Farmer John is decorating his Spring Equinox Tree (like a Christmas tree but popular about three months later). It can be modeled as a rooted mathematical tree with N (1 <= N <= 100,000) elements, labeled 1...N, with element 1 as the root of the tree. Each tree element e > 1 has a parent,

(1 <=

<= N). Element 1 has no parent (denoted '-1' in the input), of course, because it is the root of the tree.
Each element i has a corresponding subtree (potentially of size 1) rooted there. FJ would like to make sure that the subtree corresponding to element i has a total of at least

(0 <=

<= 10,000,000) ornaments scattered among its members. He would also like to minimize the total amount of time it takes him to place all the ornaments (it takes time K*

to place K ornaments at element i (1 <=

<= 100)).
Help FJ determine the minimum amount of time it takes to place ornaments that satisfy the constraints. Note that this answer might not fit into a 32-bit integer, but it will fit into a signed 64-bit integer.
For example, consider the tree below where nodes located higher on
the display are parents of connected lower nodes (1 is the root):
1
|
2
|
5
/ \
4 3
Suppose that FJ has the following subtree constraints:
Minimum ornaments the subtree requires
| Time to install an ornament
Subtree | |
root | C_i | T_i
--------+--------+-------
1 | 9 | 3
2 | 2 | 2
3 | 3 | 2
4 | 1 | 4
5 | 3 | 3
Then FJ can place all the ornaments as shown below, for a total
cost of 20:
1 [0/9(0)] legend: element# [ornaments here/ | total ornaments in subtree(node install time)]
2 [3/9(6)]
|
5 [0/6(0)]
/ \
[1/1(4)] 4 3 [5/5(10)]