I have this array:
[1,2,3,4]
and I want all the sub arrays you can get except by the single element, in this case:
[1,2],[1,2,3],[1,2,3,4],[2,3],[2,3,4],[3,4].
So I have implemented this code:
def nSub(nums:list):
"""
type nums: list
"""
l_nums = len(nums)
p = 2
for i in range(l_nums):
for j in range(p,l_nums+1):
print(nums[i:j])
p += 1
nSub([1,2,3,4])
But although it works, perhaps there is other way to implement this, so I have this question: is there a pythonic way to do this? Any help will be greatly appreciated.
Simple list comprehension should do the job:
arr = [1,2,3,4]
[arr[n1:n2]
for n1 in range(0, len(arr) + 1)
for n2 in range(n1 + 2, len(arr) + 1)]
Output:
[[1, 2], [1, 2, 3], [1, 2, 3, 4], [2, 3], [2, 3, 4], [3, 4]]
I'm trying to get a list of lists (or tuples) which follows a pattern something like this:
[1,1,1,2]
[1,1,2,2]
[1,2,2,2]
[1,2,2,3]
[1,2,3,3]
[1,2,3,4]
Using itertools.combinations_with_replacement I've gotten close, but I end up with lists which jump values for example:
[1,1,1,3]
or
[2,2,2,3]
I don't want this. I always want to start at 1, and increase until the list is filled, and then increase to the next value.
If I'm using itertools, then is there a way to remove the lists that I don't want?
Instead of using combinations, I would generate the pattern directly.
Create a list of 1's with the desired length and iterate backward, changing the list accordingly.
def generate_increment(n):
lst = [1] * n
result = []
for k in range(n-1):
lst[-1] += 1
result.append(lst[:])
for i in range(len(lst)-2, k, -1):
a, b = lst[i], lst[i+1]
if a != b:
lst[i] = b
result.append(lst[:])
return result
>>print(*generate_increment(4), sep='\n')
[1, 1, 1, 2]
[1, 1, 2, 2]
[1, 2, 2, 2]
[1, 2, 2, 3]
[1, 2, 3, 3]
[1, 2, 3, 4]
It feels like validating the results of combinations will be a bigger effort than simply creating those lists you need.
This can be done with a recursive function that adds with each step what value to add to the list until a defined size:
def gen_list(pre, size):
if size == 1:
return [pre]
res = gen_list(pre + [pre[-1]], size - 1)
res.extend(gen_list(pre + [pre[-1]+1], size-1))
return res
for l in gen_list([1], 4):
print(l)
Which prints:
[1, 1, 1, 1]
[1, 1, 1, 2]
[1, 1, 2, 2]
[1, 1, 2, 3]
[1, 2, 2, 2]
[1, 2, 2, 3]
[1, 2, 3, 3]
[1, 2, 3, 4]
I'm trying to figure out how the following backtracking solution works to generate all the permutations of integers given as a list:
def permutations(arr):
res = []
backtrack(arr, [], set(), res)
print(res)
def backtrack(arr, temp, visited, res):
if len(temp) == len(arr):
res.append(temp[:])
else:
for num in arr:
if num in visited: continue
visited.add(num)
temp.append(num)
backtrack(arr, temp, visited, res)
visited.remove(num)
temp.pop()
Executing with the following:
permutations([1, 2, 3])
The result is as expected:
[[1, 2, 3], [1, 3, 2], [2, 1, 3], [2, 3, 1], [3, 1, 2], [3, 2, 1]]
What i don't understand is the temp.pop() call at the end of the loop.
I know that pop() discards the last element of the list, but why is it necessary here?
I'd really appreciate if someone could explain this to me.
It is necessary because on the next iteration the function will append again an element, so if you didn't pop beforehand the list would keep growing in size. Since backtrack adds results only if the length of temp equals the length of the original arr it won't add anything beyond the first permutation [1, 2, 3] (because from there on, the temp list keeps growing).
We can just remove that line (temp.pop()) and see what the results look like:
[[1, 2, 3]]
Now if we also change that results are added whenever len(temp) >= len(arr) then we'll see how temp grows in size:
[[1, 2, 3], [1, 2, 3, 3], [1, 2, 3, 3, 2], [1, 2, 3, 3, 2, 3]]
We get fewer results here because for each recursive call beyond [1, 2, 3] the temp list is immediately copied without ever reaching the for loop.
I'm trying to solve a problem using backtracking and I need the permutations of numbers for it. I have this basic algorithm that does it but the problem is... the results don't come in the normal order.
def perm(a,k=0):
if(k==len(a)):
print(a)
else:
for i in range(k,len(a)):
a[k],a[i] = a[i],a[k]
perm(a, k+1)
a[k],a[i] = a[i],a[k]
Example: for [1,2,3] the normal results would be: [1,2,3] [1,3,2] [2,1,3] [2,3,1] [3,1,2] [3,2,1]
Whereas this algorithm will interchange the last 2 elements. I understand why. I just don't know how to correct this.
I don't want to use the permutations from itertools. Can the code above be easily fixed to work properly? What would be the complexity for this algorithm from above?
A recursive generator function that yields permutations in the expected order with regard to the original list:
def perm(a):
if len(a) <= 1:
yield a
else:
for i in xrange(len(a)):
for p in perm(a[:i]+a[i+1:]):
yield [a[i]]+p
a = [1, 2, 3]
for p in perm(a):
print(p)
[1, 2, 3]
[1, 3, 2]
[2, 1, 3]
[2, 3, 1]
[3, 1, 2]
[3, 2, 1]
Here's one (suboptimal, because copying lists all the time) solution:
def perm(a, prev=[]):
if not a:
print(prev)
for index, element in enumerate(a):
perm(a[:index] + a[index+1:], prev + [element])
The order it is printed out:
>>> perm([1,2,3])
[1, 2, 3]
[1, 3, 2]
[2, 1, 3]
[2, 3, 1]
[3, 1, 2]
[3, 2, 1]
I'm trying to create a pair of functions that, given a list of "starting" numbers, will recursively add to each index position up to a defined maximum value (much in the same way that a odometer works in a car--each counter wheel increasing to 9 before resetting to 1 and carrying over onto the next wheel).
The code looks like this:
number_list = []
def counter(start, i, max_count):
if start[len(start)-1-i] < max_count:
start[len(start)-1-i] += 1
return(start, i, max_count)
else:
for j in range (len(start)):
if start[len(start)-1-i-j] == max_count:
start[len(start)-1-i-j] = 1
else:
start[len(start)-1-i-j] += 1
return(start, i, max_count)
def all_values(fresh_start, i, max_count):
number_list.append(fresh_start)
new_values = counter(fresh_start,i,max_count)
if new_values != None:
all_values(*new_values)
When I run all_values([1,1,1],0,3) and print number_list, though, I get:
[[1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1],
[1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1],
[1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1], [1, 1, 1],
[1, 1, 1], [1, 1, 1], [1, 1, 1]]
Which is unfortunate. Doubly so knowing that if I replace the first line of all_values with
print(fresh_start)
I get exactly what I'm after:
[1, 1, 1]
[1, 1, 2]
[1, 1, 3]
[1, 2, 1]
[1, 2, 2]
[1, 2, 3]
[1, 3, 1]
[1, 3, 2]
[1, 3, 3]
[2, 1, 1]
[2, 1, 2]
[2, 1, 3]
[2, 2, 1]
[2, 2, 2]
[2, 2, 3]
[2, 3, 1]
[2, 3, 2]
[2, 3, 3]
[3, 1, 1]
[3, 1, 2]
[3, 1, 3]
[3, 2, 1]
[3, 2, 2]
[3, 2, 3]
[3, 3, 1]
[3, 3, 2]
[3, 3, 3]
I have already tried making a copy of fresh_start (by way of temp = fresh_start) and appending that instead, but with no change in the output.
Can anyone offer any insight as to what I might do to fix my code? Feedback on how the problem could be simplified would be welcome as well.
Thanks a lot!
temp = fresh_start
does not make a copy. Appending doesn't make copies, assignment doesn't make copies, and pretty much anything that doesn't say it makes a copy doesn't make a copy. If you want a copy, slice it:
fresh_start[:]
is a copy.
Try the following in the Python interpreter:
>>> a = [1,1,1]
>>> b = []
>>> b.append(a)
>>> b.append(a)
>>> b.append(a)
>>> b
[[1, 1, 1], [1, 1, 1], [1, 1, 1]]
>>> b[2][2] = 2
>>> b
[[1, 1, 2], [1, 1, 2], [1, 1, 2]]
This is a simplified version of what's happening in your code. But why is it happening?
b.append(a) isn't actually making a copy of a and stuffing it into the array at b. It's making a reference to a. It's like a bookmark in a web browser: when you open a webpage using a bookmark, you expect to see the webpage as it is now, not as it was when you bookmarked it. But that also means that if you have multiple bookmarks to the same page, and that page changes, you'll see the changed version no matter which bookmark you follow.
It's the same story with temp = a, and for that matter, a = [1,1,1]. temp and a are "bookmarks" to a particular array which happens to contain three ones. And b in the example above, is a bookmark to an array... which contains three bookmarks to that same array that contains three ones.
So what you do is create a new array and copy in the elements of the old array. The quickest way to do that is to take an array slice containing the whole array, as user2357112 demonstrated:
>>> a = [1,1,1]
>>> b = []
>>> b.append(a[:])
>>> b.append(a[:])
>>> b.append(a[:])
>>> b
[[1, 1, 1], [1, 1, 1], [1, 1, 1]]
>>> b[2][2] = 2
>>> b
[[1, 1, 1], [1, 1, 1], [1, 1, 2]]
Much better.
When I look at the desired output I can't help but think about using one of the numpy grid data production functions.
import numpy
first_column, second_column, third_column = numpy.mgrid[1:4,1:4,1:4]
numpy.dstack((first_column.flatten(),second_column.flatten(),third_column.flatten()))
Out[23]:
array([[[1, 1, 1],
[1, 1, 2],
[1, 1, 3],
[1, 2, 1],
[1, 2, 2],
[1, 2, 3],
[1, 3, 1],
[1, 3, 2],
[1, 3, 3],
[2, 1, 1],
[2, 1, 2],
[2, 1, 3],
[2, 2, 1],
[2, 2, 2],
[2, 2, 3],
[2, 3, 1],
[2, 3, 2],
[2, 3, 3],
[3, 1, 1],
[3, 1, 2],
[3, 1, 3],
[3, 2, 1],
[3, 2, 2],
[3, 2, 3],
[3, 3, 1],
[3, 3, 2],
[3, 3, 3]]])
Of course, the utility of this particular approach might depend on the variety of input you need to deal with, but I suspect this could be an interesting way to build the data and numpy is pretty fast for this kind of thing. Presumably if your input list has more elements you could have more min:max arguments fed into mgrid[] and then unpack / stack in a similar fashion.
Here is a simplified version of your program, which works. Comments will follow.
number_list = []
def _adjust_counter_value(counter, n, max_count):
"""
We want the counter to go from 1 to max_count, then start over at 1.
This function adds n to the counter and then returns a tuple:
(new_counter_value, carry_to_next_counter)
"""
assert max_count >= 1
assert 1 <= counter <= max_count
# Counter is in closed range: [1, max_count]
# Subtract 1 so expected value is in closed range [0, max_count - 1]
x = counter - 1 + n
carry, x = divmod(x, max_count)
# Add 1 so expected value is in closed range [1, max_count]
counter = x + 1
return (counter, carry)
def increment_counter(start, i, max_count):
last = len(start) - 1 - i
copy = start[:] # make a copy of the start
add = 1 # start by adding 1 to index
for i_cur in range(last, -1, -1):
copy[i_cur], add = _adjust_counter_value(copy[i_cur], add, max_count)
if 0 == add:
return (copy, i, max_count)
else:
# if we have a carry out of the 0th position, we are done with the sequence
return None
def all_values(fresh_start, i, max_count):
number_list.append(fresh_start)
new_values = increment_counter(fresh_start,i,max_count)
if new_values != None:
all_values(*new_values)
all_values([1,1,1],0,3)
import itertools as it
correct = [list(tup) for tup in it.product(range(1,4), range(1,4), range(1,4))]
assert number_list == correct
Since you want the counters to go from 1 through max_count inclusive, it's a little bit tricky to update each counter. Your original solution was to use several if statements, but here I have made a helper function that uses divmod() to compute each new digit. This lets us add any increment to any digit and will find the correct carry out of the digit.
Your original program never changed the value of i so my revised one doesn't either. You could simplify the program further by getting rid of i and just having increment_counter() always go to the last position.
If you run a for loop to the end without calling break or return, the else: case will then run if there is one present. Here I added an else: case to handle a carry out of the 0th place in the list. If there is a carry out of the 0th place, that means we have reached the end of the counter sequence. In this case we return None.
Your original program is kind of tricky. It has two explicit return statements in counter() and an implicit return at the end of the sequence. It does return None to signal that the recursion can stop, but the way it does it is too tricky for my taste. I recommend using an explicit return None as I showed.
Note that Python has a module itertools that includes a way to generate a counter series like this. I used it to check that the result is correct.
I'm sure you are writing this to learn about recursion, but be advised that Python isn't the best language for recursive solutions like this one. Python has a relatively shallow recursion stack, and does not automatically turn tail recursion into an iterative loop, so this could cause a stack overflow inside Python if your recursive calls nest enough times. The best solution in Python would be to use itertools.product() as I did to just directly generate the desired counter sequence.
Since your generated sequence is a list of lists, and itertools.product() produces tuples, I used a list comprehension to convert each tuple into a list, so the end result is a list of lists, and we can simply use the Python == operator to compare them.