CODEWITHSUNDEEP
pythonparameterspyomo
James Vance
James Vance

Reputation: 177

How to index a matrix in Pyomo

I'm trying to create a Pyomo index that would apply to both the column names and the row index of a pandas dataframe.

The reason being that I have constrains on each column but would also like to subset the df rows into 4 parts, with various constraints on each part. I can get this formulation to work by creating 4 seperate Params indexed only on the matrix column names, but this is inflexible if the subset needs to be a different number of parts.

My code attempting to subset the x and y of a matrix is below: 

model=ConcreteModel()
model.N = Set(initialize=col_names)
model.I = Set(initialize=full_df.index)
model.M = Param(model.I, model.N, initialize=full_df[col_names].values)

Any help would be appreciated.

Upvotes: 0

Views: 769

Answers (1)

V. Brunelle
V. Brunelle

Reputation: 1068

Param can support dictionaries, and dictionaries support tuple keys.

  1. Create a set of (i,n) tuples, where i is element of I and n is element of N. This set, instead of containing numbers or string, will contain tuples.

    model.your_set = Set(initialize=list_of_tuples)

  2. Make sure you have a dictionary or data object representation that allows one and only one numeric value per tuple (i.e. the content of your matrix). Let's unoriginally call it d.

  3. The parameter must be declared this way:

    model.M = Param(model.your_set, initialize=d)

That way, your tuples will serve as the keys of the dict.

Upvotes: 1

Related Questions