Source code for neorl.evolu.jaya

#    This file is part of NEORL.

#    Copyright (c) 2021 Exelon Corporation and MIT Nuclear Science and Engineering
#    NEORL is free software: you can redistribute it and/or modify
#    it under the terms of the MIT LICENSE

#    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
#    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
#    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
#    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
#    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
#    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
#    SOFTWARE.

# -*- coding: utf-8 -*-
#"""
#@author: Xubo
#@email: guxubo@alumni.sjtu.edu
#"""
import random
import numpy as np
import joblib
from neorl.evolu.discrete import mutate_discrete, encode_grid_to_discrete 
from neorl.evolu.discrete import decode_discrete_to_grid, encode_grid_indv_to_discrete
from neorl.utils.seeding import set_neorl_seed
from neorl.utils.tools import get_population, check_mixed_individual

[docs]class JAYA: """ JAYA algorithm :param mode: (str) problem type, either "min" for minimization problem or "max" for maximization :param bounds: (dict) input parameter type and lower/upper bounds in dictionary form. Example: ``bounds={'x1': ['int', 1, 4], 'x2': ['float', 0.1, 0.8], 'x3': ['float', 2.2, 6.2]}`` :param fit: (function) the fitness function :param npop: (int) number of individuals in the population :param int_transform: (str): method of handling int/discrete variables, choose from: ``nearest_int``, ``sigmoid``, ``minmax``. :param ncores: (int) number of parallel processors :param seed: (int) random seed for sampling """ def __init__(self, mode, bounds, fit, npop=50, int_transform ='nearest_int', ncores=1, seed=None): self.seed=seed set_neorl_seed(self.seed) assert npop > 3, '--eror: size of npop must be more than 3' self.npop= npop self.bounds=bounds self.ncores=ncores self.int_transform=int_transform if int_transform not in ["nearest_int", "sigmoid", "minmax"]: raise ValueError('--error: int_transform entered by user is invalid, must be `nearest_int`, `sigmoid`, or `minmax`') self.mode=mode if mode == 'max': self.fit=fit elif mode == 'min': def fitness_wrapper(*args, **kwargs): return -fit(*args, **kwargs) self.fit = fitness_wrapper else: raise ValueError('--error: The mode entered by user is invalid, use either `min` or `max`') #infer variable types self.var_type = np.array([bounds[item][0] for item in bounds]) #mir-grid if "grid" in self.var_type: self.grid_flag=True self.orig_bounds=bounds #keep original bounds for decoding #print('--debug: grid parameter type is found in the space') self.bounds, self.bounds_map=encode_grid_to_discrete(self.bounds) #encoding grid to int #define var_types again by converting grid to int self.var_type = np.array([self.bounds[item][0] for item in self.bounds]) else: self.grid_flag=False self.bounds = bounds self.orig_bounds=bounds self.dim = len(bounds) self.lb=[self.bounds[item][1] for item in self.bounds] self.ub=[self.bounds[item][2] for item in self.bounds] def gen_indv(self, bounds): # individual indv = [] for key in bounds: if bounds[key][0] == 'int': indv.append(random.randint(bounds[key][1], bounds[key][2])) elif bounds[key][0] == 'float': indv.append(random.uniform(bounds[key][1], bounds[key][2])) elif bounds[key][0] == 'grid': indv.append(random.sample(bounds[key][1],1)[0]) return indv def init_population(self, x0=None, verbose=False): # population pop = [] if x0: # have premary solution if verbose: print('The first individual provided by the user:', x0[0]) print('The last individual provided by the user:', x0[-1]) for i in range(self.npop): check_mixed_individual(x=x0[i], bounds=self.orig_bounds) #assert the type provided is consistent if self.grid_flag: pop.append(encode_grid_indv_to_discrete(x0[i], bounds=self.orig_bounds, bounds_map=self.bounds_map)) else: pop.append(x0[i]) else: # random init for i in range(self.npop): indv=self.gen_indv(self.bounds) pop.append(indv) # array pop = np.array(pop) return pop def ensure_bounds(self, vec): # bounds check vec_new = [] for i, (key, val) in enumerate(self.bounds.items()): # less than minimum if vec[i] < self.bounds[key][1]: vec_new.append(self.bounds[key][1]) # more than maximum if vec[i] > self.bounds[key][2]: vec_new.append(self.bounds[key][2]) # fine if self.bounds[key][1] <= vec[i] <= self.bounds[key][2]: vec_new.append(vec[i]) return vec_new def eval_pop(self, pos_array): #""" #Evaluate fitness of the population with parallel processing. #Return: #list - pop fitnesses #""" if self.ncores > 1: with joblib.Parallel(n_jobs=self.ncores) as parallel: fitness_lst = parallel(joblib.delayed(self.fit_worker)(pos_array[i, :]) for i in range(self.npop)) else: fitness_lst = [] for i in range(self.npop): fitness_lst.append(self.fit_worker(pos_array[i, :])) return fitness_lst def fit_worker(self, x): # Clip the wolf with position outside the lower/upper bounds and return same position # x=self.ensure_bounds(x) if self.grid_flag: #decode the individual back to the int/float/grid mixed space x=decode_discrete_to_grid(x,self.orig_bounds,self.bounds_map) fitness = self.fit(x) return fitness def ensure_discrete(self, vec): #""" #to mutate a vector if discrete variables exist #Params: #vec - position in vector/list form #Return: #vec - updated position vector with discrete values #""" for dim in range(self.dim): if self.var_type[dim] == 'int': vec[dim] = mutate_discrete(x_ij=vec[dim], x_min=min(vec), x_max=max(vec), lb=self.lb[dim], ub=self.ub[dim], alpha=self.b, method=self.int_transform, ) return vec
[docs] def evolute(self, ngen, x0=None, verbose=False): """ This function evolutes the MFO algorithm for number of generations. :param ngen: (int) number of generations to evolute :param x0: (list of lists) the initial individuals of the population :param verbose: (bool) print statistics to screen :return: (tuple) (best individual, best fitness, and a list of fitness history) """ self.history = {'local_fitness':[], 'global_fitness':[]} N = self.npop # population size dim = len(self.bounds) # individual length set_neorl_seed(self.seed) fitness_mat = np.zeros(N) Best_pos = np.zeros(dim) Best_score = float('-inf') # find a maximum, so the larger the better Worst_pos = np.zeros(dim) Worst_score = float('inf') ## INITIALIZE # population if x0: assert len(x0) == N, '--error: the length of x0 ({}) (initial population) must equal to number of individuals npop ({})'.format(len(x0), self.npop) pos = self.init_population(x0=x0, verbose=verbose) else: pos = self.init_population(verbose=verbose) pos = pos*1.0 #this is to account for mixed intger-cont. problems, data needs to be float # calulate fitness fitness_mat=self.eval_pop(pos) for i in range(N): if fitness_mat[i] > Best_score: Best_score = fitness_mat[i] Best_pos = pos[i, :] if fitness_mat[i] < Worst_score: Worst_score = fitness_mat[i] Worst_pos = pos[i, :] ## main loop best_scores = [] for gen in range(1, ngen+1): self.b= 1 - gen * ((1) / ngen) #mir: b decreases linearly between 1 to 0, for discrete mutation new_pos = np.zeros((N,dim)) # update pos for i in range(N): r1=np.random.random(dim) r2=np.random.random(dim) # Update pos new_pos[i,:] = ( pos[i,:] + r1*(Best_pos - abs(pos[i,:])) - r2*(Worst_pos - abs(pos[i,:])) # !! minus ) # check bounds new_pos[i,:] = self.ensure_bounds(new_pos[i,:]) new_pos[i,:] = self.ensure_discrete(new_pos[i,:]) fitness_new=self.eval_pop(new_pos) for i in range(N): if fitness_new[i] > fitness_mat[i]: pos[i,:] = new_pos[i,:] fitness_mat[i] = fitness_new[i] # update best_score and worst_score for i in range(N): if fitness_mat[i] > Best_score: Best_score = fitness_mat[i] Best_pos = pos[i, :] if fitness_mat[i] < Worst_score: Worst_score = fitness_mat[i] Worst_pos = pos[i, :] #----------------------------- #Fitness saving #----------------------------- self.last_pop=new_pos.copy() self.last_fit=np.array(fitness_new).copy() gen_avg = sum(fitness_mat) / N # current generation avg. fitness y_best = Best_score # fitness of best individual x_best = Best_pos.copy() best_scores.append(y_best) #--mir show the value wrt min/max if self.mode=='min': y_best_correct=-y_best gen_avg=-gen_avg self.history['local_fitness'].append(-np.max(fitness_new)) else: y_best_correct=y_best self.history['local_fitness'].append(np.max(fitness_new)) if verbose: print('************************************************************') print('JAYA step {}/{}, Ncores={}'.format(gen*self.npop, ngen*self.npop, self.ncores)) print('************************************************************') print('Best fitness:', np.round(y_best_correct,6)) if self.grid_flag: x_decoded = decode_discrete_to_grid(x_best, self.orig_bounds, self.bounds_map) print('Best individual:', x_decoded) else: print('Best individual:', x_best) print('Average fitness:', np.round(gen_avg,6)) print('************************************************************') #mir-grid if self.grid_flag: x_best_correct = decode_discrete_to_grid(x_best, self.orig_bounds, self.bounds_map) else: x_best_correct = x_best if verbose: print('------------------------ JAYA Summary --------------------------') print('Best fitness (y) found:', y_best_correct) print('Best individual (x) found:', x_best_correct) print('--------------------------------------------------------------') if self.mode=='min': best_scores=[-item for item in best_scores] self.last_fit=-self.last_fit self.history['global_fitness'] = best_scores #--mir return the last population for restart calculations if self.grid_flag: self.history['last_pop'] = get_population(self.last_pop, fits=self.last_fit, grid_flag=True, bounds=self.orig_bounds, bounds_map=self.bounds_map) else: self.history['last_pop'] = get_population(self.last_pop, fits=self.last_fit, grid_flag=False) return x_best_correct, y_best_correct, self.history