mirror of
https://github.com/SheffieldML/GPy.git
synced 2026-05-07 19:12:40 +02:00
Datasets.py updates should have been committed before.
This commit is contained in:
Neil Lawrence
parent
3003718ea9
commit
ec1b9190f5
6 changed files with 0 additions and 792 deletions
|
|
@ -1,149 +0,0 @@
|
|||
# Copyright (c) 2014 GPy Authors
|
||||
# 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 stabilise
|
||||
except ImportError:
|
||||
sympy_available=False
|
||||
|
||||
from ..core.mapping import Mapping, Bijective_mapping
|
||||
import numpy as np
|
||||
from scipy.special import gammaln, gamma, erf, erfc, erfcx, polygamma
|
||||
from GPy.util.functions import normcdf, normcdfln, logistic, logisticln
|
||||
from ..core.parameterization import Param
|
||||
|
||||
|
||||
if sympy_available:
|
||||
class Symbolic(Mapping):
|
||||
"""
|
||||
Symbolic likelihood.
|
||||
|
||||
Likelihood where the form of the likelihood is provided by a sympy expression.
|
||||
|
||||
"""
|
||||
def __init__(self, f=None, logZ=None, name='symbolic', param=None, func_modules=[]):
|
||||
|
||||
|
||||
if f is None:
|
||||
raise ValueError, "You must provide an argument for the function."
|
||||
|
||||
self.func_modules = func_modules
|
||||
self.func_modules += [{'gamma':gamma,
|
||||
'gammaln':gammaln,
|
||||
'erf':erf, 'erfc':erfc,
|
||||
'erfcx':erfcx,
|
||||
'polygamma':polygamma,
|
||||
'normcdf':normcdf,
|
||||
'normcdfln':normcdfln,
|
||||
'logistic':logistic,
|
||||
'logisticln':logisticln},
|
||||
'numpy']
|
||||
|
||||
super(Symbolic, self).__init__(gp_link, name=name)
|
||||
self.symbolic['function'] = f
|
||||
|
||||
# pull the variable names out of the symbolic pdf
|
||||
sym_vars = [e for e in f.atoms() if e.is_Symbol]
|
||||
self.symbolic['x'] = [e for e in sym_vars if e.name[:2]=='x_']
|
||||
if not self.symbolic['f']:
|
||||
raise ValueError('No variable x in f().')
|
||||
self.symbolic['theta'] = sorted([e for e in sym_vars if not e.name[:2]=='x_'],key=lambda e:e.name)
|
||||
|
||||
theta_names = [theta.name for theta in self.symbolic['theta']
|
||||
|
||||
# These are all the arguments need to compute the mapping.
|
||||
self.arg_list = self.symbolic['x'] + self.symbolic['theta']
|
||||
|
||||
# these are arguments for computing derivatives.
|
||||
derivative_arguments = self.arg_list
|
||||
|
||||
# Do symbolic work to compute derivatives.
|
||||
self.symbolic['derivatives'] = {theta.name : stabilise(sym.diff(f,theta)) for theta in derivative_arguments}
|
||||
|
||||
# Add parameters to the model.
|
||||
for theta in self._sym_theta:
|
||||
val = 1.0
|
||||
# TODO: need to decide how to handle user passing values for the se parameter vectors.
|
||||
if param is not None:
|
||||
if param.has_key(theta.name):
|
||||
val = param[theta.name]
|
||||
setattr(self, theta.name, Param(theta.name, val, None))
|
||||
self.add_parameters(getattr(self, theta.name))
|
||||
|
||||
# initialise code arguments
|
||||
self._arguments = {}
|
||||
|
||||
# generate the code for the pdf and derivatives
|
||||
self._gen_code()
|
||||
|
||||
def _gen_code(self):
|
||||
"""Generate the code from the symbolic parts that will be used for likleihod computation."""
|
||||
|
||||
self.code = GPy.util.function.gen_code(self.symbolic)
|
||||
|
||||
def parameters_changed(self):
|
||||
# do all the precomputation codes.
|
||||
for variable, code in self.code['precompute'].items():
|
||||
self.setattr(variable, eval(code, self.namespace))
|
||||
|
||||
def update_gradients(self, grads):
|
||||
"""
|
||||
"""
|
||||
for param, code in self.code['derivatives'].items():
|
||||
self.getattr(param).setattr('gradient',
|
||||
eval(code, self.namespace))
|
||||
pass
|
||||
|
||||
def _arguments_update(self, x):
|
||||
"""Set up argument lists for the derivatives."""
|
||||
# If we do make use of Theano, then at this point we would
|
||||
# need to do a lot of precomputation to ensure that the
|
||||
# likelihoods and gradients are computed together, then check
|
||||
# for parameter changes before updating.
|
||||
for i, fvar in enumerate(self._sym_x):
|
||||
self._arguments[fvar.name] = x[:, i]
|
||||
for theta in self._sym_theta:
|
||||
self._arguments[theta.name] = np.asarray(getattr(self, theta.name))
|
||||
|
||||
def f(self, x):
|
||||
"""
|
||||
Likelihood function given inverse link of f.
|
||||
|
||||
:param inv_link_f: inverse link of latent variables.
|
||||
:type inv_link_f: Nx1 array
|
||||
:param y: data
|
||||
:type y: Nx1 array
|
||||
:param Y_metadata: Y_metadata which is not used in student t distribution
|
||||
:returns: likelihood evaluated for this point
|
||||
:rtype: float
|
||||
"""
|
||||
self._arguments_update(inv_link_f, y)
|
||||
return self._f_function(x)
|
||||
|
||||
|
||||
def df_dX(self, X):
|
||||
"""
|
||||
Gradient of log likelihood with respect to the inverse link function.
|
||||
|
||||
:param inv_inv_link_f: latent variables (inverse link of f)
|
||||
:type inv_inv_link_f: Nx1 array
|
||||
:param y: data
|
||||
:type y: Nx1 array
|
||||
:param Y_metadata: Y_metadata
|
||||
:returns: gradient of likelihood with respect to each point.
|
||||
:rtype: Nx1 array
|
||||
|
||||
"""
|
||||
self._arguments_update(X)
|
||||
return self._derivative_code['X'](**self._arguments)
|
||||
|
||||
def df_dtheta(self, X):
|
||||
self._arguments_update(X)
|
||||
g = np.zeros((np.atleast_1d(X).shape[0], len(self._sym_theta)))
|
||||
for i, theta in enumerate(self._sym_theta):
|
||||
g[:, i:i+1] = self._derivative_code[theta.name](**self._arguments)
|
||||
return g.sum(0)
|
||||
|
||||
Loading…
Add table
Add a link
Reference in a new issue