View source on GitHub |
Computes log of the determinant of a hermitian positive definite matrix.
tf.linalg.logdet( matrix, name=None )
# Compute the determinant of a matrix while reducing the chance of over- or underflow: A = ... # shape 10 x 10 det = tf.exp(tf.linalg.logdet(A)) # scalar
Args | |
---|---|
matrix | A Tensor . Must be float16 , float32 , float64 , complex64 , or complex128 with shape [..., M, M] . |
name | A name to give this Op . Defaults to logdet . |
Returns | |
---|---|
The natural log of the determinant of matrix . |
Equivalent to numpy.linalg.slogdet, although no sign is returned since only hermitian positive definite matrices are supported.
© 2020 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 3.0.
Code samples licensed under the Apache 2.0 License.
https://www.tensorflow.org/versions/r2.3/api_docs/python/tf/linalg/logdet