def log_likelihood(self, x, u):
loglik = []
for _x in x:
_loglik = np.column_stack([
multivariate_normal_logpdf(_x, self.mean(k), self.cov[k])
for k in range(self.nb_states)
])
loglik.append(_loglik)
return loglik
def log_likelihood(self, x, u):
loglik = []
for _x, _u in zip(x, u):
_loglik = np.column_stack([
multivariate_normal_logpdf(
_x[1:, :], self.mean(k, _x[:-1, :], _u[:-1, :self.dm_act]),
self.cov[k]) for k in range(self.nb_states)
])
loglik.append(_loglik)
return loglik
def log_likelihood(self, x, u):
loglik = []
for _x, _u in zip(x, u):
T = _x.shape[0]
_loglik = np.zeros((T - 1, self.nb_states))
for k in range(self.nb_states):
_mu = self.mean(k, _x[:-1, :], _u[:-1, :self.dm_act])
j = int(self.rot_lds[k, 1])
i = int(self.rot_lds[k, 0])
T_inv = self.T_inv[j, ...]
T = self.T[j, ...]
_mu = np.einsum('kh,...h->...k', T, _mu)
_x_t = np.einsum('kh,...h->...k', T, _x[1:, :])
#_x_t = _x[1:, :]
_loglik[:, k] = multivariate_normal_logpdf(
_x_t, _mu, self.cov[i, ...])
loglik.append(_loglik)
return loglik