Defined in awkward.operations.reducers on line 761.

ak.argmax(array, axis=None, keepdims=False, mask_identity=True)
  • array – Data to find the index positions of the maximum values.

  • axis (None or int) – If None, combine all values from the array into a single scalar result; if an int, group by that axis: 0 is the outermost, 1 is the first level of nested lists, etc., and negative axis counts from the innermost: -1 is the innermost, -2 is the next level up, etc.

  • keepdims (bool) – If False, this reducer decreases the number of dimensions by 1; if True, the reduced values are wrapped in a new length-1 dimension so that the result of this operation may be broadcasted with the original array.

  • mask_identity (bool) – If True, reducing over empty lists results in None (an option type); otherwise, reducing over empty lists results in the operation’s identity.

Returns the index position of the maximum value in each group of elements from array (many types supported, including all Awkward Arrays and Records). The identity of maximization would be negative infinity, but argmax must return the position of the maximum element, which has no value for empty lists. Therefore, the identity should be masked: the argmax of an empty list is None. If mask_identity=False, the result would be -1, which is distinct from all valid index positions, but care should be taken that it is not misinterpreted as “the last element of the list.”

This operation is the same as NumPy’s argmax if all lists at a given dimension have the same length and no None values, but it generalizes to cases where they do not.

See ak.sum for a more complete description of nested list and missing value (None) handling in reducers.