def main():
    import sys
    input = sys.stdin.read
    data = input().split()
    
    n = int(data[0])
    p = list(map(int, data[1:n+1]))
    
    rounds = [(i + 1, p[i]) for i in range(n)]
    rounds.sort(reverse=True, key=lambda x: x[0])
    
    max_n = n + 2  # To handle up to N+1
    
    # Union-Find structures for next and previous available cards
    next_parent = list(range(max_n + 1))
    prev_parent = list(range(max_n + 1))
    
    def find_next(x):
        if next_parent[x] != x:
            next_parent[x] = find_next(next_parent[x])
        return next_parent[x]
    
    def find_prev(x):
        if prev_parent[x] != x:
            prev_parent[x] = find_prev(prev_parent[x])
        return prev_parent[x]
    
    score = 0
    
    for i, pi in rounds:
        target = pi + 1
        x = find_next(target)
        if x <= n:
            score += i
            # Update next_parent to point to the next available card
            next_parent[x] = find_next(x + 1)
            # Update prev_parent to point to the previous available card
            prev_parent[x] = find_prev(x - 1)
        else:
            # Find the largest available card
            y = find_prev(n)
            if y >= 1:
                if y < pi:
                    score -= i
                elif y > pi:
                    score += i
                # Update the unions for the used card y
                next_parent[y] = find_next(y + 1)
                prev_parent[y] = find_prev(y - 1)
    
    print(score)

if __name__ == "__main__":
    main()