pandas value assignment of 1 element 1 dimension numpy array set a 0 dimensional array

When setting a cell with a numpy 1-dim array, the value that is actually set in the dataframe is a 0-dimensional array. This doesn’t happen when setting multi-element 1-dim arrays:

df = pd.DataFrame({"col1": [np.array([1,2,3]), np.array([4,5,6]), np.array([45])]})
print(df)
for i, arr in df.col1.items():
    new_arr = arr + 1
    print(f"{new_arr.ndim=}")
    df.at[i, 'col1'] = new_arr
    print(f"{df.at[i, 'col1'].ndim=}")

Is this a pandas bug?
I’m using pandas 2.2.2 with python Python 3.10.14

pd: I found this because I’m doing something with numpy 1-d arrays with 1 or more strings, and using explode produced some unexpected results.

>Solution :

This is a limitation of pandas, it is not designed to work with nested structures.

What about reassigning to the original array?

for i, arr in df.col1.items():
    new_arr = arr + 1
    df.at[i, 'col1'][:] = new_arr

print(df)

Output:

        col1
0  [2, 3, 4]
1  [5, 6, 7]
2       [46]