Integer partitioning problem - for fun

[Pepper Mint]

I have a user input of 2 integers (m,n)
Then my system will generate 1 list of M (m,n < M) integers that start at m and end at M^th integer of value x_M. The formula to calculate x_M is followed by
x_0=m
x_M=x_{M-1}+n
After the list is generated I randomly delete N (N << M) rows from it and given that my system isn't allowed to remember (m,n), how can I find out what values (m,n) were ?

For example
(m,n)=(2,5)
M=5 => L={2,7,12,17,22}
deleting L_2,4 yields L={2,12,22}

 

[Borek]

In general case you can't.

I would start checking what all x_i-x_i-1 (actually every x_i-x_j) have in common.

 

[micromass]

As Borek says, you can't do this if you don't know $N$ and $M$. But you can statistically analyze what the $n$ and $m$ values likely might have been.