def __init__(self, of, maxeval, E, alfa0, N_general, N_local, correction):
"""
Initialization
:param of: any objective function to be optimized
:param maxeval: maximum allowed number of evaluations
:param E: maximum number of epochs, i.e. maximum number of the random vector generation
:param alfa0: the first dx is generated from range [-alfa0, alfa0]
:param N_general: alfa doesn't change for N_general steps during global search
:param N_local: alfa doesn't change for N_local steps during local search
:param correction: correction for x values
"""
Heuristic.__init__(self, of, maxeval)
self.k = 0 # alfa counter - it is in the article, but I don't use it
self.epoch = 0 # epoch counter
self.E = E
self.N_general = N_general
self.GSE = self.E - 2 * self.N_general # general search stop criterion,
# number of epochs during general search
self.N_local = N_local
self.alfa = [alfa0] # we will save all alpha values
self.x_curr = None # current value of x
self.alfa_best = None # alpha connected with the best solution
self.y_gen = np.zeros(
self.E + 1, dtype=float
) # we save all y values - it is needed for alpha calculation
# +1 because of the initial value
self.correction = correction
def __init__(self, of, maxeval, N, CR, F):
Heuristic.__init__(self, of, maxeval)
assert N >= 4, 'N should be at least equal to 4'
self.N = N
self.CR = CR
assert 0 <= F <= 2, 'F should be from [0; 2]'
self.F = F
def __init__(self, of, maxeval, N, CR, F):
Heuristic.__init__(self, of, maxeval)
assert N >= 4, 'N should be at least equal to 4'
self.N = N # Population size
self.n = np.size(of.a)
self.CR = CR # Crossover probability
assert 0 <= F <= 2, 'F should be from [0; 2]'
self.F = F # Differential weight
self.name = 'DE'
def __init__(self, of, maxeval, N, M, Tsel1, Tsel2, mutation, crossover):
Heuristic.__init__(self, of, maxeval)
assert M > N, 'M should be larger than N'
self.N = N # population size
self.M = M # working population size
self.Tsel1 = Tsel1 # first selection temperature
self.Tsel2 = Tsel2 # second selection temperature
self.mutation = mutation
self.crossover = crossover
def __init__(self, of, maxeval, hmax=np.inf, random_descent=False):
"""
Initialization
:param of: any objective function to be optimized
:param maxeval: maximum allowed number of evaluations
:param hmax: maximum number of local improvements (0 = Random Shooting)
:param random_descent: turns on random descent, instead of the steepest one (default)
"""
Heuristic.__init__(self, of, maxeval)
self.hmax = hmax
self.random_descent = random_descent
def __init__(self, of, maxevalFsa, maxevalGo, N, M, Tsel1, Tsel2, mutation,
crossover, T0, n0, alpha):
Heuristic.__init__(self, of, maxevalGo)
self.N = N
self.M = M
self.Tsel1 = Tsel1
self.Tsel2 = Tsel2
self.mutation = mutation
self.crossover = crossover
self.T0 = T0
self.n0 = n0
self.alpha = alpha
self.geneLenth = np.size(self.of.a)
self.maxevalFsa = maxevalFsa
self.maxevalGo = maxevalGo
def __init__(self, of, maxeval, T0, n0, alpha, mutation):
"""
Initialization
:param of: any objective function to be optimized
:param maxeval: maximum allowed number of evaluations
:param T0: initial temperature
:param n0: cooling strategy parameter - number of steps
:param alpha: cooling strategy parameter - exponent
:param mutation: mutation to be used for the specific objective function (see heur_aux.py)
"""
Heuristic.__init__(self, of, maxeval)
self.T0 = T0
self.n0 = n0
self.alpha = alpha
self.mutation = mutation
def __init__(self, of, maxeval, N):
Heuristic.__init__(self, of, maxeval)
# assert M > N, 'M should be larger than N'
self.N = N # population size
self.P = int(N/2) # parents' size
def __init__(self, of, maxeval, N):
Heuristic.__init__(self, of, maxeval)
self.N = N # population size
self.P = int(N / 2) # parents' size
