added new modeling variants, added optunacallback without pruned trial

discrete_optimization/coloring/solvers/coloring_cpsat_solver.py

@@ -31,6 +31,9 @@ class ColoringCPSatSolver(OrtoolsCPSatSolver, SolverColoringWithStartingSolution
     hyperparameters = [
         EnumHyperparameter(name="modeling", enum=ModelingCPSat),
         CategoricalHyperparameter(name="warmstart", choices=[True, False]),
+        CategoricalHyperparameter(name="value_sequence_chain", choices=[True, False]),
+        CategoricalHyperparameter(name="used_variable", choices=[True, False]),
+        CategoricalHyperparameter(name="symmetry_on_used", choices=[True, False]),
     ] + SolverColoringWithStartingSolution.hyperparameters
 
     def __init__(
@@ -112,6 +115,8 @@ def init_model_binary(self, nb_colors: int, **kwargs):
 
     def init_model_integer(self, nb_colors: int, **kwargs):
         used_variable = kwargs.get("used_variable", False)
+        value_sequence_chain = kwargs.get("value_sequence_chain", False)
+        symmetry_on_used = kwargs.get("symmetry_on_used", True)
         cp_model = CpModel()
         variables = [
             cp_model.NewIntVar(0, nb_colors - 1, name=f"c_{i}")
@@ -128,6 +133,17 @@ def init_model_integer(self, nb_colors: int, **kwargs):
                     variables[ind]
                     == self.problem.constraints_coloring.color_constraint[node]
                 )
+        if value_sequence_chain:
+            vars = [variables[i] for i in self.problem.index_subset_nodes]
+            sliding_max = [
+                cp_model.NewIntVar(0, min(i, nb_colors), name=f"m_{i}")
+                for i in range(len(vars))
+            ]
+            cp_model.Add(vars[0] == sliding_max[0])
+            self.variables["sliding_max"] = sliding_max
+            for k in range(1, len(vars)):
+                cp_model.AddMaxEquality(sliding_max[k], [sliding_max[k - 1], vars[k]])
+                cp_model.Add(sliding_max[k] <= sliding_max[k - 1] + 1)
         used = [cp_model.NewBoolVar(name=f"used_{c}") for c in range(nb_colors)]
         if used_variable:
 
@@ -147,6 +163,9 @@ def add_indicator(vars, value, presence_value, model):
                 else:
                     vars = variables
                 add_indicator(vars, j, used[j], cp_model)
+            if symmetry_on_used:
+                for j in range(nb_colors - 1):
+                    cp_model.Add(used[j] >= used[j + 1])
             cp_model.Minimize(sum(used))
         else:
             nbc = cp_model.NewIntVar(0, nb_colors, name="nbcolors")

discrete_optimization/generic_tools/callbacks/optuna.py

@@ -22,6 +22,48 @@
     logger.warning("You should install optuna to use callbacks for optuna.")
 
 
+class OptunaReportSingleFitCallback(Callback):
+    """Callback to report intermediary objective values, this is mostly useful for visualisation purposes.
+
+
+    Adapted to single objective optimization (res.fit is a float)
+
+    Args:
+        trial:
+            A :class:`optuna.trial.Trial` corresponding to the current evaluation of the
+            objective function.
+        optuna_report_nb_steps: report intermediate result every `optuna_report_nb_steps` steps
+            when the number of iterations is high, setting this to 1 could slow too much run of a single trial
+
+    """
+
+    def __init__(
+        self, trial: optuna.trial.Trial, optuna_report_nb_steps: int = 1, **kwargs
+    ) -> None:
+        self.report_nb_steps = optuna_report_nb_steps
+        self.trial = trial
+
+    def on_step_end(
+        self, step: int, res: ResultStorage, solver: SolverDO
+    ) -> Optional[bool]:
+        """Called at the end of an optimization step.
+
+        Args:
+            step: index of step
+            res: current result storage
+            solver: solvers using the callback
+
+        Returns:
+            If `True`, the optimization process is stopped, else it goes on.
+
+        """
+        if step % self.report_nb_steps == 0:
+            fit = res.best_fit
+
+            # Report current score and step to Optuna's trial.
+            self.trial.report(float(fit), step=step)
+
+
 class OptunaPruningSingleFitCallback(Callback):
     """Callback to prune unpromising trials during Optuna hyperparameters tuning.
 

examples/coloring/coloring_cpspat_solver_example.py

@@ -31,10 +31,16 @@
 
 def run_cpsat_coloring():
     logging.basicConfig(level=logging.INFO)
-    file = [f for f in get_data_available() if "gc_100_7" in f][0]
+    file = [f for f in get_data_available() if "gc_250_3" in f][0]
     color_problem = parse_file(file)
     solver = ColoringCPSatSolver(color_problem, params_objective_function=None)
-    solver.init_model(modeling=ModelingCPSat.BINARY, warmstart=True)
+    solver.init_model(
+        modeling=ModelingCPSat.INTEGER,
+        warmstart=False,
+        value_sequence_chain=False,
+        used_variable=True,
+        symmetry_on_used=True,
+    )
     p = ParametersCP.default_cpsat()
     p.time_limit = 100
     logging.info("Starting solve")

examples/coloring/optuna_full_example_coloring.py

@@ -28,6 +28,7 @@
 )
 from discrete_optimization.generic_tools.callbacks.optuna import (
     OptunaPruningSingleFitCallback,
+    OptunaReportSingleFitCallback,
 )
 from discrete_optimization.generic_tools.cp_tools import ParametersCP
 from discrete_optimization.generic_tools.do_problem import ModeOptim
@@ -40,7 +41,7 @@
 seed = 42
 optuna_nb_trials = 150
 
-study_name = f"coloring_cpsat-auto-250---"
+study_name = f"coloring_cpsat-auto-250-7"
 storage_path = "./optuna-journal.log"  # NFS path for distributed optimization
 
 # Solvers to test and their associated kwargs
@@ -61,7 +62,7 @@
 }
 
 # problem definition
-file = [f for f in get_data_available() if "gc_250_5" in f][0]
+file = [f for f in get_data_available() if "gc_250_7" in f][0]
 problem = parse_file(file)
 
 # sense of optimization
@@ -123,7 +124,8 @@ def objective(trial: Trial):
     # solve
     sol, fit = solver.solve(
         callbacks=[
-            OptunaPruningSingleFitCallback(trial=trial, **kwargs),
+            OptunaReportSingleFitCallback(trial=trial, **kwargs)
+            # OptunaPruningSingleFitCallback(trial=trial, **kwargs),
         ],
         **kwargs,
     ).get_best_solution_fit()