tf_quant_finance.math.diff

Last updated: 2023-03-16.

tf_quant_finance.math.diff#

View source

Computes the difference between elements of an array at a regular interval.

tf_quant_finance.math.diff(
    x, order=1, exclusive=False, axis=-1, dtype=None, name=None
)

For a difference along the final axis, if exclusive is True, then computes:

  result[..., i] = x[..., i+order] - x[..., i] for i < size(x) - order

This is the same as doing x[..., order:] - x[..., :-order]. Note that in this case the result Tensor is smaller in size than the input Tensor.

If exclusive is False, then computes:

  result[..., i] = x[..., i] - x[..., i-order] for i >= order
  result[..., i] = x[..., i]  for 0 <= i < order

Example#

  x = tf.constant([1, 2, 3, 4, 5])
  dx = diff(x, order=1, exclusive=False)  # Returns [1, 1, 1, 1, 1]
  dx1 = diff(x, order=1, exclusive=True)  # Returns [1, 1, 1, 1]
  dx2 = diff(x, order=2, exclusive=False)  # Returns [1, 2, 2, 2, 2]

Args:#

  • x: A Tensor of shape batch_shape + [n] and of any dtype for which arithmetic operations are permitted.

  • order: Positive Python int. The order of the difference to compute. order = 1 corresponds to the difference between successive elements. Default value: 1

  • exclusive: Python bool. See description above. Default value: False

  • axis: Python int. The axis of x along which to difference. Default value: -1 (the final axis).

  • dtype: Optional tf.DType. If supplied, the dtype for x to use when converting to Tensor. Default value: None which maps to the default dtype inferred by TF.

  • name: Python str name prefixed to Ops created by this class. Default value: None which is mapped to the default name ‘diff’.

Returns:#

  • diffs: A Tensor of the same dtype as x. If exclusive is True, then the shape is batch_shape + [n-min(order, n)], otherwise it is batch_shape + [n]. The final dimension of which contains the differences of the requested order.

Contents