Include NSGAii in SPOTPY infrastructur, fix minor padds error

.gitignore

@@ -8,7 +8,6 @@ __pycache__/
 
 #other files
 *.out
-
 *.png
 *.csv
 site/*

README.md

@@ -64,6 +64,7 @@ Some features you can use with the SPOTPY package are:
   * Fitness Scaled Chaotic Artificial Bee Colony (`FSCABC`)
   * Dynamically Dimensioned Search algorithm (`DDS`)
   * Pareto Archived - Dynamicallly Dimensioned Search algorithm (`PA-DDS`)
+  * Fast and Elitist Multiobjective Genetic Algorithm (`NSGA-II`)
 
 * Wide range of objective functions (also known as loss function, fitness function or energy function) to validate the sampled results. Available functions are
 
@@ -136,15 +137,23 @@ Some features you can use with the SPOTPY package are:
 Install
 =================
 
-Installing SPOTPY is easy. Just use:
+Classical Python options exist to install SPOTPY:
+
+From PyPi:
 
 	pip install spotpy
 
-Or, after downloading the [source code](https://pypi.python.org/pypi/spotpy "source code") and making sure python is in your OS path:
+From Conda-Forge:
+
+	conda config --add channels conda-forge
+	conda config --set channel_priority strict
+	conda install spotpy
+
+From [Source](https://pypi.python.org/pypi/spotpy):
 
 	python setup.py install
 
-	
+
 Support
 =================
 

README.rst

@@ -39,6 +39,7 @@ We want to make this task as easy as possible. Some features you can use with th
   * Fitness Scaled Chaotic Artificial Bee Colony (`FSCABC`)
   * Dynamically Dimensioned Search algorithm (`DDS`)
   * Pareto Archived - Dynamicallly Dimensioned Search algorithm (`PA-DDS`)
+  * Fast and Elitist Multiobjective Genetic Algorithm (`NSGA-II`)
 
 * Wide range of objective functions (also known as loss function, fitness function or energy function) to validate the sampled results. Available functions are
 
@@ -136,15 +137,23 @@ Documentation is available at `<https://spotpy.readthedocs.io/en/latest>`__
 Install
 -------
 
-Installing SPOTPY is easy. Just use:
+Classical Python options exist to install SPOTPY:
+
+From PyPi:
 
 	pip install spotpy
 
-Or, after downloading the source code and making sure python is in your path:
+From Conda-Forge:
+
+	conda config --add channels conda-forge
+	conda config --set channel_priority strict
+	conda install spotpy
+
+From [Source](https://pypi.python.org/pypi/spotpy):
 
 	python setup.py install
 
-	
+
 Papers citing SPOTPY
 --------------------
 See `Google Scholar <https://scholar.google.de/scholar?cites=17155001516727704728&as_sdt=2005&sciodt=0,5&hl=de>`__ for a continuously updated list.

setup.py

@@ -5,7 +5,7 @@
 
 setup(
   name = 'spotpy',
-  version = '1.5.10',
+  version = '1.5.14',
   description = 'A Statistical Parameter Optimization Tool',
   long_description=open(os.path.join(os.path.dirname(__file__),
                                        "README.rst")).read(),
@@ -15,7 +15,7 @@
   license = 'MIT',
   packages=find_packages(exclude=["tests*", "docs*"]),
   use_2to3 = True,
-  keywords = 'Monte Carlo, MCMC, MLE, SCE-UA, Simulated Annealing, DE-MCz, DREAM, ROPE, Artifical Bee Colony, DDS, PA-DDS, Uncertainty, Calibration, Model, Signatures',
+  keywords = 'Monte Carlo, MCMC, MLE, SCE-UA, Simulated Annealing, DE-MCz, DREAM, ROPE, Artifical Bee Colony, DDS, PA-DDS, NSGAii, Uncertainty, Calibration, Model, Signatures',
   classifiers = [
         'Development Status :: 4 - Beta',
         'Intended Audience :: Developers',

spotpy/__init__.py

@@ -14,6 +14,7 @@
 sampling and an analyser class for the plotting of results by the sampling.
 
 :dependencies: - Numpy >1.8 (http://www.numpy.org/) 
+               - Scipy >1.5 (https://pypi.org/project/scipy/)
                - Pandas >0.13 (optional) (http://pandas.pydata.org/)
                - Matplotlib >1.4 (optional) (http://matplotlib.org/) 
                - CMF (optional) (http://fb09-pasig.umwelt.uni-giessen.de:8000/)
@@ -40,4 +41,4 @@
 from . import describe            # Contains some helper functions to describe samplers and set-ups
 from .hydrology import signatures # Quantifies goodness of fit between simulation and evaluation data with hydrological signatures
 
-__version__ = '1.5.10'
+__version__ = '1.5.14'

spotpy/algorithms/nsgaii.py

@@ -260,7 +260,7 @@ def crowdDist2(self,x):
 
 
 
-    def sample(self, generations, n_obj, n_pop = None, skip_duplicates = False, 
+    def sample(self, generations, n_obj, n_pop = None, 
                selection = TournamentSelection(pressure=2), 
                crossover = Crossover(crossProb=0.9),
                mutation = PolynomialMutation(prob_mut=0.25,eta_mut=30)):
@@ -273,7 +273,7 @@ def sample(self, generations, n_obj, n_pop = None, skip_duplicates = False,
         self.n_pop = n_pop
         self.generations= generations
         self.set_repetiton(self.generations*self.n_pop)
-        self.skip_duplicates = skip_duplicates
+        self.skip_duplicates = True # False does not work yet
 
         Pt = np.vstack([self.parameter()['random'] for i in range(self.n_pop)])
 

spotpy/algorithms/padds.py

@@ -134,7 +134,7 @@ def get_next_x_curr(self):
             else:  # otherwise use the last generated solution
                 self.best_value.parameters, self.best_value.best_obj_val = (self.parameter_current, self.obj_func_current)
 
-            # This line is needed to get an array of data converted into a parameter object
+            # # This line is needed to get an array of data converted into a parameter object
             self.best_value.fix_format()
 
             yield rep, self.calculate_next_s_test(self.best_value.parameters, rep, self.generator_repetitions, self.r)
@@ -163,9 +163,9 @@ def calculate_initial_parameterset(self, repetitions, initial_objs, initial_para
             for i in range(initial_params.shape[0]):
                 self.parameter_current = initial_params[i]
                 if len(initial_objs[i]) > 0:
-                    self.obj_func_current = initial_objs[i]
+                    self.obj_func_current = np.array(initial_objs[i])
                 else:
-                    self.obj_func_current = self.getfitness(simulation=[], params=self.parameter_current)
+                    self.obj_func_current = np.array(self.getfitness(simulation=[], params=self.parameter_current))
 
                 if i == 0:  # Initial value
                     self.pareto_front = np.array([[self.obj_func_current, self.parameter_current]])
@@ -206,7 +206,8 @@ def sample(self, repetitions, trials=1, initial_objs=np.array([]), initial_param
             self.generator_repetitions = repetions_left
 
             for rep, x_curr, simulations in self.repeat(self.get_next_x_curr()):
-                self.obj_func_current = self.postprocessing(rep, x_curr, simulations)
+                obj_func_current = self.postprocessing(rep, x_curr, simulations)
+                self.obj_func_current = np.array(obj_func_current)
                 num_imp = np.sum(self.obj_func_current <= self.best_value.best_obj_val)
                 num_deg = np.sum(self.obj_func_current > self.best_value.best_obj_val)
 
@@ -335,6 +336,7 @@ def nd_check(nd_set_input, objective_values, parameter_set):
     dominance_flag = 0
 
     # These are simply reshaping problems if we want to loop over arrays but we have a single float given
+    objective_values = np.array(objective_values)
     try:
         like_struct_len = objective_values.shape[0]
     except IndexError:

spotpy/examples/tutorial_nsgaii.py

@@ -1,6 +1,14 @@
 #!/usr/bin/env python
 # coding: utf-8
+# -*- coding: utf-8 -*-
+'''
+Copyright 2015 by Tobias Houska
+This file is part of Statistical Parameter Estimation Tool (SPOTPY).
 
+:author: Tobias Houska
+
+This class holds example code how to use the nsgaii algorithm
+'''
 
 from __future__ import division, print_function
 from __future__ import absolute_import
@@ -24,7 +32,22 @@
 
 
 def multi_obj_func(evaluation, simulation):
-    #used to overwrite objective function in hymod example
+    """
+    This function is used to overwrite objective function in hymod example
+    
+    Parameters
+    ----------
+    evaluation : array
+        The observation data.
+    simulation : array
+        The model input
+
+    Returns
+    -------
+    list
+        Three different kinds of objective functions.
+
+    """
     like1 = abs(spotpy.objectivefunctions.pbias(evaluation, simulation))
     like2 = spotpy.objectivefunctions.rmse(evaluation, simulation)
     like3 = spotpy.objectivefunctions.rsquared(evaluation, simulation)*-1
@@ -40,61 +63,7 @@ def multi_obj_func(evaluation, simulation):
     sp_setup=spot_setup(obj_func= multi_obj_func)
     sampler = spotpy.algorithms.NSGAII(spot_setup=sp_setup, dbname='NSGA2', dbformat="csv")
 
-    sampler.sample(generations, n_obj= 3, n_pop = n_pop, skip_duplicates = skip_duplicates)
-    #sampler.sample(generations=5, paramsamp=40)
-
-
-#    # user config
-#    
-#    n_var = 5
-#    
-#    
-#    last = None
-#    first = None
-#    
-#    # output calibration 
-#    
-#    df = pd.read_csv("NSGA2.csv")
-#    
-#    if last:
-#        df = df.iloc[-last:,:]
-#    elif first:
-#        df = df.iloc[:first,:]
-#    else:
-#        pass
-#    
-#    
-#    
-#    print(len(df))
-#    # plot objective functions
-#    fig = plt.figure()
-#    for i,name in enumerate(df.columns[:n_obj]):
-#        ax = fig.add_subplot(n_obj,1,i +1)
-#        df.loc[::5,name].plot(lw=0.5,figsize=(18,8),ax = ax,color="black")
-#        plt.title(name)
-#    plt.show()
-#    
-#    last = 100
-#    first = None
-#    
-#    x,y,z = df.iloc[-last:,0],df.iloc[-last:,1],df.iloc[-last:,2]
-#    fig = plt.figure()
-#    ax = fig.add_subplot(111, projection='3d')
-#    ax.scatter(x,y,z,marker="o")
-#    ax.set_xlabel("f0")
-#    ax.set_ylabel("f1")
-#    ax.set_zlabel("f2")
-#    plt.show()
-#    
-#    # plot parameters
-#    fig = plt.figure()
-#    for i,name in enumerate(df.columns[n_obj:8]):
-#        ax = fig.add_subplot(5,1,i +1)
-#        df.loc[:,name].plot(lw=0.5,figsize=(18,8),ax = ax,color="black")
-#        plt.title(name)
-#    plt.show()
-#    
-
+    sampler.sample(generations, n_obj= 3, n_pop = n_pop)
 
 
     results = sampler.getdata()

spotpy/examples/tutorial_padds.py

@@ -78,4 +78,4 @@ def objectivefunction(self, simulation, evaluation, params):
 spot_setup = padds_spot_setup()
 
 sampler = spotpy.algorithms.padds(spot_setup, dbname='padds_hymod', dbformat='csv')
-res = sampler.sample(10000,trials=1)
+res = sampler.sample(2000,trials=1)