import numpy as np


def solve(area):
    next_area = np.array(area)
    if next_area.shape[0] <= 1:
        return 0
    count = 0
    while True:
        max_ind = np.argmax(next_area)
        min_ind = np.argmin(next_area)
        if max_ind == min_ind:
            break
        next_area = next_area[0:min_ind] if max_ind < min_ind else next_area[min_ind + 1: next_area.shape[0]]
        count += 1
    max_ind = np.argmax(area)
    count += solve(area[0:max_ind])
    count += solve(area[max_ind + 1: area.shape[0]])
    return count

N = int(input())
input_np = np.array(input().split()).astype(np.int32)

print(solve(input_np))