[RTM] Ipopt options were problematic

bioptim/interfaces/solver_options.py

@@ -166,11 +166,11 @@ def tol(self):
 
         @property
         def dual_inf_tol(self):
-            return self._constr_viol_tol
+            return self._dual_inf_tol
 
         @property
         def constr_viol_tol(self):
-            return self._tol
+            return self._constr_viol_tol
 
         @property
         def compl_inf_tol(self):
@@ -230,7 +230,7 @@ def warm_start_bound_push(self):
 
         @property
         def warm_start_slack_bound_frac(self):
-            return self._warm_start_init_point
+            return self._warm_start_slack_bound_frac
 
         @property
         def warm_start_bound_frac(self):
@@ -256,10 +256,10 @@ def set_tol(self, val: float):
             self._tol = val
 
         def set_dual_inf_tol(self, val: float):
-            self._constr_viol_tol = val
+            self._dual_inf_tol = val
 
         def set_constr_viol_tol(self, val: float):
-            self._tol = val
+            self._constr_viol_tol = val
 
         def set_compl_inf_tol(self, val: float):
             self._compl_inf_tol = val

tests/test_solver_options.py

@@ -0,0 +1,122 @@
+from bioptim import Solver
+from bioptim.misc.enums import SolverType
+
+
+class FakeSolver:
+    def __init__(
+        self,
+        options_common: dict = None,
+    ):
+        self.options_common = options_common
+
+
+def test_ipopt_solver_options():
+    solver = Solver.IPOPT()
+    assert solver.type == SolverType.IPOPT
+    assert solver.show_online_optim is False
+    assert solver.show_options is None
+    assert solver.tol == 1e-6
+    assert solver.dual_inf_tol == 1.0
+    assert solver.constr_viol_tol == 0.0001
+    assert solver.compl_inf_tol == 0.0001
+    assert solver.acceptable_tol == 1e-6
+    assert solver.acceptable_dual_inf_tol == 1e10
+    assert solver.acceptable_constr_viol_tol == 1e-2
+    assert solver.acceptable_compl_inf_tol == 1e-2
+    assert solver.max_iter == 1000
+    assert solver.hessian_approximation == "exact"
+    assert solver.limited_memory_max_history == 50
+    assert solver.linear_solver == "mumps"
+    assert solver.mu_init == 0.1
+    assert solver.warm_start_init_point == "no"
+    assert solver.warm_start_mult_bound_push == 0.001
+    assert solver.warm_start_slack_bound_push == 0.001
+    assert solver.warm_start_bound_push == 0.001
+    assert solver.warm_start_slack_bound_frac == 0.001
+    assert solver.warm_start_bound_frac == 0.001
+    assert solver.bound_push == 0.01
+    assert solver.bound_frac == 0.01
+    assert solver.print_level == 5
+    assert solver.c_compile is False
+
+    solver.set_linear_solver("ma57")
+    assert solver.linear_solver == "ma57"
+    solver.set_tol(2)
+    assert solver.tol == 2
+    solver.set_dual_inf_tol(3)
+    assert solver.dual_inf_tol == 3
+    solver.set_constr_viol_tol(4)
+    assert solver.constr_viol_tol == 4
+    solver.set_compl_inf_tol(5)
+    assert solver.compl_inf_tol == 5
+    solver.set_acceptable_tol(6)
+    assert solver.acceptable_tol == 6
+    solver.set_acceptable_dual_inf_tol(7)
+    assert solver.acceptable_dual_inf_tol == 7
+    solver.set_acceptable_constr_viol_tol(8)
+    assert solver.acceptable_constr_viol_tol == 8
+    solver.set_acceptable_compl_inf_tol(9)
+    assert solver.acceptable_compl_inf_tol == 9
+    solver.set_maximum_iterations(10)
+    assert solver.max_iter == 10
+    solver.set_hessian_approximation("hello bioptim")
+    assert solver.hessian_approximation == "hello bioptim"
+    solver.set_limited_memory_max_history(11)
+    assert solver.limited_memory_max_history == 11
+    solver.set_mu_init(12)
+    assert solver.mu_init == 12
+    solver.set_warm_start_init_point("how are you?")
+    assert solver.warm_start_init_point == "how are you?"
+    solver.set_warm_start_mult_bound_push(13)
+    assert solver.warm_start_mult_bound_push == 13
+    solver.set_warm_start_slack_bound_push(14)
+    assert solver.warm_start_slack_bound_push == 14
+    solver.set_warm_start_bound_push(15)
+    assert solver.warm_start_bound_push == 15
+    solver.set_warm_start_slack_bound_frac(16)
+    assert solver.warm_start_slack_bound_frac == 16
+    solver.set_warm_start_bound_frac(17)
+    assert solver.warm_start_bound_frac == 17
+    solver.set_bound_push(18)
+    assert solver.bound_push == 18
+    solver.set_bound_frac(19)
+    assert solver.bound_frac == 19
+    solver.set_print_level(20)
+    assert solver.print_level == 20
+    solver.set_c_compile(True)
+    assert solver.c_compile is True
+
+    solver.set_convergence_tolerance(21)
+    assert solver.tol == 21
+    assert solver.acceptable_tol == 21
+    assert solver.compl_inf_tol == 21
+    assert solver.acceptable_compl_inf_tol == 21
+
+    solver.set_constraint_tolerance(22)
+    assert solver.constr_viol_tol == 22
+    assert solver.acceptable_constr_viol_tol == 22
+
+    solver.set_warm_start_options(42)
+    assert solver.warm_start_init_point == "yes"
+    assert solver.mu_init == 42
+    assert solver.warm_start_mult_bound_push == 42
+    assert solver.warm_start_slack_bound_push == 42
+    assert solver.warm_start_bound_push == 42
+    assert solver.warm_start_slack_bound_frac == 42
+    assert solver.warm_start_bound_frac == 42
+
+    solver.set_initialization_options(44)
+    assert solver.bound_push == 44
+    assert solver.bound_frac == 44
+
+    solver.set_option_unsafe(666, "mysterious option")
+    assert solver.__dict__["_mysterious option"] == 666
+
+    fake_solver = FakeSolver(options_common={"ipopt.casino_gain": 777})
+    solver_dict = solver.as_dict(fake_solver)
+    assert solver_dict["ipopt.casino_gain"] == 777
+    assert solver_dict["ipopt.tol"] == 21
+    assert not "_c_compile" in solver_dict
+    assert not "type" in solver_dict
+    assert not "show_online_optim" in solver_dict
+    assert not "show_options" in solver_dict