apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-12 05:22:38 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Remove symbolic import.

Browse source
This commit is contained in:
Neil Lawrence 2014-10-16 15:45:10 +01:00
parent 4be2ef267d
commit b8a09bfff7
4 changed files with 0 additions and 187 deletions
Download patch file Download diff file Expand all files Collapse all files

47
GPy/likelihoods/negative_binomial.py
View file

@ -1,47 +0,0 @@
# Copyright (c) 2014 The GPy authors (see AUTHORS.txt)
# Licensed under the BSD 3-clause license (see LICENSE.txt)
try:
import sympy as sym
sympy_available=True
from sympy.utilities.lambdify import lambdify
from GPy.util.symbolic import gammaln, logisticln
except ImportError:
sympy_available=False
import numpy as np
import link_functions
from symbolic import Symbolic
from scipy import stats
class Negative_binomial(Symbolic):
"""
Negative binomial
.. math::
p(y_{i}|\pi(f_{i})) = \left(\frac{r}{r+f_i}\right)^r \frac{\Gamma(r+y_i)}{y!\Gamma(r)}\left(\frac{f_i}{r+f_i}\right)^{y_i}
.. Note::
Y takes non zero integer values..
link function should have a positive domain, e.g. log (default).
.. See also::
symbolic.py, for the parent class
"""
def __init__(self, gp_link=None, dispersion=1.0):
parameters = {'dispersion':dispersion}
if gp_link is None:
gp_link = link_functions.Identity()
dispersion = sym.Symbol('dispersion', positive=True, real=True)
y_0 = sym.Symbol('y_0', nonnegative=True, integer=True)
f_0 = sym.Symbol('f_0', positive=True, real=True)
gp_link = link_functions.Log()
log_pdf=dispersion*sym.log(dispersion) - (dispersion+y_0)*sym.log(dispersion+f_0) + gammaln(y_0+dispersion) - gammaln(y_0+1) - gammaln(dispersion) + y_0*sym.log(f_0)
#log_pdf= -(dispersion+y)*logisticln(f-sym.log(dispersion)) + gammaln(y+dispersion) - gammaln(y+1) - gammaln(dispersion) + y*(f-sym.log(dispersion))
super(Negative_binomial, self).__init__(log_pdf=log_pdf, parameters=parameters, gp_link=gp_link, name='Negative_binomial')
# TODO: Check this.
self.log_concave = False

43
GPy/likelihoods/skew_exponential.py
View file

@ -1,43 +0,0 @@
# Copyright (c) 2014 The GPy authors (see AUTHORS.txt)
# Licensed under the BSD 3-clause license (see LICENSE.txt)
import sympy as sym
#from GPy.util.symbolic import normcdfln
import numpy as np
import link_functions
from symbolic import Symbolic
from scipy import stats
class Skew_exponential(Symbolic):
"""
Negative binomial
.. math::
.. Note::
Y takes real values.
link function is identity
.. See also::
symbolic.py, for the parent class
"""
def __init__(self, gp_link=None, shape=1.0, scale=1.0):
parameters={'scale':scale, 'shape':shape}
if gp_link is None:
gp_link = link_functions.Identity()
#func_modules = [{'exp':clip_exp}]
scale = sym.Symbol('scale', positive=True, real=True)
shape = sym.Symbol('shape', real=True)
y_0 = sym.Symbol('y_0', real=True)
f_0 = sym.Symbol('f_0', real=True)
log_pdf=sym.log(shape)-sym.log(scale)-((y_0-f_0)/scale) + normcdfln(shape*(y_0-f_0)/scale)
super(Skew_exponential, self).__init__(log_pdf=log_pdf, gp_link=gp_link, name='Skew_exponential', parameters=parameters)
# TODO: Check this.
self.log_concave = True

52
GPy/likelihoods/skew_normal.py
View file

@ -1,52 +0,0 @@
# Copyright (c) 2014 The GPy authors (see AUTHORS.txt)
# Licensed under the BSD 3-clause license (see LICENSE.txt)
try:
import sympy as sym
sympy_available=True
from sympy.utilities.lambdify import lambdify
from GPy.util.symbolic import normcdfln, normcdf
except ImportError:
sympy_available=False
import numpy as np
from GPy.util.functions import clip_exp
import link_functions
from symbolic import Symbolic
from scipy import stats
class Skew_normal(Symbolic):
"""
Skew Normal distribution.
.. math::
.. Note::
Y takes real values.
link function is identity
.. See also::
symbolic.py, for the parent class
"""
def __init__(self, gp_link=None, shape=1.0, scale=1.0):
parameters = {'scale': scale, 'shape':shape}
if gp_link is None:
gp_link = link_functions.Identity()
# # this likelihood has severe problems with likelihoods saturating exponentials, so clip_exp is used in place of the true exp as a solution for dealing with the numerics.
# func_modules = [{'exp':clip_exp}]
func_modules = []
scale = sym.Symbol('scale', positive=True, real=True)
shape = sym.Symbol('shape', real=True)
y_0 = sym.Symbol('y_0', real=True)
f_0 = sym.Symbol('f_0', real=True)
log_pdf=-sym.log(scale)-1./2*sym.log(2*sym.pi)-1./2*((y_0-f_0)/scale)**2 + sym.log(2) + normcdfln(shape*(y_0-f_0)/scale)
super(Skew_normal, self).__init__(log_pdf=log_pdf, parameters=parameters, gp_link=gp_link, name='Skew_normal', func_modules=func_modules)
self.log_concave = True

45
GPy/likelihoods/sstudent_t.py
View file

@ -1,45 +0,0 @@
# Copyright (c) 2014 The GPy authors (see AUTHORS.txt)
# Licensed under the BSD 3-clause license (see LICENSE.txt)
import sympy as sym
from sympy.utilities.lambdify import lambdify
# does not exist! JH from GPy.util.symbolic import gammaln
import numpy as np
import link_functions
from symbolic import Symbolic
from scipy import stats
class SstudentT(Symbolic):
"""
Symbolic variant of the Student-t distribution.
.. math::
.. Note::
Y takes real values.
link function is identity
.. See also::
symbolic.py, for the parent class
"""
def __init__(self, gp_link=None, deg_free=5.0, t_scale2=1.0):
parameters = {'deg_free':5.0, 't_scale2':1.0}
if gp_link is None:
gp_link = link_functions.Identity()
# this likelihood has severe problems with likelihoods saturating ...
y_0 = sym.Symbol('y_0', real=True)
f_0 = sym.Symbol('f_0', real=True)
nu = sym.Symbol('nu', positive=True, real=True)
t_scale2 = sym.Symbol('t_scale2', positive=True, real=True)
log_pdf = (gammaln((nu + 1) * 0.5)
- gammaln(nu * 0.5)
- 0.5*sym.log(t_scale2 * nu * sym.pi)
- 0.5*(nu + 1)*sym.log(1 + (1/nu)*(((y_0-f_0)**2)/t_scale2)))
super(SstudentT, self).__init__(log_pdf=log_pdf, parameters=parameters, gp_link=gp_link, name='SstudentT')
self.log_concave = False