Set from a List of Sets

Question

I am working in Python. I have a large list L = [X_{1}, X_{2}, ....,X_{n}] where each X_{i} is a set. It is possible that for some distinct indices i, j we have X_{i} = X_{j} . I want to create a list (or a set) L_{1} = [Y_{1}, Y_{2}, ... ,Y_{k}] such that the set (in the mathematical sense) {X_{1},..., X_{n}} = {Y_{1}, ..., Y_{k}} and L_{1} is of least possible length (in other-words L_{1} has no repetitions).

For subsequent applications I would need to access the elements of L_{1} . So I cannot use the frozenset option.

Using set(L) gives an error TypeError: unhashable type: 'set' .

I think I can solve it using a loop construction but I am wondering if there is a more elegant solution to this problem in Python.

Answer 1

If you want to eliminate duplicate sets, you'll need to convert them to frozenset() instances:

unique_sets = {frozenset(s) for s in L}

or if you are using Python 3:

unique_sets = set(map(frozenset, L))

If you need to end up with a list of mutable sets again you can produce a list of set() objects again with:

unique_set_list = [set(s) for s in unique_sets]

You could combine the two into a single list comprehension with a little loss of readability:

seen = set()
unique_set_list = [ms for ms, ims in ((s, frozenset(s)) for s in L)
                   if not (ims in seen or seen.add(ims))]