tensorflow/core/api_def/base_api/api_def_LowerBound.pbtxt

op {
  graph_op_name: "LowerBound"
  visibility: HIDDEN
  in_arg {
    name: "sorted_inputs"
    description: <<END
2-D Tensor where each row is ordered.
END
  }
  in_arg {
    name: "values"
    description: <<END
2-D Tensor with the same numbers of rows as `sorted_search_values`. Contains
the values that will be searched for in `sorted_search_values`.
END
  }
  out_arg {
    name: "output"
    description: <<END
A `Tensor` with the same shape as `values`.  It contains the first scalar index
into the last dimension where values can be inserted without changing the
ordered property.
END
  }
  summary: "Applies lower_bound(sorted_search_values, values) along each row."
  description: <<END
Each set of rows with the same index in (sorted_inputs, values) is treated
independently.  The resulting row is the equivalent of calling
`np.searchsorted(sorted_inputs, values, side='left')`.

The result is not a global index to the entire 
`Tensor`, but rather just the index in the last dimension.

A 2-D example:
  sorted_sequence = [[0, 3, 9, 9, 10],
                     [1, 2, 3, 4, 5]]
  values = [[2, 4, 9],
            [0, 2, 6]]

  result = LowerBound(sorted_sequence, values)

  result == [[1, 2, 2],
             [0, 1, 5]]
END
}