Pymoo Python: TypeError: _evaluate() got an unexpected keyword argument 'algorithm'

Question

I am trying to set up my optimization with Python's Pymoo library, I am using their 'getting started' guide but passing my own independent variables and also not using constraints. I get the same using the example functions from the guide (I have commented them out in the code below).

Here is the code:

class MyProblem(Problem): 
    
    def __init__(self,total,G,t):                      
        super().__init__(n_var = 3,  # 2 in the case of the example from guide
                         n_obj = 2, 
                         n_constr = 0, 
                         #xl = np.array([-1.0,0.0]),    # for example from guide
                         #xu = np.array([1.0, 10.0]),
                         xl = np.array([-1.0,0.0, -1.0]), 
                         xu = np.array([1.0, 10.0, 1.0]),
                         elementwise_evaluation = True)
        self.total = total,         # my own independent variables 
        self.G = G,
        self.t = t
    
    def _evaluate(self, x, out):   
        f1 = 1/3*self.total*(1+2*((x[0]-x[2])*np.exp(-self.t/x[1]) + x[2]))
        f2 = 1/3*self.total*self.G*(1-((x[0]-x[2])*np.exp(-self.t/x[1]) + x[2]))
        #f1 = x[0]**2 + x[1]**2         # example from guide
        #f2 = (x[0]-1)**2 + x[1]**2
        
        out["F"] = np.column_stack([f1, f2])
        
elementwise_problem = MyProblem(total,G,t)

problem = elementwise_problem

algorithm = NSGA2(pop_size = 100,
                  n_offspring = 10, 
                  sampling = get_sampling('real_random'),
                  crossover = get_crossover('real_sbx', prob = 0.9, eta = 15),
                  mutation = get_mutation('real_pm',eta = 20),
                  eliminate_duplicates = True)

termination = get_termination("n_gen", 40)

# method 1
results = minimize(problem,
                   algorithm,
                   termination,
                   seed = 1, 
                   save_history = True,
                   verbose = True)

# method 2
obj = copy.deepcopy(algorithm)

obj.setup(problem, termination = termination, seed = 1)

# until the termination criterion has not been met
while obj.has_next():
    # perform an iteration of the algorithm
    obj.next()
    # access the algorithm to print some intermediate outputs
    print(f"gen: {obj.n_gen} n_nds: {len(obj.opt)} constr: {obj.opt.get('CV').min()} ideal: {obj.opt.get('F').min(axis=0)}")
    

result = obj.result()

When I print out the kwargs in the _evaluate_elementwise method in the Problem class, indeed I get it is the algorithm object:

{'algorithm': <pymoo.algorithms.nsga2.NSGA2 object at 0x00000212D12413C8>}

I struggle to see how it might be taking the algorithm object as an argument to _evalute, which accepts (_x,_out,*args,**kwargs). If anyone is more familiar with this package I'd appreciate the help tremendously !

Here is the full trackeback:

Keyword args: {'algorithm': <pymoo.algorithms.nsga2.NSGA2 object at 0x00000212D12413C8>} Traceback (most recent call last):

File "", line 6, in verbose = True)

File "C:\Users\anaconda3\lib\site-packages\pymoo\optimize.py", line 85, in minimize res = algorithm.solve()

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\algorithm.py", line 226, in solve self._solve(self.problem)

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\algorithm.py", line 321, in _solve self.next()

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\algorithm.py", line 243, in next self.initialize()

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\algorithm.py", line 215, in initialize self._initialize()

File "C:\Users\anaconda3\lib\site-packages\pymoo\algorithms\genetic_algorithm.py", line 81, in _initialize self.evaluator.eval(self.problem, pop, algorithm=self)

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\evaluator.py", line 78, in eval self._eval(problem, pop[I], **kwargs)

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\evaluator.py", line 97, in _eval **kwargs)

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\problem.py", line 284, in evaluate out = self._evaluate_elementwise(X, calc_gradient, out, *args, **kwargs)

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\problem.py", line 413, in _evaluate_elementwise [ret.append(func(x)) for x in X]

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\problem.py", line 413, in [ret.append(func(x)) for x in X]

File "C:\Users\anaconda3\lib\site-packages\pymoo\model\problem.py", line 400, in func self._evaluate(_x, _out, *args, **kwargs)

TypeError: _evaluate() got an unexpected keyword argument 'algorithm'

Answer 1

It seems like your error occurs because of the lack of *args, **kwargs in _evaluate function. I edited your code a bit, you may check it:

class MyProblem(Problem):
    
    total = 5.0        # my own independent variables 
    G = 6.0
    t = 7.0
        
    def __init__(self):                      
        super().__init__(n_var = 3,  # 2 in the case of the example from guide
                         n_obj = 2, 
                         n_constr = 0, 
                         #xl = np.array([-1.0,0.0]),    # for example from guide
                         #xu = np.array([1.0, 10.0]),
                         xl = np.array([-1.0,0.0, -1.0]), 
                         xu = np.array([1.0, 10.0, 1.0]),
                         elementwise_evaluation = True)

    
    def _evaluate(self, x, out, *args, **kwargs): # added *args, **kwargs   
        f1 = 1/3*self.total*(1+2*((x[0]-x[2])*np.exp(-self.t/x[1]) + x[2]))
        f2 = 1/3*self.total*self.G*(1-((x[0]-x[2])*np.exp(-self.t/x[1]) + x[2]))
        #f1 = x[0]**2 + x[1]**2         # example from guide
        #f2 = (x[0]-1)**2 + x[1]**2
        
        out["F"] = np.column_stack([f1, f2])
        
elementwise_problem = MyProblem()

#problem = elementwise_problem

algorithm = NSGA2(pop_size = 100,
                  n_offspring = 10, 
                  sampling = get_sampling('real_random'),
                  crossover = get_crossover('real_sbx', prob = 0.9, eta = 15),
                  mutation = get_mutation('real_pm',eta = 20),
                  eliminate_duplicates = True)

termination = get_termination("n_gen", 40)

# method 1
results = minimize(elementwise_problem,
                   algorithm,
                   termination,
                   seed = 1, 
                   save_history = True,
                   verbose = True)