Update to pymoo 0.6.1 (#25)

sb_arch_opt/algo/arch_sbo/infill.py

@@ -38,7 +38,14 @@
 from pymoo.core.algorithm import filter_optimum
 from pymoo.util.normalization import Normalization
 from pymoo.util.nds.non_dominated_sorting import NonDominatedSorting
-from pymoo.algorithms.moo.nsga2 import RankAndCrowdingSurvival, calc_crowding_distance
+from pymoo.algorithms.moo.nsga2 import RankAndCrowdingSurvival
+
+try:
+    # pymoo < 0.6.1
+    from pymoo.algorithms.moo.nsga2 import calc_crowding_distance
+except ImportError:
+    # pymoo >= 0.6.1
+    from pymoo.operators.survival.rank_and_crowding.metrics import calc_crowding_distance
 
 __all__ = ['SurrogateInfill', 'FunctionEstimateInfill', 'ConstrainedInfill', 'FunctionEstimateConstrainedInfill',
            'ExpectedImprovementInfill', 'MinVariancePFInfill', 'ConstraintStrategy', 'MeanConstraintPrediction',

sb_arch_opt/algo/pymoo_interface/md_mating.py

@@ -116,13 +116,14 @@ def _do(self, problem, pop, n_offsprings, parents=False, **kwargs):
             crossover = self.crossover[clazz]
             assert crossover.n_parents == n_parents_crossover and crossover.n_offsprings == n_offspring_crossover
 
-            _parents = [[Individual(X=parent.X[x_idx]) for parent in parents] for parents in pop]
+            _parents = [Population.new(X=[parent.X[x_idx].astype(float) for parent in parents]) for parents in pop]
+            # _parents = [[Individual(X=parent.X[x_idx]) for parent in parents] for parents in pop]
 
-            _vars = [var_defs[e] for e in list_of_vars]
+            _vars = {i: var_defs[e] for i, e in enumerate(list_of_vars)}
             _xl, _xu = None, None
 
             if clazz in [Real, Integer]:
-                _xl, _xu = np.array([v.bounds for v in _vars]).T
+                _xl, _xu = np.array([v.bounds for v in _vars.values()]).T
 
             _problem = Problem(vars=_vars, xl=_xl, xu=_xu)
 
@@ -133,7 +134,7 @@ def _do(self, problem, pop, n_offsprings, parents=False, **kwargs):
                 _off = mutation(_problem, _off, **kwargs)
 
                 # Sometimes NaN's might sneak into the outputs, try again if this is the case
-                x_off = _off.get('X')[:n_offsprings, :]
+                x_off = _off.get('X')[:n_offsprings, :].astype(float)
                 if np.any(np.isnan(x_off)):
                     continue
                 break

sb_arch_opt/pareto_front.py

@@ -37,11 +37,17 @@
 from pymoo.core.evaluator import Evaluator
 from pymoo.visualization.scatter import Scatter
 from pymoo.core.initialization import Initialization
-from pymoo.algorithms.moo.nsga2 import calc_crowding_distance
 from pymoo.util.nds.non_dominated_sorting import NonDominatedSorting
 from pymoo.termination.default import DefaultMultiObjectiveTermination, DefaultSingleObjectiveTermination
 from sb_arch_opt.sampling import HierarchicalExhaustiveSampling, HierarchicalSampling
 
+try:
+    # pymoo < 0.6.1
+    from pymoo.algorithms.moo.nsga2 import calc_crowding_distance
+except ImportError:
+    # pymoo >= 0.6.1
+    from pymoo.operators.survival.rank_and_crowding.metrics import calc_crowding_distance
+
 __all__ = ['CachedParetoFrontMixin']
 
 

setup.py

@@ -52,7 +52,7 @@ def _get_readme():
         license='MIT',
         install_requires=[
             'numpy',
-            'pymoo~=0.6.0.1',
+            'pymoo~=0.6.1',
             'scipy',
             'deprecated',
             'pandas',