apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-04-27 22:06:22 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

domain and trtansformations namespace prettyfying

Browse source
This commit is contained in:
Max Zwiessele 2013-10-11 17:26:04 +01:00
parent 780daac03e
commit 2018368d8d
5 changed files with 36 additions and 41 deletions
Download patch file Download diff file Expand all files Collapse all files

18
GPy/core/domains.py
View file

@ -6,21 +6,21 @@ Created on 4 Jun 2013
(Hyper-)Parameter domains defined for :py:mod:`~GPy.core.priors` and :py:mod:`~GPy.kern`.
These domains specify the legitimate realm of the parameters to live in.
:const:`~GPy.core.domains.REAL` :
:const:`~GPy.core.domains._REAL` :
real domain, all values in the real numbers are allowed
:const:`~GPy.core.domains.POSITIVE`:
:const:`~GPy.core.domains._POSITIVE`:
positive domain, only positive real values are allowed
:const:`~GPy.core.domains.NEGATIVE`:
same as :const:`~GPy.core.domains.POSITIVE`, but only negative values are allowed
:const:`~GPy.core.domains._NEGATIVE`:
same as :const:`~GPy.core.domains._POSITIVE`, but only negative values are allowed
:const:`~GPy.core.domains.BOUNDED`:
:const:`~GPy.core.domains._BOUNDED`:
only values within the bounded range are allowed,
the bounds are specified withing the object with the bounded range
'''
REAL = 'real'
POSITIVE = "positive"
NEGATIVE = 'negative'
BOUNDED = 'bounded'
_REAL = 'real'
_POSITIVE = "positive"
_NEGATIVE = 'negative'
_BOUNDED = 'bounded'

8
GPy/core/model.py
View file

@ -9,7 +9,7 @@ from GPy.util.misc import opt_wrapper
from parameterized import Parameterized
import multiprocessing as mp
import numpy as np
from GPy.core.domains import POSITIVE, REAL
from GPy.core.domains import _POSITIVE, _REAL
from numpy.linalg.linalg import LinAlgError
# import numdifftools as ndt
@ -93,8 +93,8 @@ class Model(Parameterized):
# check constraints are okay
if what.domain is POSITIVE:
constrained_positive_indices = [i for i, t in zip(self.constrained_indices, self.constraints) if t.domain is POSITIVE]
if what.domain is _POSITIVE:
constrained_positive_indices = [i for i, t in zip(self.constrained_indices, self.constraints) if t.domain is _POSITIVE]
if len(constrained_positive_indices):
constrained_positive_indices = np.hstack(constrained_positive_indices)
else:
@ -107,7 +107,7 @@ class Model(Parameterized):
print '\n'.join([n for i, n in enumerate(self._get_param_names()) if i in unconst])
print '\n'
self.constrain_positive(unconst)
elif what.domain is REAL:
elif what.domain is _REAL:
assert not np.any(which[:, None] == self.all_constrained_indices()), "constraint and prior incompatible"
else:
raise ValueError, "prior not recognised"

13
GPy/core/parameter.py
View file

@ -41,7 +41,7 @@ class Parameterized(object):
- m.name regular expression matches all parameters beginning with name
- m['name'] regular expression matches all parameters with name
Handling of constraining, fixing and tieing parameters together:
Handling of constraining, fixing and tieing parameters:
@ -74,28 +74,23 @@ class Parameterized(object):
self.__dict__[p.name] = p
#===========================================================================
# Optimization handling:
# Optimization handles:
#===========================================================================
def _get_params(self):
return numpy.hstack([x._get_params() for x in self._params])#numpy.fromiter(itertools.chain(*itertools.imap(lambda x: x._get_params(), self._params)), dtype=numpy.float64, count=sum(self.parameter_sizes))
return numpy.hstack([x._get_params() for x in self._params])#numpy.fromiter(itertools.chain(*itertools.imap(lambda x: x._get_params(), self._params)), dtype=numpy.float64, count=sum(self.parameter_sizes))
def _set_params(self, params):
[p._set_params(params[s]) for p,s in itertools.izip(self._params,self._param_slices)]
def _get_params_transformed(self):
p = self._get_params()
[numpy.put(p, ind, c.finv(p[ind])) for c,ind in self._constraints.iteritems() if c is not __fixed__]
if self._ties_fixes is not None:
return p[self._ties_fixes]
return p
def _set_params_transformed(self, p):
if self._ties_fixes is not None: tmp = self._get_params(); tmp[self._ties_fixes] = p; p = tmp; del tmp
[numpy.put(p, ind, c.f(p[ind])) for c,ind in self._constraints.iteritems() if c is not __fixed__]
[numpy.put(p, f, p[t]) for f,t in self._ties.iter_from_to_indices()]
self._set_params(p)
def _handle_ties(self):
if not self._init:
self._set_params_transformed(self._get_params_transformed())
@ -466,7 +461,7 @@ class ParamConcatenation(object):
return "\n".join(map(repr,self.params))
if __name__ == '__main__':
X = numpy.random.randn(3,3,2)
X = numpy.random.randn(4,2)
p = Param("q_mean", X)
p1 = Param("q_variance", numpy.random.rand(*p.shape))
p2 = Param("Y", numpy.random.randn(p.shape[0],1))

12
GPy/core/priors.py
View file

@ -6,7 +6,7 @@ import numpy as np
import pylab as pb
from scipy.special import gammaln, digamma
from ..util.linalg import pdinv
from GPy.core.domains import REAL, POSITIVE
from GPy.core.domains import _REAL, _POSITIVE
import warnings
class Prior:
@ -32,7 +32,7 @@ class Gaussian(Prior):
.. Note:: Bishop 2006 notation is used throughout the code
"""
domain = REAL
domain = _REAL
def __init__(self, mu, sigma):
self.mu = float(mu)
self.sigma = float(sigma)
@ -62,7 +62,7 @@ class LogGaussian(Prior):
.. Note:: Bishop 2006 notation is used throughout the code
"""
domain = POSITIVE
domain = _POSITIVE
def __init__(self, mu, sigma):
self.mu = float(mu)
self.sigma = float(sigma)
@ -92,7 +92,7 @@ class MultivariateGaussian:
.. Note:: Bishop 2006 notation is used throughout the code
"""
domain = REAL
domain = _REAL
def __init__(self, mu, var):
self.mu = np.array(mu).flatten()
self.var = np.array(var)
@ -147,7 +147,7 @@ class Gamma(Prior):
.. Note:: Bishop 2006 notation is used throughout the code
"""
domain = POSITIVE
domain = _POSITIVE
def __init__(self, a, b):
self.a = float(a)
self.b = float(b)
@ -198,7 +198,7 @@ class inverse_gamma(Prior):
.. Note:: Bishop 2006 notation is used throughout the code
"""
domain = POSITIVE
domain = _POSITIVE
def __init__(self, a, b):
self.a = float(a)
self.b = float(b)

26
GPy/core/transformations.py
View file

@ -3,10 +3,10 @@
import numpy as np
from GPy.core.domains import POSITIVE, NEGATIVE, BOUNDED
from GPy.core.domains import _POSITIVE,_NEGATIVE, _BOUNDED
import sys
import weakref
lim_val = -np.log(sys.float_info.epsilon)
_lim_val = -np.log(sys.float_info.epsilon)
class Transformation(object):
domain = None
@ -29,13 +29,13 @@ class Transformation(object):
raise NotImplementedError
class Logexp(Transformation):
domain = POSITIVE
domain = _POSITIVE
def f(self, x):
return np.where(x>lim_val, x, np.log(1. + np.exp(x)))
return np.where(x>_lim_val, x, np.log(1. + np.exp(x)))
def finv(self, f):
return np.where(f>lim_val, f, np.log(np.exp(f) - 1.))
return np.where(f>_lim_val, f, np.log(np.exp(f) - 1.))
def gradfactor(self, f):
return np.where(f>lim_val, 1., 1 - np.exp(-f))
return np.where(f>_lim_val, 1., 1 - np.exp(-f))
def initialize(self, f):
if np.any(f < 0.):
print "Warning: changing parameters to satisfy constraints"
@ -44,7 +44,7 @@ class Logexp(Transformation):
return '+ve'
class NegativeLogexp(Transformation):
domain = NEGATIVE
domain = _NEGATIVE
logexp = Logexp()
def f(self, x):
return -self.logexp.f(x) # np.log(1. + np.exp(x))
@ -62,7 +62,7 @@ class LogexpClipped(Logexp):
min_bound = 1e-10
log_max_bound = np.log(max_bound)
log_min_bound = np.log(min_bound)
domain = POSITIVE
domain = _POSITIVE
_instances = []
def __new__(cls, lower=1e-6, *args, **kwargs):
if cls._instances:
@ -97,9 +97,9 @@ class LogexpClipped(Logexp):
class Exponent(Transformation):
# TODO: can't allow this to go to zero, need to set a lower bound. Similar with negative Exponent below. See old MATLAB code.
domain = POSITIVE
domain = _POSITIVE
def f(self, x):
return np.where(x<lim_val, np.where(x>-lim_val, np.exp(x), np.exp(-lim_val)), np.exp(lim_val))
return np.where(x<_lim_val, np.where(x>-_lim_val, np.exp(x), np.exp(-_lim_val)), np.exp(_lim_val))
def finv(self, x):
return np.log(x)
def gradfactor(self, f):
@ -112,7 +112,7 @@ class Exponent(Transformation):
return '+ve'
class NegativeExponent(Exponent):
domain = NEGATIVE
domain = _NEGATIVE
def f(self, x):
return -Exponent.f(x)
def finv(self, f):
@ -125,7 +125,7 @@ class NegativeExponent(Exponent):
return '-ve'
class Square(Transformation):
domain = POSITIVE
domain = _POSITIVE
def f(self, x):
return x ** 2
def finv(self, x):
@ -138,7 +138,7 @@ class Square(Transformation):
return '+sq'
class Logistic(Transformation):
domain = BOUNDED
domain = _BOUNDED
_instances = []
def __new__(cls, lower=1e-6, upper=1e-6, *args, **kwargs):
if cls._instances: