
n = 3
tf_dct = dct_ops.dct(signals, n=n,norm="ortho")
print(tf_dct)

Output:

tf.Tensor(
[[ 0.70799171 -0.32111126  0.04285833]
 [ 1.21643631 -0.44179349 -0.26358989]
 [ 0.77752386  0.3508207   0.2477764 ]
 [ 0.75556746 -0.17665882  0.10850445]], shape=(4, 3), dtype=float64)