import random #--------粒 子 群 最 適 化------------------------------ #評価関数: z = 1000*x+2000*y def criterion(x, y): stock_c=w-0.75*(x)-2/7*(y) stock_d=h-0.25*(x)-5/7*(y) z=1000*x+2000*y if stock_c<0 or stock_d<0 or x<0 or y<0: z=-z return z #粒子の位置の更新を行う関数 def update_position(x, y, vx, vy): stock_c=w-0.75*(x+vx)-2/7*(y+vy) stock_d=h-0.25*(x+vx)-5/7*(y+vy) if stock_c<0 or stock_d<0 or x+vx<0 or y+vy<0: new_x = x new_y = y else: new_x = x + vx new_y = y + vy return new_x, new_y #粒子の速度の更新を行う関数 def update_velocity(x, y, vx, vy, p, g, w=0.5, ro_max=1, c1=2.05, c2=2.05): #パラメーターroはランダムに与える ro1 = random.uniform(0, ro_max) ro2 = random.uniform(0, ro_max) phi=c1+c2 K=2/abs(2-phi-(phi*phi-4*phi)**0.5) #粒子速度の更新を行う new_vx =K * ( w * vx + c1 * ro1 * (p[0] - x) + c2 * ro2 * (g[0] - x) ) new_vy =K * ( w * vy + c1 * ro1 * (p[1] - y) + c2 * ro2 * (g[1] - y) ) # new_vx = w * vx + ro1 * (p[0] - x) + ro2 * (g[0] - x) # new_vy = w * vy + ro1 * (p[1] - y) + ro2 * (g[1] - y) return new_vx, new_vy def main(): N = 200 #粒子の数 x_min, x_max = 0, 2*w y_min, y_max = 0, 2*h #粒子位置, 速度, パーソナルベスト, グローバルベストの初期化を行う ps = [[random.uniform(x_min, 0.7*w), random.uniform(y_min, 0.7*h)] for i in range(N)] vs = [[0.0, 0.0] for i in range(N)] personal_best_positions = list(ps) personal_best_scores = [criterion(p[0], p[1]) for p in ps] best_particle = personal_best_scores.index(max(personal_best_scores)) global_best_position = personal_best_positions[best_particle] T = 100 #世代数(ループの回数) for t in range(T): for n in range(N): x, y = ps[n][0], ps[n][1] vx, vy = vs[n][0], vs[n][1] p = personal_best_positions[n] #粒子の位置の更新を行う new_x, new_y = update_position(x, y, vx, vy) ps[n] = [new_x, new_y] #粒子の速度の更新を行う new_vx, new_vy = update_velocity(new_x, new_y, vx, vy, p, global_best_position) vs[n] = [new_vx, new_vy] #評価値を求め, パーソナルベストの更新を行う score = criterion(new_x, new_y) if score > personal_best_scores[n]: personal_best_scores[n] = score personal_best_positions[n] = [new_x, new_y] #グローバルベストの更新を行う # f.write(str(max(personal_best_scores))+"\n") best_particle = personal_best_scores.index(max(personal_best_scores)) global_best_position = personal_best_positions[best_particle] #最適解 # print(max(personal_best_scores))#, global_best_position) return max(personal_best_scores) #-------------------------------------------------------------- w, h=[int(i) for i in input().split()] best=0 for i in range(1): # with open('result.txt', 'w') as f: best=max(best, main()) # print(best) print(best)