Proper way to define a matrix using pyomo's sets

Question

I have been given a pyomo deterministic optimization model which I have transformed in a stochastic one using stochastic colocation methods.

However, I am having issues to save the output. I have defined the number of stochastic nodes (number of values of the state variables)

m.particle_number = Param(initialize = count)
m.K = Set(initialize = range(1,value(m.particle_number)+1))

After this, the model defines a series of lists of length m.n, number of points in which the model is evaluated. Here is an example.

position_phi     = list(range(0,value(m.n)+1))

I need to turn these 1D lists into m.K dimensions. I have tried to write

position_phi     = list(range(0,value(m.n)+1), range(0,value(m.K)+1))

instead. However, I get the following error:

TypeError: Cannot evaluate object with unknown type: SimpleSet

Can someone explain to me why I can’t construct a matrix of size m.n*m.k?

Asked By: slow_learner

||

Source

Answer 1

I’m not completely clear what your model is attempting to do, but you can create a cross product of two sets in Pyomo by multiplying the two.

import pyomo.environ as pe
m = pe.ConcreteModel()
m.K = pe.Set(initialize=[1, 2, 3])
m.n = pe.Set(initialize=[4, 5, 6, 7])
m.position_phi = m.K * m.n

This will make the elements of m.position_phi be [(1, 4), (1, 5), …]

You can then use this set as follows:

m.whatever = pe.Param(m.position_phi, intitialize={(1, 4):4, (3, 6):6}, default=0)

And then when you call m.pprint() you’ll see something like this:

3 Set Declarations
    K : Dim=0, Dimen=1, Size=3, Domain=None, Ordered=False, Bounds=(1, 3)
        [1, 2, 3]
    n : Dim=0, Dimen=1, Size=4, Domain=None, Ordered=False, Bounds=(4, 7)
        [4, 5, 6, 7]
    position_phi : Dim=0, Dimen=2, Size=12, Domain=None, Ordered=False, Bounds=None
        Virtual

1 Param Declarations
    whatever : Size=12, Index=position_phi, Domain=Any, Default=0, Mutable=False
        Key    : Value
        (1, 4) :     4
        (3, 6) :     6

4 Declarations: K n position_phi whatever

Also, calling list() with more than 1 argument in Python should be throwing a TypeError.

Answered By: Conor

Answer 2

We have a matrix N (4 X4)

−1 0 0 0
1 −1 0 0
0 1 −1 0
0 0 1 −1

And 4 differential equations ds/dt = Nv
S and v are 4-dimensional vectors
Is there anyway I can in pyomo make a vector matrix operation and declare the 4 differential equations without declaring each of the equations individually
Thanks

Answered By: Lakshmi N Sridhar

Proper way to define a matrix using pyomo's sets

Question:

Answers: