pymoo 0.6.1.5.dev0__cp311-cp311-macosx_11_0_arm64.whl

pymoo/util/misc.py

@@ -0,0 +1,460 @@
+from collections import OrderedDict
+from datetime import datetime
+from itertools import combinations
+
+import numpy as np
+
+
+from pymoo.core.population import Population
+from pymoo.core.sampling import Sampling
+
+
+
+def parameter_less(F, CV, fmax=None, inplace=False):
+    assert len(F) == len(CV)
+
+    if not inplace:
+        F = np.copy(F)
+
+    if fmax is None:
+        fmax = np.max(F)
+
+    param_less = fmax + CV
+
+    infeas = (CV > 0).flatten()
+    F[infeas] = param_less[infeas]
+
+    return F
+
+
+def swap(M, a, b):
+    tmp = M[a]
+    M[a] = M[b]
+    M[b] = tmp
+
+
+# repairs a numpy array to be in bounds
+def repair(X, xl, xu):
+    larger_than_xu = X[0, :] > xu
+    X[0, larger_than_xu] = xu[larger_than_xu]
+
+    smaller_than_xl = X[0, :] < xl
+    X[0, smaller_than_xl] = xl[smaller_than_xl]
+
+    return X
+
+
+def unique_rows(a):
+    a = np.ascontiguousarray(a)
+    unique_a = np.unique(a.view([('', a.dtype)] * a.shape[1]))
+    return unique_a.view(a.dtype).reshape((unique_a.shape[0], a.shape[1]))
+
+
+def parameter_less_constraints(F, CV, F_max=None):
+    if F_max is None:
+        F_max = np.max(F)
+    has_constraint_violation = CV > 0
+    F[has_constraint_violation] = CV[has_constraint_violation] + F_max
+    return F
+
+
+def random_permutations(n, l, concat=True):
+    P = []
+    for i in range(n):
+        P.append(np.random.permutation(l))
+    if concat:
+        P = np.concatenate(P)
+    return P
+
+
+def get_duplicates(M):
+    res = []
+    I = np.lexsort([M[:, i] for i in reversed(range(0, M.shape[1]))])
+    S = M[I, :]
+
+    i = 0
+
+    while i < S.shape[0] - 1:
+        l = []
+        while np.all(S[i, :] == S[i + 1, :]):
+            l.append(I[i])
+            i += 1
+        if len(l) > 0:
+            l.append(I[i])
+            res.append(l)
+        i += 1
+
+    return res
+
+
+# -----------------------------------------------
+# Euclidean Distance
+# -----------------------------------------------
+
+def func_euclidean_distance(a, b):
+    return np.sqrt(((a - b) ** 2).sum(axis=1))
+
+
+def func_norm_euclidean_distance(xl, xu):
+    return lambda a, b: np.sqrt((((a - b) / (xu - xl)) ** 2).sum(axis=1))
+
+
+def norm_eucl_dist_by_bounds(A, B, xl, xu, **kwargs):
+    return vectorized_cdist(A, B, func_dist=func_norm_euclidean_distance(xl, xu), **kwargs)
+
+
+def norm_eucl_dist(problem, A, B, **kwargs):
+    return norm_eucl_dist_by_bounds(A, B, *problem.bounds(), **kwargs)
+
+
+# -----------------------------------------------
+# Manhatten Distance
+# -----------------------------------------------
+
+def func_manhatten_distance(a, b):
+    return np.abs(a - b).sum(axis=1)
+
+
+def func_norm_manhatten_distance(xl, xu):
+    return lambda a, b: np.abs((a - b) / (xu - xl)).sum(axis=1)
+
+
+def norm_manhatten_dist_by_bounds(A, B, xl, xu, **kwargs):
+    return vectorized_cdist(A, B, func_dist=func_norm_manhatten_distance(xl, xu), **kwargs)
+
+
+def norm_manhatten_dist(problem, A, B, **kwargs):
+    return norm_manhatten_dist_by_bounds(A, B, *problem.bounds(), **kwargs)
+
+
+# -----------------------------------------------
+# Tchebychev Distance
+# -----------------------------------------------
+
+
+def func_tchebychev_distance(a, b):
+    return np.abs(a - b).max(axis=1)
+
+
+def func_norm_tchebychev_distance(xl, xu):
+    return lambda a, b: np.abs((a - b) / (xu - xl)).max(axis=1)
+
+
+def norm_tchebychev_dist_by_bounds(A, B, xl, xu, **kwargs):
+    return vectorized_cdist(A, B, func_dist=func_norm_tchebychev_distance(xl, xu), **kwargs)
+
+
+def norm_tchebychev_dist(problem, A, B, **kwargs):
+    return norm_tchebychev_dist_by_bounds(A, B, *problem.bounds(), **kwargs)
+
+
+# -----------------------------------------------
+# Others
+# -----------------------------------------------
+
+
+def cdist(A, B, **kwargs):
+    from scipy.spatial import distance
+    return distance.cdist(A.astype(float), B.astype(float), **kwargs)
+
+
+def vectorized_cdist(A, B, func_dist=func_euclidean_distance, fill_diag_with_inf=False, **kwargs) -> object:
+    assert A.ndim <= 2 and B.ndim <= 2
+
+    A, only_row = at_least_2d_array(A, extend_as="row", return_if_reshaped=True)
+    B, only_column = at_least_2d_array(B, extend_as="row", return_if_reshaped=True)
+
+    u = np.repeat(A, B.shape[0], axis=0)
+    v = np.tile(B, (A.shape[0], 1))
+
+    D = func_dist(u, v, **kwargs)
+    M = np.reshape(D, (A.shape[0], B.shape[0]))
+
+    if fill_diag_with_inf:
+        np.fill_diagonal(M, np.inf)
+
+    if only_row and only_column:
+        M = M[0, 0]
+    elif only_row:
+        M = M[0]
+    elif only_column:
+        M = M[:, [0]]
+
+    return M
+
+
+def covert_to_type(problem, X):
+    if problem.vtype == float:
+        return X.astype(np.double)
+    elif problem.vtype == int:
+        return np.round(X).astype(int)
+    elif problem.vtype == bool:
+        return X < (problem.xu - problem.xl) / 2
+
+
+def find_duplicates(X, epsilon=1e-16):
+    # calculate the distance matrix from each point to another
+    D = cdist(X, X)
+
+    # set the diagonal to infinity
+    D[np.triu_indices(len(X))] = np.inf
+
+    # set as duplicate if a point is really close to this one
+    is_duplicate = np.any(D <= epsilon, axis=1)
+
+    return is_duplicate
+
+
+def at_least_2d(*args, **kwargs):
+    ret = tuple([at_least_2d_array(arg, **kwargs) for arg in args])
+    if len(ret) == 1:
+        ret = ret[0]
+    return ret
+
+
+def at_least_2d_array(x, extend_as="row", return_if_reshaped=False):
+    if x is None:
+        return x
+    elif not isinstance(x, np.ndarray):
+        x = np.array([x])
+
+    has_been_reshaped = False
+
+    if x.ndim == 1:
+        if extend_as.startswith("r"):
+            x = x[None, :]
+        elif extend_as.startswith("c"):
+            x = x[:, None]
+        else:
+            raise Exception("The option `extend_as` should be either `row` or `column`.")
+
+        has_been_reshaped = True
+
+    if return_if_reshaped:
+        return x, has_been_reshaped
+    else:
+        return x
+
+
+def to_1d_array_if_possible(x):
+    if not isinstance(x, np.ndarray):
+        x = np.array([x])
+
+    if x.ndim == 2:
+        if x.shape[0] == 1 or x.shape[1] == 1:
+            x = x.flatten()
+
+    return x
+
+
+def stack(*args, flatten=True):
+    if not flatten:
+        ps = np.concatenate([e[None, ...] for e in args])
+    else:
+        ps = np.row_stack(args)
+    return ps
+
+
+def all_except(x, *args):
+    if len(args) == 0:
+        return x
+    else:
+        H = set(args) if len(args) > 5 else args
+        I = [k for k in range(len(x)) if k not in H]
+        return x[I]
+
+
+def all_combinations(A, B):
+    u = np.repeat(A, B.shape[0], axis=0)
+    v = np.tile(B, A.shape[0])
+    return np.column_stack([u, v])
+
+
+def pop_from_sampling(problem, sampling, n_initial_samples, pop=None):
+    # the population type can be different - (different type of individuals)
+    if pop is None:
+        pop = Population()
+
+    # provide a whole population object - (individuals might be already evaluated)
+    if isinstance(sampling, Population):
+        pop = sampling
+
+    else:
+        # if just an X array create a pop
+        if isinstance(sampling, np.ndarray):
+            pop = pop.new("X", sampling)
+
+        elif isinstance(sampling, Sampling):
+            # use the sampling
+            pop = sampling.do(problem, n_initial_samples, pop=pop)
+
+        else:
+            return None
+
+    return pop
+
+
+def evaluate_if_not_done_yet(evaluator, problem, pop, algorithm=None):
+    I = np.where(pop.get("F") == None)[0]
+    if len(I) > 0:
+        pop[I] = evaluator.process(problem, pop[I], algorithm=algorithm)
+
+
+def set_if_none(kwargs, str, val):
+    if str not in kwargs:
+        kwargs[str] = val
+
+
+def set_if_none_from_tuples(kwargs, *args):
+    for key, val in args:
+        if key not in kwargs:
+            kwargs[key] = val
+
+
+def calc_perpendicular_distance(N, ref_dirs):
+    u = np.tile(ref_dirs, (len(N), 1))
+    v = np.repeat(N, len(ref_dirs), axis=0)
+
+    norm_u = np.linalg.norm(u, axis=1)
+
+    scalar_proj = np.sum(v * u, axis=1) / norm_u
+    proj = scalar_proj[:, None] * u / norm_u[:, None]
+    val = np.linalg.norm(proj - v, axis=1)
+    matrix = np.reshape(val, (len(N), len(ref_dirs)))
+
+    return matrix
+
+
+def distance_of_closest_points_to_others(X):
+    D = vectorized_cdist(X, X)
+    np.fill_diagonal(D, np.inf)
+    return D.argmin(axis=1), D.min(axis=1)
+
+
+def time_to_int(t):
+    vals = [int(e) for e in t.split(":")][::-1]
+    s = vals[0]
+    if len(vals) > 1:
+        s += 60 * vals[1]
+    if len(vals) > 2:
+        s += 3600 * vals[2]
+    return s
+
+
+def powerset(iterable):
+    for n in range(len(iterable) + 1):
+        yield from combinations(iterable, n)
+
+
+def intersect(a, b):
+    H = set()
+    for entry in b:
+        H.add(entry)
+
+    ret = []
+    for entry in a:
+        if entry in H:
+            ret.append(entry)
+
+    return ret
+
+
+def has_feasible(pop):
+    return np.any(pop.get("FEAS"))
+
+
+def to_numpy(a):
+    return np.array(a)
+
+
+def termination_from_tuple(termination):
+    from pymoo.core.termination import Termination
+
+    # get the termination if provided as a tuple - create an object
+    if termination is not None and not isinstance(termination, Termination):
+        from pymoo.termination import get_termination
+        if isinstance(termination, str):
+            termination = get_termination(termination)
+        else:
+            termination = get_termination(*termination)
+
+    return termination
+
+
+def unique_and_all_indices(arr):
+    sort_indexes = np.argsort(arr)
+    arr = np.asarray(arr)[sort_indexes]
+    vals, first_indexes, inverse, counts = np.unique(arr,
+                                                     return_index=True, return_inverse=True, return_counts=True)
+    indexes = np.split(sort_indexes, first_indexes[1:])
+    for x in indexes:
+        x.sort()
+    return vals, indexes
+
+
+def from_dict(D, *keys):
+    return [D.get(k) for k in keys]
+
+
+def list_of_dicts_unique(l, k):
+    return list(OrderedDict([(e[k], None) for e in l]).keys())
+
+
+def list_of_dicts_filter(l, *pairs):
+    return [e for e in l if all(e[k] == v for (k, v) in pairs)]
+
+
+def logical_op(func, a, b, *args):
+    ret = func(a, b)
+    for c in args:
+        ret = func(ret, c)
+    return ret
+
+
+def replace_nan_by(x, val, inplace=False):
+    is_nan = np.isnan(x)
+    if np.sum(is_nan) > 0:
+        if not inplace:
+            x = x.copy()
+        x[is_nan] = val
+    return x
+
+
+def set_defaults(kwargs, defaults, overwrite=False, func_get=lambda x: x):
+    for k, v in defaults.items():
+        if overwrite or k not in kwargs:
+            kwargs[k] = func_get(v)
+
+
+def filter_params(params, prefix, delete_prefix=True):
+    ret = {}
+    for k, v in params.items():
+        if k.startswith(prefix):
+            if delete_prefix:
+                k = k[len(prefix):]
+            ret[k] = v
+    return ret
+
+
+def where_is_what(x):
+    H = {}
+    for k, e in enumerate(x):
+        if e not in H:
+            H[e] = []
+        H[e].append(k)
+    return H
+
+
+def crossover_mask(X, M):
+    # convert input to output by flatting along the first axis
+    _X = np.copy(X)
+    _X[0][M] = X[1][M]
+    _X[1][M] = X[0][M]
+    return _X
+
+
+def row_at_least_once_true(M):
+    _, d = M.shape
+    for k in np.where(~np.any(M, axis=1))[0]:
+        M[k, np.random.randint(d)] = True
+    return M

pymoo/util/mnn.py

@@ -0,0 +1,70 @@
+import numpy as np
+from scipy.spatial.distance import pdist, squareform
+
+def calc_mnn(X, n_remove=0):
+    return calc_mnn_base(X, n_remove=n_remove, twonn=False)
+
+def calc_2nn(X, n_remove=0):
+    return calc_mnn_base(X, n_remove=n_remove, twonn=True)
+
+def calc_mnn_base(X, n_remove=0, twonn=False):
+
+    N = X.shape[0]
+    M = X.shape[1]
+    
+    if N <= M:
+        return np.full(N, np.inf)
+
+    if n_remove <= (N - M):
+        if n_remove < 0:
+            n_remove = 0
+        else:
+            pass
+    else:
+        n_remove = N - M
+    
+    if twonn:
+        M = 2
+
+    extremes_min = np.argmin(X, axis=0)
+    extremes_max = np.argmax(X, axis=0)
+    
+    min_vals = np.min(X, axis=0)
+    max_vals = np.max(X, axis=0)
+
+    extremes = np.concatenate((extremes_min, extremes_max))
+    
+    X = (X - min_vals) / (max_vals - min_vals)
+    
+    H = np.arange(N)
+    
+    D = squareform(pdist(X, metric="sqeuclidean"))
+    Dnn = np.partition(D, range(1, M+1), axis=1)[:, 1:M+1]
+    d = np.prod(Dnn, axis=1)
+    d[extremes] = np.inf
+    
+    n_removed = 0
+
+    #While n_remove not acheived
+    while n_removed < (n_remove - 1):
+
+        #Obtain element to drop
+        _d = d[H]
+        _k = np.argmin(_d)
+        k = H[_k]
+        H = H[H != k]
+
+        #Update index
+        n_removed = n_removed + 1
+        if n_removed == n_remove:
+            break
+
+        else:
+
+            D[:, k] = np.inf
+            Dnn[H] = np.partition(D[H], range(1, M+1), axis=1)[:, 1:M+1]
+            d[H] = np.prod(Dnn[H], axis=1)
+            d[extremes] = np.inf
+    
+    return d
+

pymoo/util/nds/dominance_degree_non_dominated_sort.py

@@ -0,0 +1,159 @@
+"""Module which implements Dominance Degree Approaches for Non-dominated Sorting.
+
+For the original work see:
+    DDA-NS https://ieeexplore.ieee.org/document/7469397
+    DDA-ENS https://ieeexplore.ieee.org/document/9282978
+
+Adapted from https://github.com/rsenwar/Non-Dominated-Sorting-Algorithms/tree/master
+"""
+
+
+from typing import Literal, List
+import numpy as np
+
+
+def construct_comp_matrix(vec: np.ndarray, sorted_idx: np.ndarray) -> np.ndarray:
+    """
+    const_comp_mat construct the comparison matrix from a row-vector vec.
+
+    Parameters
+    ----------
+    vec : np.ndarray
+        The vector of scores for the population on a single objective
+    sorted_idx : np.ndarray
+        The indices which would sort `vec`
+
+    Returns
+    -------
+    np.ndarray
+        The comparison matrix indicating whether each member in the population dominates the other member for the
+        objective in `vec`
+    """
+    n = vec.shape[0]
+    c = np.zeros(shape=(n, n), dtype=np.int32)
+
+    # the elements of the b(0)-th row in C are all set to 1
+    c[sorted_idx[0], :] = 1
+
+    for i in range(1, n):
+        if vec[sorted_idx[i]] == vec[sorted_idx[i - 1]]:
+            # the rows in C corresponding to the same elements in w are identical
+            c[sorted_idx[i]] = c[sorted_idx[i - 1]]
+        else:
+            c[sorted_idx[i], sorted_idx[i:]] = 1
+
+    return c
+
+
+def construct_domination_matrix(f_scores: np.ndarray, **kwargs) -> np.ndarray:
+    """
+    construct_domination_matrix calculates the dominance degree matrix for a set of vectors.
+
+    The dominance degree indicate the degree of dominance of a solution, which is the number of
+    objectives for which it is the dominating solution.
+
+    Parameters
+    ----------
+    f_scores : np.ndarray
+        an N x M matrix of N (population size) objective function values for M objectives
+    """
+    d = np.zeros((f_scores.shape[0], f_scores.shape[0]), dtype=np.int32)
+    b = np.apply_over_axes(np.argsort, f_scores, axes=0)
+    for vec, srt in zip(f_scores.T, b.T):
+        d += construct_comp_matrix(vec, srt)
+    d = np.where(
+        np.logical_and(d == f_scores.shape[-1], d.T == f_scores.shape[-1]), 0, d
+    )
+    return d
+
+
+def dda_ns(f_scores: np.ndarray, **kwargs) -> List[List[int]]:
+    """
+    dda_ns runs the DDA-NS algorithm.
+
+    Parameters
+    ----------
+    f_scores : np.ndarray
+        an N x M matrix of N (population size) objective function values for M objectives
+
+    Returns
+    -------
+    List[List[int]]
+        A list of members of each Pareto front. The index in the outer most list corresponds to the level in the Pareto front
+        while the value in the inner-most list is the id of the member of the population belonging to that front.
+    """
+    d_mx = construct_domination_matrix(f_scores)
+    max_d = np.empty((f_scores.shape[0],), dtype=np.int32)
+
+    fronts = []
+    count = 0
+    while count < f_scores.shape[0]:
+        # Max(D) is the row vector containing the maximum elements from each column of D
+        np.max(d_mx, out=max_d, axis=0)
+        front = [i for i, m_d in enumerate(max_d) if 0 <= m_d < f_scores.shape[-1]]
+        count += len(front)
+        d_mx[front] = -1
+        d_mx[:, front] = -1
+        fronts.append(front)
+
+    return fronts
+
+
+def dda_ens(f_scores: np.ndarray, **kwargs) -> List[List[int]]:
+    """
+    dda_ens runs the DDA-ENS (efficient DDA) algorithm
+
+    Parameters
+    ----------
+    f_scores : np.ndarray
+        The N x M matrix of N (population size) objective function values for M objectives
+
+    Returns
+    -------
+    List[List[int]]
+        an N x M matrix of N (population size) objective function values for M objectives
+    """
+    d_mx = construct_domination_matrix(f_scores)
+
+    fronts: List[List[int]] = []
+    for s in np.lexsort(f_scores.T):
+        isinserted = False
+        for fk in fronts:
+            if not (d_mx[fk, s] == f_scores.shape[1]).any():
+                fk.append(s)
+                isinserted = True
+                break
+        if not isinserted:
+            fronts.append([s])
+    return fronts
+
+
+def dominance_degree_non_dominated_sort(
+    f_scores: np.ndarray, strategy: Literal["efficient", "fast"] = "efficient"
+) -> List[List[int]]:
+    """
+    dominance_degree_non_dominated_sort performs the non-dominating sort with the specified algorithm
+
+    Parameters
+    ----------
+    f_scores : np.ndarray
+        The N x M matrix of N (population size) objective function values for M objectives
+    strategy : Literal["efficient", "fast"], optional
+        The dominance degree algorithm to use, by default "efficient"
+
+    Returns
+    -------
+    List[List[int]]
+        A list of members of each Pareto front. The index in the outer most list corresponds to the level in the Pareto front
+        while the value in the inner-most list is the id of the member of the population belonging to that front.
+
+    Raises
+    ------
+    ValueError
+        If an invalid strategy is specified
+    """
+    if strategy == "efficient":
+        return dda_ens(f_scores)
+    if strategy == "fast":
+        return dda_ns(f_scores)
+    raise ValueError("Invalid search strategy")