Function searchSorted

Searches for where a value would go in a sorted sequence.

This is not a method for checking containment (like javascript in).

The typical use case for this operation is "binning", "bucketing", or "discretizing". The values are assigned to bucket-indices based on the edges listed in 'sortedSequence'. This operation returns the bucket-index for each value.

The side argument controls which index is returned if a value lands exactly on an edge.

The axis is not settable for this operation. It always operates on the innermost dimension (axis=-1). The operation will accept any number of outer dimensions.

Note: This operation assumes that 'sortedSequence' is sorted along the innermost axis, maybe using 'sort(..., axis=-1)'. If the sequence is not sorted no error is raised and the content of the returned tensor is not well defined.

const edges = tf.tensor1d([-1, 3.3, 9.1, 10.0]);
let values = tf.tensor1d([0.0, 4.1, 12.0]);
const result1 = tf.searchSorted(edges, values, 'left');
result1.print(); // [1, 2, 4]

const seq = tf.tensor1d([0, 3, 9, 10, 10]);
values = tf.tensor1d([0, 4, 10]);
const result2 = tf.searchSorted(seq, values, 'left');
result2.print(); // [0, 2, 3]
const result3 = tf.searchSorted(seq, values, 'right');
result3.print(); // [1, 2, 5]

const sortedSequence = tf.tensor2d([[0., 3., 8., 9., 10.],
                                    [1., 2., 3., 4., 5.]]);
values = tf.tensor2d([[9.8, 2.1, 4.3],
                      [0.1, 6.6, 4.5, ]]);
const result4 = tf.searchSorted(sortedSequence, values, 'left');
result4.print(); // [[4, 1, 2], [0, 5, 4]]
Copy

Doc