Can I generate a pulp problem with a MPS file?

Question

I have a MPS file like this:

* ENCODING=ISO-8859-1
NAME          test.lp
ROWS
 N  obj1       
 L  Constraint1
 G  Constraint2
COLUMNS
    x            obj1                              -3
    x            Constraint1                        1
    x            Constraint2                        2
    MARK0000  'MARKER'                 'INTORG'
    y            obj1                              -2
    y            Constraint1                        2
    y            Constraint2                       -1
    MARK0001  'MARKER'                 'INTEND'
RHS
    rhs          Constraint1                        5
    rhs          Constraint2                        3
BOUNDS
 UP bnd          x                                 10
 LI bnd          y            0
ENDATA

Can I generate a pulp problem with it? It can generate automatic or I have to read each line?

I've search on the internet and I can export it like this:

var1, prob1 = LpProblem.fromMPS("test.mps")

I am thinking to generate the problem by take each variable in var1 and so on. Have any way to do it automatically? Thanks

Answer 1

Yes, you can generate a pulp problem from an mps file.

Please note that the method pulp.LpProblem.fromMPS("test.mps") that you've found imports a problem from an MPS file, rather than exporting it.

You can find more information on how to import and export models using PuLP here: How to import and export models in PuLP

Having said that, the mps file you've included in your example is missing the information about the problem's "sense" (whether to minimize or maximize your objective value). When you import your problem using pulp.LpProblem.fromMPS("test.mps"), PuLP will interpret it as being a minimization problem, when no sense is included inside the mps file. If you want to maximize the objective value, you can include the parameter sense to .fromMPS() method like so:

import pulp

mps_filename = "test.mps"

# Include:
# sense = -1, to maximize the objective function
# sense = 1 (default), to minimize the objective function
lpvars, model = pulp.LpProblem.fromMPS(mps_filename, sense=-1)

status = model.solve(pulp.PULP_CBC_CMD(msg=False))
status_str = pulp.LpStatus[status]
print(status_str)
# Prints: 'Optimal'

NOTES

The mps format is an industry standard. But it is not very flexible with some information not being stored. It stores only variables and constraints. It does not store the values of variables. In other words, after importing the model, you need to call prob1.solve() to re-generate the variables optimal values.