# tf.matrix_transpose

### Aliases:

• tf.linalg.transpose
• tf.matrix_transpose
tf.matrix_transpose(
a,
name='matrix_transpose',
conjugate=False
)


Defined in tensorflow/python/ops/array_ops.py.

See the guide: Math > Matrix Math Functions

Transposes last two dimensions of tensor a.

For example:

x = tf.constant([[1, 2, 3], [4, 5, 6]])
tf.matrix_transpose(x)  # [[1, 4],
#  [2, 5],
#  [3, 6]]

x = tf.constant([[1 + 1j, 2 + 2j, 3 + 3j],
[4 + 4j, 5 + 5j, 6 + 6j]])
tf.matrix_transpose(x, conjugate=True)  # [[1 - 1j, 4 - 4j],
#  [2 - 2j, 5 - 5j],
#  [3 - 3j, 6 - 6j]]

# Matrix with two batch dimensions.
# x.shape is [1, 2, 3, 4]
# tf.matrix_transpose(x) is shape [1, 2, 4, 3]


Note that tf.matmul provides kwargs allowing for transpose of arguments. This is done with minimal cost, and is preferable to using this function. E.g.

# Good!  Transpose is taken at minimal additional cost.
tf.matmul(matrix, b, transpose_b=True)

# Inefficient!
tf.matmul(matrix, tf.matrix_transpose(b))


#### Args:

• a: A Tensor with rank >= 2.
• name: A name for the operation (optional).
• conjugate: Optional bool. Setting it to True is mathematically equivalent to tf.conj(tf.matrix_transpose(input)).

#### Returns:

A transposed batch matrix Tensor.

#### Raises:

• ValueError: If a is determined statically to have rank < 2.

#### Numpy Compatibility

In numpy transposes are memory-efficient constant time operations as they simply return a new view of the same data with adjusted strides.

TensorFlow does not support strides, matrix_transposes return a new tensor with the items permuted.