""" Statistical Ensembles """ from csb.numeric import log, exp from abc import ABCMeta, abstractmethod class StatisticalEnsemble(object): __metaclass__ = ABCMeta def __call__(self, raw_energies): return exp(-self.energy(raw_energies)) def log_prob(self, raw_energies): return -self.energy(raw_energies) @abstractmethod def energy(self, raw_energies): """ Transforms the raw energies as if they were observed in this statistical ensemble """ pass def gradient(self, raw_energies): raise NotImplementedError() class BoltzmannEnsemble(StatisticalEnsemble): def __init__(self, beta=1.): self._beta = float(beta) @property def beta(self): """ Inverse temperature """ return self._beta @beta.setter def beta(self, value): value = float(value) if value <= 0.: raise ValueError("Inverse temperature {0} < 0".formate(value)) self._beta = value def energy(self, raw_energies): return raw_energies * self._beta class FermiEnsemble(BoltzmannEnsemble): def __init__(self, beta=1., e_max=0.): super(FermiEnsemble, self).__init__(beta) self._e_max = float(e_max) @property def e_max(self): """ Maximum energy """ return self._e_max @e_max.setter def e_max(self, value): self._e_max = float(value) def energy(self, raw_energies): from numpy import isinf if isinf(self.beta): m = (raw_energies >= self.e_max).astype('f') return - m * log(0.) else: x = 1 + exp(self.beta * (raw_energies - self.e_max)) return log(x) class TsallisEnsemble(StatisticalEnsemble): def __init__(self, q=1., e_min=0.): self._q = q self._e_min = e_min @property def q(self): """ q-analoge of the temperature """ return self._q @q.setter def q(self, value): if value <= 0.: raise ValueError("Inverse temperature {0} < 0".formate(value)) self._q = value @property def e_min(self): """ lower bound of the energy """ return self._e_min @e_min.setter def e_min(self, value): self._e_min = value def energy(self, raw_energies): q = self.q e_min = self.e_min if (q < 1 + 1e-10): return raw_energies * q else: return log(1 + (raw_energies - e_min) * (q - 1)) * q / (q - 1) + e_min class CompositeEnsemble(StatisticalEnsemble): def __init__(self, ensembles=[]): self._ensembles = ensembles @property def ensembles(self): """ Collection of statistical ensembles """ return self._ensembles @ensembles.setter def ensembles(self, value): if not isinstance(value, list): if len(value) > 0: if not isinstance(value[0], StatisticalEnsemble): raise ValueError("Not a list of statistical ensembles") else: self._enesmbles = value else: self._enesmbles = value def energy(self, raw_energies): return sum([self._ensembles[i].energy(raw_energies[i]) for i in range(len(self.ensembles))], 0) def gradient(self, raw_energies): return sum([self._ensembles[i].gradient(raw_energies[i]) for i in range(len(self.ensembles))], 0)