Candy Distribution

Kids like candies, so much that they start beating each other if the candies are not fairly distributed. So at your next party, you better start thinking before you buy the candies.

If there are $K$ kids,
you of course need $K \cdot
X$ candies for a fair distribution, where $X$ is a positive natural number. But
you learned that always at least one kid loses one candy, so
you better be prepared with **exactly** one spare
candy, resulting in $(K \cdot X)
+ 1$ candies.

Usually, the candies are packed into bags with a fixed number of candies $C$ per bag. You will buy some of these bags so that the above constraints are fulfilled.

The first line gives the number of test cases $t$ ($0 < t < 100$). Each test case is specified by two integers $K$ and $C$ on a single line, where $K$ is the number of kids and $C$ the number of candies in one bag ($1 \le K, C \le 10^9$). As your money is limited, you will never buy more than $10^9$ candy bags.

For each test case, print one line. If there is no such
number of candy bags to fulfill the above constraints, print
“`IMPOSSIBLE`”. Otherwise print the number of
candy bags you want to buy. If there is more than one solution,
any will do.

Sample Input 1 | Sample Output 1 |
---|---|

5 10 5 10 7 1337 23 123454321 42 999999937 142857133 |
IMPOSSIBLE 3 872 14696943 166666655 |