python中的带有字典的尾递归

Question

I am trying to implement a tail-recursive algorithm for enumerating sets. I am searching for lists of numbers with particular properties, but I found a problem in my recursion.

I made a minimal working version with a list (included below) but changing the list to a dictionary and trying the same algorithm gave strange results.

For the purposes of the example,I know I don't need a dictionary but for the non-minimal algorithm I do. Here is my MWE:

# This is the broken algorithm
def find_solutions_dict(actual, may, solutions):

    if (2 not in actual):
        actual[2] = []

    if (len(may) == 0):
        solutions.append(actual.copy())
        return

    # Move onto the next section, excluding the first option                       
    find_solutions_dict(actual, may[1:], solutions)

    new_overlaps = actual.copy()
    new_overlaps[2].append( may[0] )
    find_solutions_dict(new_overlaps, may[1:],solutions)

# However, this one works
def find_solutions_list(actual, may, solutions):
    if (len(may) == 0):
        solutions.append(actual[:])
        return

    find_solutions_list(actual, may[1:], solutions)

    new_overlaps = actual[:]
    new_overlaps.append( may[0])
    find_solutions_list(new_overlaps, may[1:], solutions)

# Test
sol = []
may = [1,2,3]

find_solutions_list([],may, sol)
# Prints all 8 subsets of [1,2,3]
print sol


sol2 = []
find_solutions_dict({}, may, sol2)
# Has 8 entries, but there are many duplicates
print sol2

Where is the error that makes the dictionary algorithm repeat entries, but the list one work?

Answer 1

dict.copy is a shallow copy, not a deep copy. That means, the list in the dictionary is not copied, but shared between the copies. Use copy.deepcopy from the copy -module instead.