And the Winner Is... Ourselves!

Let us remind you how the total penalty for this contest is calculated.

• When you solve a problem at $T$ minutes, $T+20V$ is added to your penalty, where $V$ is the number of incorrect verdicts (excluding compile errors) received on that problem.
• If you do not solve a problem before the contest ends, the incorrect verdicts on that problem are not counted toward the penalty.

Here is bad news for all of you: we, the problem setters, are planning to enter the competition and solve our own problems.

We know our problems very well, so we can solve all of them before the contest ends. Moreover, we can predict exactly how long each problem takes to solve and how many incorrect verdicts (excluding compile errors) we will get on each problem. Depending on the order in which we solve the problems, our total penalty can differ. What is the minimum penalty if we solve all the problems?

The input consists of 11 lines. The $i$-th line contains two space-separated integers $D\_i$ and $V\_i$, where $D\_i$ is the number of minutes required to solve the $i$-th problem and $V\_i$ is the number of incorrect verdicts on the $i$-th problem.

For each $i$, $1 \leq D\_i$ and $0 \leq V\_i \leq 1\ 000$. Also, $\sum\_{i=1}^{11} D\_i \leq 300$.

Output the minimum penalty if we solve all the problems.

The sample input does not necessarily reflect the actual difficulty of the problems.

The problem statement does not necessarily reflect the actual situation of the contest.