# -*- coding: utf-8 -*-
""" The generic KL-UCB policy for one-parameter exponential distributions, using a different exploration time step for each arm (:math:`\log(t_k) + c \log(\log(t_k))` instead of :math:`\log(t) + c \log(\log(t))`).
- It is designed to be used with the wrapper :class:`GLR_UCB`.
- By default, it assumes Bernoulli arms.
- Reference: [Garivier & Cappé - COLT, 2011](https://arxiv.org/pdf/1102.2490.pdf).
"""
from __future__ import division, print_function # Python 2 compatibility
__author__ = "Lilian Besson"
__version__ = "0.9"
from math import log
import numpy as np
np.seterr(divide='ignore') # XXX dangerous in general, controlled here!
try:
from .kullback import klucbBern
from .klUCBloglog import klUCBloglog
except (ImportError, SystemError):
from kullback import klucbBern
from klUCBloglog import klUCBloglog
#: Default value for the constant c used in the computation of KL-UCB index.
c = 3 #: Default value when using :math:`f(t) = \log(t) + c \log(\log(t))`, as :class:`klUCB_forGLR` is inherited from :class:`klUCBloglog`.
#: Default value for the tolerance for computing numerical approximations of the kl-UCB indexes.
TOLERANCE = 1e-4
# --- Class
[docs]class klUCB_forGLR(klUCBloglog):
""" The generic KL-UCB policy for one-parameter exponential distributions, using a different exploration time step for each arm (:math:`\log(t_k) + c \log(\log(t_k))` instead of :math:`\log(t) + c \log(\log(t))`).
- It is designed to be used with the wrapper :class:`GLR_UCB`.
- By default, it assumes Bernoulli arms.
- Reference: [Garivier & Cappé - COLT, 2011](https://arxiv.org/pdf/1102.2490.pdf).
"""
[docs] def __init__(self, nbArms, tolerance=TOLERANCE, klucb=klucbBern, c=c, lower=0., amplitude=1.):
super(klUCB_forGLR, self).__init__(nbArms, tolerance=tolerance, klucb=klucb, c=c, lower=lower, amplitude=amplitude)
self.t_for_each_arm = np.zeros(nbArms, dtype=int) #: Keep in memory not only the global time step :math:`t`, but also let the possibility for :class:`GLR_UCB` to use a different time steps :math:`t_k` for each arm, in the exploration function :math:`f(t) = \log(t_k) + 3 \log(\log(t_k))`.
[docs] def computeIndex(self, arm):
r""" Compute the current index, at time t and after :math:`N_k(t)` pulls of arm k:
.. math::
\hat{\mu}_k(t) &= \frac{X_k(t)}{N_k(t)}, \\
U_k(t) &= \sup\limits_{q \in [a, b]} \left\{ q : \mathrm{kl}(\hat{\mu}_k(t), q) \leq \frac{\log(t_k) + c \log(\log(t_k))}{N_k(t)} \right\},\\
I_k(t) &= U_k(t).
If rewards are in :math:`[a, b]` (default to :math:`[0, 1]`) and :math:`\mathrm{kl}(x, y)` is the Kullback-Leibler divergence between two distributions of means x and y (see :mod:`Arms.kullback`),
and c is the parameter (default to 1).
.. warning:: The only difference with :class:`klUCB` is that a custom :math:`t_k` is used for each arm k, instead of a common :math:`t`. This policy is designed to be used with :class:`GLR_UCB`.
"""
if self.pulls[arm] < 1:
return float('+inf')
else:
# XXX We could adapt tolerance to the value of self.t
return self.klucb(self.rewards[arm] / self.pulls[arm], self.c * log(self.t_for_each_arm[arm]) / self.pulls[arm], self.tolerance)
[docs] def computeAllIndex(self):
""" Compute the current indexes for all arms, in a vectorized manner."""
indexes = self.klucb_vect(self.rewards / self.pulls, self.c * np.log(self.t_for_each_arm) / self.pulls, self.tolerance)
indexes[self.pulls < 1] = float('+inf')
self.index[:] = indexes