Computes the reverse mode backpropagated gradient of the Cholesky algorithm.
For an explanation see "Differentiation of the Cholesky algorithm" by Iain Murray http://arxiv.org/abs/1602.07527.
Constants
String | OP_NAME | The name of this op, as known by TensorFlow core engine |
Public Methods
Output<T> |
asOutput()
Returns the symbolic handle of the tensor.
|
static <T extends TNumber> CholeskyGrad<T> | |
Output<T> |
output()
Symmetrized version of df/dA .
|
Inherited Methods
Constants
public static final String OP_NAME
The name of this op, as known by TensorFlow core engine
Public Methods
public Output<T> asOutput ()
Returns the symbolic handle of the tensor.
Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.
public static CholeskyGrad<T> create (Scope scope, Operand<T> l, Operand<T> grad)
Factory method to create a class wrapping a new CholeskyGrad operation.
Parameters
scope | current scope |
---|---|
l | Output of batch Cholesky algorithm l = cholesky(A). Shape is `[..., M, M]`. Algorithm depends only on lower triangular part of the innermost matrices of this tensor. |
grad | df/dl where f is some scalar function. Shape is `[..., M, M]`. Algorithm depends only on lower triangular part of the innermost matrices of this tensor. |
Returns
- a new instance of CholeskyGrad