def __init__(self, solver):
super(CplexResults, self).__init__()
self.wallclock_time = None
self.solution_loader = PersistentSolutionLoader(solver=solver)
def _parse_sol(self):
solve_vars = self._writer.get_ordered_vars()
solve_cons = self._writer.get_ordered_cons()
results = Results()
f = open(self._filename + '.sol', 'r')
all_lines = list(f.readlines())
f.close()
termination_line = all_lines[1]
if 'Optimal Solution Found' in termination_line:
results.termination_condition = TerminationCondition.optimal
elif 'Problem may be infeasible' in termination_line:
results.termination_condition = TerminationCondition.infeasible
elif 'problem might be unbounded' in termination_line:
results.termination_condition = TerminationCondition.unbounded
elif 'Maximum Number of Iterations Exceeded' in termination_line:
results.termination_condition = TerminationCondition.maxIterations
elif 'Maximum CPU Time Exceeded' in termination_line:
results.termination_condition = TerminationCondition.maxTimeLimit
else:
results.termination_condition = TerminationCondition.unknown
n_cons = len(solve_cons)
n_vars = len(solve_vars)
dual_lines = all_lines[12:12 + n_cons]
primal_lines = all_lines[12 + n_cons:12 + n_cons + n_vars]
rc_upper_info_line = all_lines[12 + n_cons + n_vars + 1]
assert rc_upper_info_line.startswith('suffix')
n_rc_upper = int(rc_upper_info_line.split()[2])
assert 'ipopt_zU_out' in all_lines[12 + n_cons + n_vars + 2]
upper_rc_lines = all_lines[12 + n_cons + n_vars + 3:12 + n_cons +
n_vars + 3 + n_rc_upper]
rc_lower_info_line = all_lines[12 + n_cons + n_vars + 3 + n_rc_upper]
assert rc_lower_info_line.startswith('suffix')
n_rc_lower = int(rc_lower_info_line.split()[2])
assert 'ipopt_zL_out' in all_lines[12 + n_cons + n_vars + 3 +
n_rc_upper + 1]
lower_rc_lines = all_lines[12 + n_cons + n_vars + 3 + n_rc_upper +
2:12 + n_cons + n_vars + 3 + n_rc_upper +
2 + n_rc_lower]
self._dual_sol = dict()
self._primal_sol = ComponentMap()
self._reduced_costs = ComponentMap()
for ndx, dual in enumerate(dual_lines):
dual = float(dual)
con = solve_cons[ndx]
self._dual_sol[con] = dual
for ndx, primal in enumerate(primal_lines):
primal = float(primal)
var = solve_vars[ndx]
self._primal_sol[var] = primal
for rcu_line in upper_rc_lines:
split_line = rcu_line.split()
var_ndx = int(split_line[0])
rcu = float(split_line[1])
var = solve_vars[var_ndx]
self._reduced_costs[var] = rcu
for rcl_line in lower_rc_lines:
split_line = rcl_line.split()
var_ndx = int(split_line[0])
rcl = float(split_line[1])
var = solve_vars[var_ndx]
if var in self._reduced_costs:
if abs(rcl) > abs(self._reduced_costs[var]):
self._reduced_costs[var] = rcl
else:
self._reduced_costs[var] = rcl
for var in solve_vars:
if var not in self._reduced_costs:
self._reduced_costs[var] = 0
if results.termination_condition == TerminationCondition.optimal and self.config.load_solution:
for v, val in self._primal_sol.items():
v.set_value(val, skip_validation=True)
if self._writer.get_active_objective() is None:
results.best_feasible_objective = None
else:
results.best_feasible_objective = value(
self._writer.get_active_objective().expr)
elif results.termination_condition == TerminationCondition.optimal:
if self._writer.get_active_objective() is None:
results.best_feasible_objective = None
else:
obj_expr_evaluated = replace_expressions(
self._writer.get_active_objective().expr,
substitution_map={
id(v): val
for v, val in self._primal_sol.items()
},
descend_into_named_expressions=True,
remove_named_expressions=True)
results.best_feasible_objective = value(obj_expr_evaluated)
elif self.config.load_solution:
raise RuntimeError(
'A feasible solution was not found, so no solution can be loaded.'
'Please set opt.config.load_solution=False and check '
'results.termination_condition and '
'resutls.best_feasible_objective before loading a solution.')
results.solution_loader = PersistentSolutionLoader(solver=self)
return results
def _parse_soln(self):
results = Results()
f = open(self._filename + '.soln', 'r')
all_lines = list(f.readlines())
f.close()
termination_line = all_lines[0].lower()
obj_val = None
if termination_line.startswith('optimal'):
results.termination_condition = TerminationCondition.optimal
obj_val = float(termination_line.split()[-1])
elif 'infeasible' in termination_line:
results.termination_condition = TerminationCondition.infeasible
elif 'unbounded' in termination_line:
results.termination_condition = TerminationCondition.unbounded
elif termination_line.startswith('stopped on time'):
results.termination_condition = TerminationCondition.maxTimeLimit
obj_val = float(termination_line.split()[-1])
elif termination_line.startswith('stopped on iterations'):
results.termination_condition = TerminationCondition.maxIterations
obj_val = float(termination_line.split()[-1])
else:
results.termination_condition = TerminationCondition.unknown
first_con_line = None
last_con_line = None
first_var_line = None
last_var_line = None
for ndx, line in enumerate(all_lines):
if line.strip('*').strip().startswith('0'):
if first_con_line is None:
first_con_line = ndx
else:
last_con_line = ndx - 1
first_var_line = ndx
last_var_line = len(all_lines) - 1
self._dual_sol = dict()
self._primal_sol = dict()
self._reduced_costs = dict()
symbol_map = self._writer.symbol_map
for line in all_lines[first_con_line:last_con_line+1]:
split_line = line.strip('*')
split_line = split_line.split()
name = split_line[1]
orig_name = name[:-3]
if orig_name == 'obj_const_con':
continue
con = symbol_map.bySymbol[orig_name]()
dual_val = float(split_line[-1])
if con in self._dual_sol:
if abs(dual_val) > abs(self._dual_sol[con]):
self._dual_sol[con] = dual_val
else:
self._dual_sol[con] = dual_val
for line in all_lines[first_var_line:last_var_line+1]:
split_line = line.strip('*')
split_line = split_line.split()
name = split_line[1]
if name == 'obj_const':
continue
val = float(split_line[2])
rc = float(split_line[3])
var = symbol_map.bySymbol[name]()
self._primal_sol[id(var)] = (var, val)
self._reduced_costs[id(var)] = (var, rc)
if (self.version() < (2, 10, 2) and
self._writer.get_active_objective() is not None and
self._writer.get_active_objective().sense == maximize):
if obj_val is not None:
obj_val = -obj_val
for con, dual_val in self._dual_sol.items():
self._dual_sol[con] = -dual_val
for v_id, (v, rc_val) in self._reduced_costs.items():
self._reduced_costs[v_id] = (v, -rc_val)
if results.termination_condition == TerminationCondition.optimal and self.config.load_solution:
for v_id, (v, val) in self._primal_sol.items():
v.set_value(val, skip_validation=True)
if self._writer.get_active_objective() is None:
results.best_feasible_objective = None
else:
results.best_feasible_objective = obj_val
elif results.termination_condition == TerminationCondition.optimal:
if self._writer.get_active_objective() is None:
results.best_feasible_objective = None
else:
results.best_feasible_objective = obj_val
elif self.config.load_solution:
raise RuntimeError('A feasible solution was not found, so no solution can be loaded.'
'Please set opt.config.load_solution=False and check '
'results.termination_condition and '
'resutls.best_feasible_objective before loading a solution.')
results.solution_loader = PersistentSolutionLoader(solver=self)
return results