Please type the answer with full process clearly. Thank you very much!

(10 pts.) If we add one more partition step in find-pivot function in selecting problem, and use 3 as the size of each subarray, the algorithm would become: function find-pivot (A, k) Partition A into n/3 subarrays; Let M be the list of medians of these n/3 subarrays; Partition M into n/9 subarrays; Let M' be the list of medians of these n/9 subarrays; return selection(M',|M']/2) end function; Analyze the running time of the new selection algorithm with the find-pivot function described above