Index names of decision variables implicitly changes when added via gurobi pandas method

Answered

Madhushini Narayana Prasad

• May 31, 2023 05:58

I have decision variables with index names like AB-101, AN-102, AV-104 etc. These indices become AB_101 when added via gurobipy_pandas.add_vars method and remain as-is when added via model.addVars method. This creates confusion while trying to fetch these values post-processing. Is there a way to mitigate this?

Related to

• gurobipy-pandas

0

• 

  Riley Clement

  Gurobi Staff

  • May 31, 2023 06:43

  Hi Madhushini,

  Can you share how you are creating variables?  This may influence what can be done.

  For example in the following snippet:

  import gurobipy as gp
  import gurobipy_pandas as gppd
  import pandas as pd
  
  m = gp.Model()
  vars = gppd.add_vars(
      model=m, 
      pandas_obj=pd.Series(range(10)),
      name=pd.Series([f"AB-{100+i}" for i in range(10)]),
  )

  the variable names are AB-100, AB-101, ... , AB-109 and there is no conversion to underscores (at least not with pandas 2).

  - Riley

  0
• 

  Madhushini Narayana Prasad

  • May 31, 2023 06:55

  

  

  

  Hi Riley,

  I am reading these values from an excel sheet. Please see the code snippet below:

  lstNames = ['AB-101','AC-102','BG-203']

  weekLst = list(range(0, 21))

  tmp_data = pandas.DataFrame(list(product(lstNames, weekLst)), columns=['Name', 'Age'])

  opt_data=tmp_data.set_index(["Name","Age"])

   

  m = gp.Model()

  CV_StartVars = gppd.add_vars(m, opt_data, name="dens", lb=0, ub=GRB.INFINITY, vtype=GRB.CONTINUOUS)

  CV_EndVars = m.addVars(lstNames, lb=0, ub=GRB.INFINITY, vtype=GRB.CONTINUOUS, name="Yale" )

   

  m.optimize()

  var_vals = m.getVars()

  print(var_vals)

   

  

   

  0
• 

  Riley Clement

  Gurobi Staff

  • May 31, 2023 23:24
  • Edited

  Hi Madhushini,

  I can confirm that gurobipy-pandas is deliberately replacing "-" with "_" in the names.  As an aside, it is also replacing "+", "-", "*", "^" with "_". 

  If you wish to avoid this default behavior you can add the index_formatter argument to add_vars:

  CV_StartVars = gppd.add_vars(
      m,
      opt_data,
      name="dens",
      lb=0,
      ub=GRB.INFINITY,
      vtype=GRB.CONTINUOUS,
      index_formatter="disable",
  )

  - Riley

  0
• 

  Madhushini Narayana Prasad

  • May 31, 2023 18:17

  

  

  

  Thanks Riley. Unfortunately it gives me the following error when I try adding the index_formatter to add_vars:

  File "/opt/anaconda3/envs/shrimpfeed/lib/python3.9/site-packages/gurobipy_pandas/api.py", line 158, in add_vars
      return add_vars_from_dataframe(
    File "/opt/anaconda3/envs/shrimpfeed/lib/python3.9/site-packages/gurobipy_pandas/variables.py", line 170, in add_vars_from_dataframe
      namearg = create_names(name, data.index, index_formatter)
    File "/opt/anaconda3/envs/shrimpfeed/lib/python3.9/site-packages/gurobipy_pandas/util.py", line 39, in create_names
      mapper = create_mapper(index_formatter)
    File "/opt/anaconda3/envs/shrimpfeed/lib/python3.9/site-packages/gurobipy_pandas/index_mappers.py", line 21, in create_mapper
      if arg == "disable":
  ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
  Warning: environment still referenced so free is deferred
  Warning: remote job 1655ab5b-ea63-42b3-8f35-2441dae9dcfe on server http://gurobi-dev.cargill.com:7183 killed because environment was freed

  0
• 

  Riley Clement

  Gurobi Staff

  • May 31, 2023 23:20

  Hi Madhushini,

  Apologies, I did not pick up this error.  I've since realized the simplest solution is to use

  index_formatter="disable"

  I have updated my answer above to reflect this.

  - Riley

  0
• 

  Madhushini Narayana Prasad

  • June 01, 2023 17:07

  

  

  

  It works now, thanks a lot for your help and support Riley.

  0

Please sign in to leave a comment.