ScatterNdNonAliasingAdd

tensorflow C++ API

tensorflow::ops::ScatterNdNonAliasingAdd

Applies sparse addition to input using individual values or slices.

Summary

fromupdatesaccording to indicesindices. The updates are non-aliasing:inputis only modified in-place if no other operations will use it. Otherwise, a copy ofinputis made. This operation has a gradient with respect to bothinputandupdates.

inputis aTensorwith rankPandindicesis aTensorof rankQ.

indicesmust be integer tensor, containing indices intoinput. It must be shape[d_0, ..., d_{Q-2}, K]where0 < K <= P.

The innermost dimension ofindices(with lengthK) corresponds to indices into elements (ifK = P) or(P-K)-dimensional slices (ifK < P) along theKth dimension ofinput.

updatesisTensorof rankQ-1+P-Kwith shape:

``` [d_0, ..., d_{Q-2}, input.shape[K], ..., input.shape[P-1]]. ```

For example, say we want to add 4 scattered elements to a rank-1 tensor to 8 elements. In Python, that addition would look like this:

input = tf.constant([1,2,3,4,5,6,7,8])
indices = tf.constant([[4],[3],[1],[7]])
updates = tf.constant([9,10,11,12])
output = tf.scatter_nd_non_aliasing_add(input, indices, updates)
with tf.Session()as sess:
    print(sess.run(output))

The resulting valueoutputwould look like this:

[1, 13, 3, 14, 14, 6, 7, 20]

See tf.scatter_nd for more details about how to make updates to slices.

Arguments:

scope: A Scope object
input: A Tensor.
indices: A Tensor . Must be one of the following types:int32,int64. A tensor of indices into input.
updates: A Tensor. Must have the same type as ref. A tensor of updated values to add to input.

Returns:

Output: A Tensorwith the same shape as input, containing values of inputupdated with updates.