How to acheive cumulative sum of cumulative 3D arrays in gurobi python

Awaiting user input

Gnanadeepan D

• August 27, 2021 17:10

Say I have X1 and X2 which is based on some operations between variables and parameters of the model and so X1 and X2 will be variables.

x1=

[[[1, 6], [2, 7], [3, 8], [4, 9], [5, 10]],

 [[1, 6], [2, 7], [3, 8], [4, 9], [5, 10]],

 [[1, 6], [2, 7], [3, 8], [4, 9], [5, 10]],

 [[1, 6], [2, 7], [3, 8], [4, 9], [5, 10]]]

x2=

[[[0, 0], [1, 6], [2, 7], [3, 8], [4, 9]],

 [[0, 0], [1, 6], [2, 7], [3, 8], [4, 9]],

 [[0, 0], [1, 6], [2, 7], [3, 8], [4, 9]],

 [[0, 0], [1, 6], [2, 7], [3, 8], [4, 9]]]

Then I need the following, that is to check the cumulative sum of cumulative sum of arrays y1>=y2

y1=

[[[1, 21], [3, 28], [6, 36], [10, 45], [15, 55]],

 [[1, 21], [3, 28], [6, 36], [10, 45], [15, 55]],

 [[1, 21], [3, 28], [6, 36], [10, 45], [15, 55]],

 [[1, 21], [3, 28], [6, 36], [10, 45], [15, 55]]]

y2=

[[[0, 10], [1, 16], [3, 23], [6, 31], [10, 40]],

 [[0, 10], [1, 16], [3, 23], [6, 31], [10, 40]],

 [[0, 10], [1, 16], [3, 23], [6, 31], [10, 40]],

 [[0, 10], [1, 16], [3, 23], [6, 31], [10, 40]]]

I have tried

for l in range(2):

   for j in range (5):

      for k in range (4):

            o.addConstr(grb.quicksum(grb.quicksum(x1[i,jj,ll] for jj in n if jj <=j)for ll in d if ll<=l)  >= (grb.quicksum(grb.quicksum(x2[i,jj,ll] for jj in n if jj <=j)for ll in d if ll<=l)))

but that is not correct, Can you help? This is a simple snippet of my complex model. If my complete model needed for clear picture, I can provide those too.

 

0

• 

  Jaromił Najman

  Gurobi Staff

  • September 06, 2021 12:48

  Hi,

  I am not sure what you are trying to achieve. A rather standard way of summing over all values of a multidimensional list would be the use of \(\texttt{for}\)-loops

  res = 0
  for i in x1:
    for j in i:
      for k in j:
        res += k
  
  print(res)

  However, this will not work when \(\texttt{x1}\) is an optimization variable.

  You can introduce \(\texttt{x1,x2}\) as a 3 dimensional optimization variable via the addVars function

  x1 = model.addVars(2,3,4)
  x2 = model.addVars(4,5,6)

  and then sum over all variable using the quicksum function

  model.addConstr(gp.quicksum(x1[i,j,k] for i in range(2) for j in range(3) for k in range(4)
               >= gp.quicksum(x2[i,j,k] for i in range(4) for j in range(5) for k in range(6))

  Does this answer you question?

  Best regards,
  Jaromił

  0

Please sign in to leave a comment.