def __init__(self, mu, sig, w=1):
Potential.__init__(self, symmetric=False)
self.mu = np.array(mu)
self.sig = np.matrix(sig)
self.prec = self.sig.I
det = np.linalg.det(self.sig)
p = float(len(mu))
if det == 0:
raise NameError("The covariance matrix can't be singular")
self.coefficient = w / (pow(2 * pi, p * 0.5) * pow(det, 0.5))
def __init__(self, A, b, c):
"""
:param A: an array of shape [v1, v2, ..., v_Nd, Nc, Nc], where v1, ..., v_Nd are the number of states of the Nd
discrete nodes in the scope, and Nc is the number of continuous nodes
:param b: an array of shape [v1, v2, ..., v_Nd, Nc]
:param c: an array of shape [v1, v2, ..., v_Nd]
:return:
"""
Potential.__init__(self, symmetric=False)
self.A = A
self.b = b
self.c = c
self.Nd = len(c.shape) # num disc nodes
self.Nc = int(b.shape[-1]) # num cont nodes
self.log_potential = LogHybridQuadratic(A, b, c)
def __init__(self, A, b, c):
Potential.__init__(self, symmetric=False)
self.A = np.array(A)
self.b = np.array(b)
self.c = c
self.log_potential = LogQuadratic(A, b, c)
def __init__(self, distant_cof, scaling_cof, max_threshold):
Potential.__init__(self, symmetric=True)
self.distant_cof = distant_cof
self.scaling_cof = scaling_cof
self.max_threshold = max_threshold
self.v = pow(e, -self.max_threshold / self.scaling_cof)
def __init__(self, mu, sig):
Potential.__init__(self, symmetric=True)
self.mu = mu
self.sig = sig
def __init__(self, coeff, sig):
Potential.__init__(self, symmetric=True)
self.coeff = coeff
self.sig = sig
def __init__(self, table, symmetric=False):
Potential.__init__(self, symmetric=symmetric)
self.table = table