List replication operator
Steven D'Aprano
steve+comp.lang.python at pearwood.info
Thu May 24 22:30:30 EDT 2018
On Thu, 24 May 2018 16:05:32 -0700, Paul wrote:
> How would one make a multi-dimensional list now, with truly-separate sub
> lists? Is there just no way to do it with the replication operator?
Correct. Let's say you want to make a 1-D list with three items
initialised to zero. This works brilliantly:
py> [0]*3
[0, 0, 0]
This seems like it ought to create a 3x3 2-D list:
py> y = [[0]*3]*3
py> y
[[0, 0, 0], [0, 0, 0], [0, 0, 0]]
but alas, it's a trap:
py> y[0][0] = 1
py> y
[[1, 0, 0], [1, 0, 0], [1, 0, 0]]
The current recommended solution is:
py> y = [[0]*3 for _ in range(3)]
py> y[0][0] = 1
py> y
[[1, 0, 0], [0, 0, 0], [0, 0, 0]]
To get a three-dimensional 3x3x3 list is even more work:
py> z = [[[0]*3 for _ in range(3)] for _ in range(3)]
py> z[0][0][0] = 1
py> z[1][1][1] = 2
py> z[2][2][2] = 3
py> z
[[[1, 0, 0], [0, 0, 0], [0, 0, 0]],
[[0, 0, 0], [0, 2, 0], [0, 0, 0]],
[[0, 0, 0], [0, 0, 0], [0, 0, 3]]]
With my suggestion, we get:
x = [0]**3 # one-dimensional
y = [[0]**3]**3 # two-dimensional
z = [[[0]**3]**3]**3 # three-dimensional
Or there's MRAB's suggestion of using @ instead.
The one-dimensional case can be optimized by using regular * replication
instead of ** duplication, but that's an optimization for immutable
objects.
Here's a subclass that implements a simple version of this for testing:
class ML(list):
def __pow__(self, other):
import copy
L = []
for i in range(other):
L.extend(copy.deepcopy(obj) for obj in self)
return ML(L)
And in use to generate a 3-D list:
py> z = ML([ML([ML([0])**3])**3])**3
py> z[0][0][0] = 1
py> z[1][1][1] = 2
py> z[2][2][2] = 3
py> z
[[[1, 0, 0], [0, 0, 0], [0, 0, 0]],
[[0, 0, 0], [0, 2, 0], [0, 0, 0]],
[[0, 0, 0], [0, 0, 0], [0, 0, 3]]]
The repeated calls to ML() are ugly and are only there because the []
syntax creates ordinary lists, not my subclass.
--
Steve
More information about the Python-list
mailing list