Take the 5×4 coordinate grid and add a 'mine' column of 0s (safe) and 1s (mine). Seed with np.random.seed(0) immediately before drawing (inside solve, right before np.random.binomial: it's global, shared state, so seeding anywhere else won't reproduce the same numbers), giving each square a 40% chance of being a mine. Return the DataFrame.

Output

    x  y  mine
0   0  0     0
1   0  1     1
2   0  2     1
3   0  3     0
4   1  0     0
.. .. ..   ...
15  3  3     0
16  4  0     0
17  4  1     1
18  4  2     1
19  4  3     1

[20 rows x 3 columns]