That's cool.. but I have the same question that other, why is the reason to sort?

The best possible time complexity for this task is linear.
This implementation has linear-logarithmic time complexity and it also corrupts the initial data.
Conciseness can never justify poor performance. Sorry, but it's definitely not the best practice.

1. You are wrong - to sort primitives Java uses the so-called Dual-Pivot Quicksort. This algorithm is based on quicksort, as obvious from its name, but in fact, comprises many technics to ensure linear-logarithmic performance even with the massive data.

2. Your solution is computationally intensive as well, it performs iteration through the entire array B for every element in array A. That gives O(n * m) time complexity.
   And this task can be solved in a linear time (try to refine your algorithm, or you may look at my solution).
   Additionally, your code has a side effect - it mutates array B.

sort is pretty kind of simplest way but costs too much because it takes O(nlog(n)) for fastest merge sort,but at least,it looks pretty clear and readable,well done!Arr

Sorting isn't cheap. Isn't there a way to evaluate the efficency of a solution?