Solving optimisation sub-instances in Parallel, using Pyomo ( Traceback )

Question

I am trying to solve an energy model with Benders Decomposition.

In the model we are creating a master model and several sub models.

And I want to solve the sub models in parallel, and I saw an example here.

This is what I am using in the code:

from pyomo.opt.base import SolverFactory
from pyomo.opt.parallel import SolverManagerFactory
from pyomo.opt.parallel.manager import solve_all_instances

subs = []
for m in range(0, len(supportsteps)-1):
    subs.append(urbs.create_model(data,
                        range(supportsteps[m], supportsteps[m+1]+1),
                        supportsteps, type=1))

solver_manager = SolverManagerFactory("pyro")

solve_all_instances(solver_manager, 'gurobi', subs)

Which gives an error:

Error Message

So what I am doing wrong?

Or, is it not possible to solve them in parallel?

Answer 1

The error message that you're seeing means that SolverManagerFactory("pyro") gave you None. It's possible that pyro isn't installed or on your PATH.

Try installing the Pyomo extras: conda install -c conda-forge pyomo.extras or pyomo install-extras