pymoo 0.6.1.5.dev0__cp313-cp313-macosx_10_13_universal2.whl

pymoo/vendor/gta.py

@@ -0,0 +1,617 @@
+"""Dynamic Multi-objective Optimization Problems
+This Library is a GTA(a and m) benchmarks from the website of the author,
+and the implementation of the FDA, DIMP, DMOP, HE benchmarks of the CEC2015 competition
+nt: severity of change
+taut: frequency of change
+tauT: maximum number of generation
+tau : current generation
+
+"""
+# !/bin/python
+
+import numpy as np
+from random import randint
+from math import floor, fabs, sin, pi, cos, sqrt
+
+## Parameter configuration ##
+LOWER_BOUND = [0.0] + 20 * [-1.0]
+UPPER_BOUND = 21 * [1.0]
+ERR_MSG = "x is outside decision boundary or dimension of x is not correct"
+DELTA_STATE = 1
+
+
+## Define component functions ##
+def beta_uni(x, t, g, obj_num=2):
+    """This function is used to calculate the unimodal beta function. Input are
+    the decision variable (x), time (t) and g function (g).
+    """
+    beta = [0.0] * obj_num
+    for i in range(obj_num - 1, len(x)):
+        beta[(i + 1) % obj_num] += (x[i] - g(x, t)) * (x[i] - g(x, t))
+
+    beta = [(2.0 / int(len(LOWER_BOUND) / obj_num)) * b for b in beta]
+    return beta
+
+
+def beta_multi(x, t, g, obj_num=2):
+    """This function is used to calculate the multi-modal beta function. Input
+    are the decision variable (x), time (t) and g function (g).
+    """
+    beta = [0.0] * obj_num
+    for i in range(obj_num - 1, len(x)):
+        beta[(i + 1) % obj_num] += (x[i] - g(x, t)) * (x[i] - g(x, t)) * \
+                                   (1 + np.abs(np.sin(4 * np.pi * (x[i] - g(x, t)))))
+
+    beta = [(2.0 / int(len(LOWER_BOUND) / obj_num)) * b for b in beta]
+    return beta
+
+
+def beta_mix(x, t, g, obj_num=2):
+    """This function is used to calculate the mixed unimodal and multi-modal
+    beta function. Input are the decision variable (x), time (t) and g function
+    (g).
+    """
+    beta = [0.0] * obj_num
+    k = int(abs(5.0 * (int(DELTA_STATE * int(t) / 5.0) % 2) - (DELTA_STATE * int(t) % 5)))
+
+    for i in range(obj_num - 1, len(x)):
+        temp = 1.0 + (x[i] - g(x, t)) * (x[i] - g(x, t)) - np.cos(2 * np.pi * k * (x[i] - g(x, t)))
+        beta[(i + 1) % obj_num] += temp
+    beta = [(2.0 / int(len(LOWER_BOUND) / obj_num)) * b for b in beta]
+    return beta
+
+
+def alpha_conv(x):
+    """This function is used to calculate the alpha function with convex POF.
+    Input is decision variable (x).
+    """
+    return [x[0], 1 - np.sqrt(x[0])]
+
+
+def alpha_disc(x):
+    """This function is used to calculate the alpha function with discrete POF.
+    Input is decision variable (x).
+    """
+    return [x[0], 1.5 - np.sqrt(x[0]) - 0.5 * np.sin(10 * np.pi * x[0])]
+
+
+def alpha_mix(x, t):
+    """This function is used to calculate the alpha function with mixed
+    continuous POF and discrete POF.
+    """
+    k = int(abs(5.0 * (int(DELTA_STATE * int(t) / 5.0) % 2) - (DELTA_STATE * int(t) % 5)))
+    return [x[0], 1 - np.sqrt(x[0]) + 0.1 * k * (1 + np.sin(10 * np.pi * x[0]))]
+
+
+def alpha_conf(x, t):
+    """This function is used to calculate the alpha function with time-varying
+    conflicting objective. Input are decision variable (x) and time (t).
+    """
+    k = int(abs(5.0 * (int(DELTA_STATE * int(t) / 5.0) % 2) - (DELTA_STATE * int(t) % 5)))
+    return [x[0], 1 - np.power(x[0], \
+                               np.log(1 - 0.1 * k) / np.log(0.1 * k + np.finfo(float).eps))]
+
+
+def alpha_conf_3obj_type1(x, t):
+    """This function is used to calculate the alpha unction with time-varying
+    conflicting objective (3-objective, type 1). Input are decision variable
+    (x) and time (t).
+    """
+    k = int(abs(5.0 * (int(DELTA_STATE * int(t) / 5.0) % 2) - (DELTA_STATE * int(t) % 5)))
+    alpha1 = fix_numerical_instability(np.cos(0.5 * x[0] * np.pi) * np.cos(0.5 * x[1] * np.pi))
+    alpha2 = fix_numerical_instability(np.cos(0.5 * x[0] * np.pi) * np.sin(0.5 * x[1] * np.pi))
+    alpha3 = fix_numerical_instability(np.sin(0.5 * x[0] * np.pi + 0.25 * (k / 5.0) * np.pi))
+    return [alpha1, alpha2, alpha3]
+
+
+def alpha_conf_3obj_type2(x, t):
+    """This function is used to calculate the alpha unction with time-varying
+    conflicting objective (3-objective, type 2). Input are decision variable (x)
+    and time (t).
+    """
+    k = int(abs(5.0 * (int(DELTA_STATE * int(t) / 5.0) % 2) - (DELTA_STATE * int(t) % 5)))
+    k_ratio = (5.0 - k) / 5.0
+    alpha1 = fix_numerical_instability(np.cos(0.5 * x[0] * np.pi) * np.cos(0.5 * x[1] * np.pi * k_ratio))
+    alpha2 = fix_numerical_instability(np.cos(0.5 * x[0] * np.pi) * np.sin(0.5 * x[1] * np.pi * k_ratio))
+    alpha3 = fix_numerical_instability(np.sin(0.5 * x[0] * np.pi))
+    return [alpha1, alpha2, alpha3]
+
+
+def g(x, t):
+    """This function is used to calculate the g function used in the paper.
+    Input are decision variable (x) and time (t).
+    """
+    return np.sin(0.5 * np.pi * (t - x[0]))
+
+
+## Utility functions ##
+def check_boundary(x, upper_bound=UPPER_BOUND, lower_bound=LOWER_BOUND):
+    """Check the dimension of x and whether it is in the decision boundary. x is
+    decision variable, upper_bound and lower_bound are upperbound and lowerbound
+    lists of the decision space
+    """
+    if len(x) != len(upper_bound) or len(x) != len(lower_bound):
+        return False
+
+    output = True
+    for e, upp, low in zip(x, upper_bound, lower_bound):
+        output = output and (e >= low) and (e <= upp)
+    return output
+
+
+def check_boundary_3obj(x, upper_bound=UPPER_BOUND, lower_bound=LOWER_BOUND):
+    """Check the dimension of x and whether it is in the decision boundary. x is
+    decision variable, upper_bound and lower_bound are upperbound and lowerbound
+    lists of the decision space
+    """
+    lower_bound = [0.0] + lower_bound
+    upper_bound = [1.0] + upper_bound
+    if len(x) != len(upper_bound) or len(x) != len(lower_bound):
+        return False
+
+    output = True
+    for e, upp, low in zip(x, upper_bound, lower_bound):
+        output = output and (e >= low) and (e <= upp)
+    return output
+
+
+def fix_numerical_instability(x):
+    """Check whether x is close to zero, sqrt(0.5) or not. If it is close to
+    these two values, changes x to the value. Otherwise, return x.
+    """
+    if np.allclose(0.0, x):
+        return 0.0
+
+    if np.allclose(np.sqrt(0.5), x):
+        return np.sqrt(0.5)
+    return x
+
+
+def additive(alpha, beta):
+    """Additive form of the benchmark problem.
+    """
+    return [a + b for a, b in zip(alpha, beta)]
+
+
+#    return [alpha[0] + beta[0], alpha[1] + beta[1]]
+
+
+def multiplicative(alpha, beta):
+    """Multiplicative form of the benchmark problem.
+    """
+    return [a * (1 + b) for a, b in zip(alpha, beta)]
+
+
+#    return [alpha[0]*(1 + beta[0]), alpha[1]*(1 + beta[1])]
+
+
+## Benchmark functions ##
+def DB1a(x, t):
+    """DB1a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_uni(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB1m(x, t):
+    """DB1m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_uni(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB2a(x, t):
+    """DB2a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_multi(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB2m(x, t):
+    """DB2m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_multi(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB3a(x, t):
+    """DB3a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_mix(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB3m(x, t):
+    """DB3m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conv(x)
+        beta = beta_mix(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB4a(x, t):
+    """DB4a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_disc(x)
+        beta = beta_mix(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB4m(x, t):
+    """DB4m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_disc(x)
+        beta = beta_mix(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB5a(x, t):
+    """DB5a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_mix(x, t)
+        beta = beta_multi(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB5m(x, t):
+    """DB5m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_mix(x, t)
+        beta = beta_multi(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB6a(x, t):
+    """DB6a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_mix(x, t)
+        beta = beta_mix(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB6m(x, t):
+    """DB6m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_mix(x, t)
+        beta = beta_mix(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB7a(x, t):
+    """DB7a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf(x, t)
+        beta = beta_multi(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB7m(x, t):
+    """DB7m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf(x, t)
+        beta = beta_multi(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB8a(x, t):
+    """DB8a dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf(x, t)
+        beta = beta_mix(x, t, g)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB8m(x, t):
+    """DB8m dynamic benchmark problem
+    """
+    if check_boundary(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf(x, t)
+        beta = beta_mix(x, t, g)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB9a(x, t):
+    """DB9a dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type1(x, t)
+        beta = beta_multi(x, t, g, obj_num=3)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB9m(x, t):
+    """DB9m dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type1(x, t)
+        beta = beta_multi(x, t, g, obj_num=3)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB10a(x, t):
+    """DB10a dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type1(x, t)
+        beta = beta_mix(x, t, g, obj_num=3)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB10m(x, t):
+    """DB10m dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type1(x, t)
+        beta = beta_mix(x, t, g, obj_num=3)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB11a(x, t):
+    """DB11a dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type2(x, t)
+        beta = beta_multi(x, t, g, obj_num=3)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB11m(x, t):
+    """DB11m dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type2(x, t)
+        beta = beta_multi(x, t, g, obj_num=3)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB12a(x, t):
+    """DB12a dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type2(x, t)
+        beta = beta_mix(x, t, g, obj_num=3)
+        return additive(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def DB12m(x, t):
+    """DB12m dynamic benchmark problem
+    """
+    if check_boundary_3obj(x, UPPER_BOUND, LOWER_BOUND):
+        alpha = alpha_conf_3obj_type2(x, t)
+        beta = beta_mix(x, t, g, obj_num=3)
+        return multiplicative(alpha, beta)
+    else:
+        raise Exception(ERR_MSG)
+
+
+def fda2_deb(x, t):
+    f1 = x[0]
+    H = 2 * np.sin(0.5 * np.pi * (t - 1))
+    XII = x[1:6]
+    XIII = x[6:13]
+    g = 1 + np.sum(np.power(XII, 2))
+    Htemp = np.sum(np.power((XIII - H / 4), 2))
+    h = 1 - np.power((f1 / g), np.power(2, H + Htemp))
+    f2 = g * h
+    return [f1, f2]
+
+
+def FDA4(X, t):
+    """FDA4 dynamic benchmark problem
+    """
+    XII = X[2:]
+    G = fabs(sin(0.5 * pi * t))
+    g = sum([pow(xi - G, 2) for xi in XII])
+    f1 = (1 + g) * cos(X[0] * pi / 2) * cos(X[1] * pi / 2)
+    f2 = (1 + g) * cos(X[0] * pi / 2) * sin(X[1] * pi / 2)
+    f3 = (1 + g) * sin(X[0] * pi / 2)
+    return [f1, f2, f3]
+
+
+def FDA5(X, t):
+    """FDA5 dynamic benchmark problem
+    """
+    XII = X[2:]
+    G = fabs(sin(0.5 * pi * t))
+    g = G + sum([pow(xi - G, 2) for xi in XII])
+    F = 1 + 100 * pow(sin(0.5 * pi * t), 4)
+    y = lambda i: pow(X[i], F)
+    f1 = (1 + g) * cos(y(0) * pi / 2) * cos(y(1) * pi / 2)
+    f2 = (1 + g) * cos(y(0) * pi / 2) * sin(y(1) * pi / 2)
+    f3 = (1 + g) * sin(y(0) * pi / 2)
+    return [f1, f2, f3]
+
+
+def DIMP2(X, t):
+    """DIMP2 dynamic benchmark problem
+    """
+    n = len(X)
+    XII = X[1:]
+    g = 1.0 + 2.0 * (len(XII))
+    for k in range(1, n):
+        G = sin(pow(0.5 * pi * t + 2.0 * pi * float(k + 1) / float(n + 1.0), 2))
+        g += pow(X[k] - G, 2) - 2.0 * cos(3.0 * pi * (X[k] - G))
+    f1 = X[0]
+    h = 1 - sqrt(f1 / g)
+    f2 = g * h
+    return [f1, f2]
+
+
+def dMOP2(X, t):
+    """dMOP2 dynamic benchmark problem
+    """
+    XII = X[1:]
+    G = sin(0.5 * pi * t)
+    g = 1 + 9 * sum([pow(xi - G, 2) for xi in XII])
+    H = 0.75 * sin(0.5 * pi * t) + 1.25
+    f1 = X[0]
+    h = 1 - pow((f1 / g), H)
+    f2 = g * h
+    return [f1, f2]
+
+
+def dMOP3(X, tau, nt, taut, r, rIteration):
+    """dMOP3 dynamic benchmark problem
+    """
+    if tau % taut == 0 and rIteration != tau:
+        r = randint(0, 9)
+        rIteration = tau
+
+    XII = X[:r] + X[r + 1:]
+    t = float(1) / float(nt)
+    t = t * floor(float(tau) / float(taut))
+    G = sin(0.5 * pi * t)
+    g = 1 + 9 * sum([pow(xi - G, 2) for xi in XII])
+    H = 0.75 * sin(0.5 * pi * t)
+    f1 = X[r]
+    f2 = 1 - pow((f1 / g), H)
+    return [f1, f2, r, rIteration]
+
+
+def HE2(X, t):
+    """HE2 dynamic benchmark problem
+    """
+    n = 30
+    XII = X[1:]
+    H = 0.75 * sin(0.5 * pi * t) + 1.25
+    g = 1 + (9 / (n - 1)) * sum(XII)
+    f1 = X[0]
+    h = 1 - pow(sqrt(f1 / g), H) - pow(f1 / g, H) * sin(10 * pi * f1)
+    f2 = g * h
+    return [f1, f2]
+
+
+def HE7(X, t):
+    """HE7 dynamic benchmark problem
+    """
+
+    def _f1(input1):
+        value = input1[0]
+        ssum = 0.0
+        index = 0
+        for k in range(2, len(input1), 2):
+            val = 6 * pi * value + k * pi / len(input1)
+            ssum += pow(input1[k] - (0.3 * value * value * cos(4 * val) + 0.6 * value) * cos(val), 2)
+            index += 1
+        ssum *= 2.0 / index
+        ssum += value
+        return ssum
+
+    def _g(input1):
+        value = input1[0]
+        ssum = 0.0
+        index = 0
+        for k in range(1, len(input1), 2):
+            val = 6 * pi * value + k * pi / len(input1)
+            ssum += pow(input1[k] - (0.3 * value * value * cos(4 * val) + 0.6 * value) * sin(val), 2)
+            index += 1
+        ssum *= 2.0 / index
+        ssum += 2.0 - sqrt(value)
+        return ssum
+
+    f1 = _f1(X)
+    g = _g(X)
+    H = 0.75 * sin(0.5 * pi * t) + 1.25
+    h = 1 - pow(f1 / g, H)
+    f2 = g * h
+    return [f1, f2]
+
+
+def HE9(X, t):
+    """HE9 dynamic benchmark problem
+    """
+
+    def _f1(input1):
+        value = input1[0]
+        ssum = 0.0
+        index = 0
+        for k in range(2, len(input1), 2):
+            ssum += pow(input1[k] - sin(6 * pi * value + k * pi / len(input1)), 2)
+            index += 1
+        ssum *= 2.0 / index
+        ssum += value
+        return ssum
+
+    def _g(input1):
+        value = input1[0]
+        ssum = 0.0
+        index = 0
+        for k in range(1, len(input1), 2):
+            ssum += pow(input1[k] - sin(6 * pi * value + k * pi / len(input1)), 2)
+            index += 1
+        ssum *= 2.0 / index
+        ssum += 2.0 - pow(value, 2)
+        return ssum
+
+    f1 = _f1(X)
+    g = _g(X)
+    H = 0.75 * sin(0.5 * pi * t) + 1.25
+    h = 1 - pow(f1 / g, H)
+    f2 = g * h
+    return [f1, f2]