Open
Description
Summary
On /pyomo/solvers/plugins/solvers/CBCplugin.py line 418, the command line for the CBC solver is created, with the following arguments:
cmd.extend(["-stat=1", "-solve", "-solu", self._soln_file])
This causes the command line to have the options above as the last ones in the CLI, in particular the option solve.
Then there's the initialSolve option which, as CBC cli reports, is used to solve the original problem just as a continuous problem.
Indeed, it is used by PuLP to implement its mip option (see here)
However, passing initialSolve as an option via pyomo puts it before solve, listed in the previous list, causing it to be ignored.
Note: this may not be a big issue specifically for initialSolve if one could easily compute the linear relaxation of a model in other ways. It could still be an issue if other CBC options get ignored because of CLI construction.
For linear relaxation, I found this thread on google groups where they mention TransformationFactory('core.relax_integer_vars') as a pyomo-way of transforming a MILP model in a LP one, however, fyi, it does not appear to be documented elsewhere.
Steps to reproduce the issue
Consider
m = ConcreteModel()
m.x = Var([0,1,2,3], domain=Binary)
m.weight=Constraint(expr=sum(m.x[i] for i in list(range(4)))<= 3.5)
m.value=Objective(expr=sum(m.x[i] for i in range(4)), sense=maximize)
m.write("example.lp")
solver = SolverFactory("cbc")
solver.options={"initialSolve":""}
solver.solve(m, tee=True)Its optimal solution is 3.5, but it is solved as a MIP and thus the solution is 3 .
Log
Welcome to the CBC MILP Solver
Version: 2.10.7
Build Date: Feb 14 2022
command line - /usr/bin/cbc -printingOptions all -import /tmp/tmpb1phpxks.pyomo.lp **-initialSolve -stat=1 -solve** -solu /tmp/tmpb1phpxks.pyomo.soln (default strategy 1)
Option for printingOptions changed from normal to all
CoinLpIO::readLp(): Maximization problem reformulated as minimization
Coin0009I Switching back to maximization to get correct duals etc
Presolve 0 (-2) rows, 0 (-5) columns and 0 (-5) elements
Empty problem - 0 rows, 0 columns and 0 elements
Optimal - objective value 3.5
After Postsolve, objective 3.5, infeasibilities - dual 0 (0), primal 0 (0)
Optimal objective 3.5 - 0 iterations time 0.002, Presolve 0.00
Presolve 1 (-1) rows, 4 (-1) columns and 4 (-1) elements
Statistics for presolved model
Original problem has 4 integers (4 of which binary)
Presolved problem has 4 integers (4 of which binary)
==== 0 zero objective 1 different
4 variables have objective of 1
==== absolute objective values 1 different
4 variables have objective of 1
==== for integers 0 zero objective 1 different
4 variables have objective of 1
==== for integers absolute objective values 1 different
4 variables have objective of 1
===== end objective counts
Problem has 1 rows, 4 columns (4 with objective) and 4 elements
There are 4 singletons with objective
Column breakdown:
0 of type 0.0->inf, 0 of type 0.0->up, 0 of type lo->inf,
0 of type lo->up, 0 of type free, 0 of type fixed,
0 of type -inf->0.0, 0 of type -inf->up, 4 of type 0.0->1.0
Row breakdown:
0 of type E 0.0, 0 of type E 1.0, 0 of type E -1.0,
0 of type E other, 0 of type G 0.0, 0 of type G 1.0,
0 of type G other, 0 of type L 0.0, 0 of type L 1.0,
1 of type L other, 0 of type Range 0.0->1.0, 0 of type Range other,
0 of type Free
Continuous objective value is 3.5 - 0.00 seconds
Cgl0004I processed model has 0 rows, 0 columns (0 integer (0 of which binary)) and 0 elements
Cbc3007W No integer variables - nothing to do
Cuts at root node changed objective from -3 to -1.79769e+308
Probing was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Gomory was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Knapsack was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Clique was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
MixedIntegerRounding2 was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
FlowCover was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
TwoMirCuts was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
ZeroHalf was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Result - Optimal solution found
Objective value: 3.00000000
Enumerated nodes: 0
Total iterations: 0
Time (CPU seconds): 0.00
Time (Wallclock seconds): 0.00
Total time (CPU seconds): 0.00 (Wallclock seconds): 0.00
Solving the example.lp file with CBC CLI, specifying just "initialSolve" works as intended.
$ cbc example.lp -initialSolve
Welcome to the CBC MILP Solver
Version: 2.10.7
Build Date: Feb 14 2022
command line - cbc example.lp -initialSolve (default strategy 1)
CoinLpIO::readLp(): Maximization problem reformulated as minimization
Coin0009I Switching back to maximization to get correct duals etc
Presolve 0 (-2) rows, 0 (-5) columns and 0 (-5) elements
Empty problem - 0 rows, 0 columns and 0 elements
Optimal - objective value 3.5
After Postsolve, objective 3.5, infeasibilities - dual 0 (0), primal 0 (0)
Optimal objective 3.5 - 0 iterations time 0.002, Presolve 0.00
Total time (CPU seconds): 0.00 (Wallclock seconds): 0.00
while adding -solve after it recreates the issue
$ cbc example.lp -initialSolve -solve
Welcome to the CBC MILP Solver
Version: 2.10.7
Build Date: Feb 14 2022
command line - cbc example.lp -initialSolve -solve (default strategy 1)
CoinLpIO::readLp(): Maximization problem reformulated as minimization
Coin0009I Switching back to maximization to get correct duals etc
Presolve 0 (-2) rows, 0 (-5) columns and 0 (-5) elements
Empty problem - 0 rows, 0 columns and 0 elements
Optimal - objective value 3.5
After Postsolve, objective 3.5, infeasibilities - dual 0 (0), primal 0 (0)
Optimal objective 3.5 - 0 iterations time 0.002, Presolve 0.00
Continuous objective value is 3.5 - 0.00 seconds
Cgl0004I processed model has 0 rows, 0 columns (0 integer (0 of which binary)) and 0 elements
Cbc3007W No integer variables - nothing to do
Cuts at root node changed objective from -3 to -1.79769e+308
Probing was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Gomory was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Knapsack was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Clique was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
MixedIntegerRounding2 was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
FlowCover was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
TwoMirCuts was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
ZeroHalf was tried 0 times and created 0 cuts of which 0 were active after adding rounds of cuts (0.000 seconds)
Result - Optimal solution found
Objective value: 3.00000000
Enumerated nodes: 0
Total iterations: 0
Time (CPU seconds): 0.00
Time (Wallclock seconds): 0.00
Total time (CPU seconds): 0.00 (Wallclock seconds): 0.00