CODEWITHSUNDEEP
pythonpython-3.xpulp
Suraj Nagabhushana Ponnaganti
Suraj Nagabhushana Ponnaganti

Reputation: 1257

How to create a 3 dimensional dictionary from 3 lists in Python

I have 3 lists

i = ['a','b','c']
j = ['1','2','3']
l = ['01','02']

I want to create a dictionary looking something like this:

D = {'a':{'1':{'01':10, '02':100}, {'2':{'01':20,'02':50}}}

I want to be able to loop over i, j and l and do D[i][j][l] to get the value associated with it.

Ds = {}
l_ = 0
for i_ in range(len(i)):
    j_ = 0
    for j_ in range(len(j)):
        l_ = 0
        for l_ in range(len(l)): 
            Ds[i[i_]] = {j[j_]:{l[l_]:'some value'}}

I tried this, but this does not work.

Extra: I am writing a Linear program in Pulp and I want to create a dictionary of demand values for a particular sku i, in period j on line l so that I can do Dijl*Xijtl = some_value summed over i, j ,l for constraints or something like that. These demand values are constants. This could be cartesian product or comprehension. Is there a function similar to LpVariable.dicts()?

Upvotes: 1

Views: 2177

Answers (5)

Jose Figueroa
Jose Figueroa

Reputation: 61

Since you are using PuLP, you may want to look into pulp's makeDict().

Suppose you have the following:

i = ['a', 'b']
j = ['1', '2', '3']
l = ['01', '02']

and that for your demand values you have:

value_a_1_01 = 10
value_a_1_02 = 20
value_a_2_01 = 30
value_a_2_02 = 40
value_a_3_01 = 50
value_a_3_02 = 60

value_b_1_01 = 70
value_b_1_02 = 80
value_b_2_01 = 90
value_b_2_02 = 100
value_b_3_01 = 110
value_b_3_02 = 120

You can then create an array that contains your demand data like this:

array = [[{'01': 10, '02': 20}, {'01': 30, '02': 40},
          {'01': 50, '02': 60}],
         [{'01': 70, '02': 80}, {'01': 90, '02': 100},
          {'01': 110, '02': 120}]]

and now define your dictionary:

D = pulp.makeDict([i, j], array)

EDIT: Since the data set up was a bit messy before, setting up the array like this makes it much cleaner (and easier to edit):

array = [
         [[10, 20], [30, 40], [50, 60]], 
         [[70, 80], [90, 100], [110, 120]]
        ]

Now the parameters for makeDict() should be:

 D = pulp.makeDict([i, j, l], array)

Upvotes: 1

chthonicdaemon
chthonicdaemon

Reputation: 19810

I think this reads really nicely as a comprehension:

D = {i_: {j_: {l_: "value" for l_ in l} for j_ in j} for i_ in i}

Upvotes: 0

Stuart Mitchell
Stuart Mitchell

Reputation: 1059

Another way of doing this would just be to have a dictionary that is keyed by tuple values i.e 

D ={}
for i in I:
   for j in J:
      for k in K:
         D[i,j,k] = your value

Note in pulp it would be more common to donate your sets with a capital letter and iterate with the smaller letter.

Upvotes: 0

hiro protagonist
hiro protagonist

Reputation: 46899

this is a variant using product and setdefault:

from itertools import product

D = {}
for ii, jj, ll in product(i, j, l):
    D.setdefault(ii, {})
    D[ii].setdefault(jj, {})
    D[ii][jj][ll] = None

where i used None as placeholder for your value.

Upvotes: 1

Dmitry
Dmitry

Reputation: 2096

If you want "Cartesian product" of lists, you can use:

i = ['a','b','c']
j = ['1','2','3']
l = ['01','02']

Ds = {}

for i_ in i:
    item_i = {}
    for j_ in j:
        item_j = {}
        for l_ in l:
            item_j[l_] = '' # Your value                                                                                                                      
        item_i[j_] = item_j
    Ds[i_] = item_i

print Ds
# {'a': {'1': {'02': '', '01': ''}, '3': {'02': '', '01': ''}, '2': {'02': '', '01': ''}}, 'c': {'1': {'02': '', '01': ''}, '3': {'02': '', '01': ''}, '2': {'02': '', '01': ''}}, 'b': {'1': {'02': '', '01': ''}, '3': {'02': '', '01': ''}, '2': {'02': '', '01': ''}}}

Upvotes: 1

Related Questions