apunkt/GPy
1
0
Fork 0
mirror of https://github.com/SheffieldML/GPy.git synced 2026-05-18 13:55:14 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'devel' of github.com:SheffieldML/GPy into devel

Browse source
This commit is contained in:
James Hensman 2013-09-20 14:37:39 +01:00
commit 759cf08016
24 changed files with 278 additions and 158 deletions
Download patch file Download diff file Expand all files Collapse all files

8
GPy/core/fitc.py
View file

@ -11,18 +11,20 @@ from sparse_gp import SparseGP
class FITC(SparseGP): class FITC(SparseGP):
""" """
sparse FITC approximation
Sparse FITC approximation
:param X: inputs :param X: inputs
:type X: np.ndarray (num_data x Q) :type X: np.ndarray (num_data x Q)
:param likelihood: a likelihood instance, containing the observed data :param likelihood: a likelihood instance, containing the observed data
:type likelihood: GPy.likelihood.(Gaussian | EP) :type likelihood: GPy.likelihood.(Gaussian | EP)
:param kernel : the kernel (covariance function). See link kernels :param kernel: the kernel (covariance function). See link kernels
:type kernel: a GPy.kern.kern instance :type kernel: a GPy.kern.kern instance
:param Z: inducing inputs (optional, see note) :param Z: inducing inputs (optional, see note)
:type Z: np.ndarray (M x Q) | None :type Z: np.ndarray (M x Q) | None
:param normalize_(X|Y) : whether to normalize the data before computing (predictions will be in original scales) :param normalize_(X|Y): whether to normalize the data before computing (predictions will be in original scales)
:type normalize_(X|Y): bool :type normalize_(X|Y): bool
""" """
def __init__(self, X, likelihood, kernel, Z, normalize_X=False): def __init__(self, X, likelihood, kernel, Z, normalize_X=False):

4
GPy/core/mapping.py
View file

@ -49,6 +49,7 @@ class Mapping(Parameterized):
def plot(self, plot_limits=None, which_data='all', which_parts='all', resolution=None, levels=20, samples=0, fignum=None, ax=None, fixed_inputs=[], linecol=Tango.colorsHex['darkBlue']): def plot(self, plot_limits=None, which_data='all', which_parts='all', resolution=None, levels=20, samples=0, fignum=None, ax=None, fixed_inputs=[], linecol=Tango.colorsHex['darkBlue']):
""" """
Plot the mapping. Plot the mapping.
Plots the mapping associated with the model. Plots the mapping associated with the model.
@ -79,8 +80,7 @@ class Mapping(Parameterized):
:type fixed_inputs: a list of tuples :type fixed_inputs: a list of tuples
:param linecol: color of line to plot. :param linecol: color of line to plot.
:type linecol: :type linecol:
:param levels: for 2D plotting, the number of contour levels to use :param levels: for 2D plotting, the number of contour levels to use is ax is None, create a new figure
is ax is None, create a new figure
""" """
# TODO include samples # TODO include samples

17
GPy/core/model.py
View file

@ -56,10 +56,11 @@ class Model(Parameterized):
def set_prior(self, regexp, what): def set_prior(self, regexp, what):
""" """
Sets priors on the model parameters. Sets priors on the model parameters.
Notes **Notes**
-----
Asserts that the prior is suitable for the constraint. If the Asserts that the prior is suitable for the constraint. If the
wrong constraint is in place, an error is raised. If no wrong constraint is in place, an error is raised. If no
constraint is in place, one is added (warning printed). constraint is in place, one is added (warning printed).
@ -185,8 +186,8 @@ class Model(Parameterized):
be handled silently. If _all_ runs fail, the model is reset to the be handled silently. If _all_ runs fail, the model is reset to the
existing parameter values. existing parameter values.
Notes **Notes**
-----
:param num_restarts: number of restarts to use (default 10) :param num_restarts: number of restarts to use (default 10)
:type num_restarts: int :type num_restarts: int
:param robust: whether to handle exceptions silently or not (default False) :param robust: whether to handle exceptions silently or not (default False)
@ -195,7 +196,9 @@ class Model(Parameterized):
:type parallel: bool :type parallel: bool
:param num_processes: number of workers in the multiprocessing pool :param num_processes: number of workers in the multiprocessing pool
:type numprocesses: int :type numprocesses: int
**kwargs are passed to the optimizer. They can be:
\*\*kwargs are passed to the optimizer. They can be:
:param max_f_eval: maximum number of function evaluations :param max_f_eval: maximum number of function evaluations
:type max_f_eval: int :type max_f_eval: int
:param max_iters: maximum number of iterations :param max_iters: maximum number of iterations
@ -203,9 +206,7 @@ class Model(Parameterized):
:param messages: whether to display during optimisation :param messages: whether to display during optimisation
:type messages: bool :type messages: bool
..Note: If num_processes is None, the number of workes in the multiprocessing pool is automatically .. note:: If num_processes is None, the number of workes in the multiprocessing pool is automatically set to the number of processors on the current machine.
set to the number of processors on the current machine.
""" """
initial_parameters = self._get_params_transformed() initial_parameters = self._get_params_transformed()

12
GPy/core/parameterized.py
View file

@ -231,17 +231,19 @@ class Parameterized(object):
def constrain_fixed(self, regexp, value=None): def constrain_fixed(self, regexp, value=None):
""" """
Arguments
---------
:param regexp: which parameters need to be fixed. :param regexp: which parameters need to be fixed.
:type regexp: ndarray(dtype=int) or regular expression object or string :type regexp: ndarray(dtype=int) or regular expression object or string
:param value: the vlaue to fix the parameters to. If the value is not specified, :param value: the vlaue to fix the parameters to. If the value is not specified,
the parameter is fixed to the current value the parameter is fixed to the current value
:type value: float :type value: float
Notes
----- **Notes**
Fixing a parameter which is tied to another, or constrained in some way will result in an error. Fixing a parameter which is tied to another, or constrained in some way will result in an error.
To fix multiple parameters to the same value, simply pass a regular expression which matches both parameter names, or pass both of the indexes
To fix multiple parameters to the same value, simply pass a regular expression which matches both parameter names, or pass both of the indexes.
""" """
matches = self.grep_param_names(regexp) matches = self.grep_param_names(regexp)
overlap = set(matches).intersection(set(self.all_constrained_indices())) overlap = set(matches).intersection(set(self.all_constrained_indices()))

12
GPy/core/sparse_gp.py
View file

@ -16,16 +16,17 @@ class SparseGP(GPBase):
:type X: np.ndarray (num_data x input_dim) :type X: np.ndarray (num_data x input_dim)
:param likelihood: a likelihood instance, containing the observed data :param likelihood: a likelihood instance, containing the observed data
:type likelihood: GPy.likelihood.(Gaussian | EP | Laplace) :type likelihood: GPy.likelihood.(Gaussian | EP | Laplace)
:param kernel : the kernel (covariance function). See link kernels :param kernel: the kernel (covariance function). See link kernels
:type kernel: a GPy.kern.kern instance :type kernel: a GPy.kern.kern instance
:param X_variance: The uncertainty in the measurements of X (Gaussian variance) :param X_variance: The uncertainty in the measurements of X (Gaussian variance)
:type X_variance: np.ndarray (num_data x input_dim) | None :type X_variance: np.ndarray (num_data x input_dim) | None
:param Z: inducing inputs (optional, see note) :param Z: inducing inputs (optional, see note)
:type Z: np.ndarray (num_inducing x input_dim) | None :type Z: np.ndarray (num_inducing x input_dim) | None
:param num_inducing : Number of inducing points (optional, default 10. Ignored if Z is not None) :param num_inducing: Number of inducing points (optional, default 10. Ignored if Z is not None)
:type num_inducing: int :type num_inducing: int
:param normalize_(X|Y) : whether to normalize the data before computing (predictions will be in original scales) :param normalize_(X|Y): whether to normalize the data before computing (predictions will be in original scales)
:type normalize_(X|Y): bool :type normalize_(X|Y): bool
""" """
def __init__(self, X, likelihood, kernel, Z, X_variance=None, normalize_X=False): def __init__(self, X, likelihood, kernel, Z, X_variance=None, normalize_X=False):
@ -306,10 +307,11 @@ class SparseGP(GPBase):
def predict(self, Xnew, X_variance_new=None, which_parts='all', full_cov=False): def predict(self, Xnew, X_variance_new=None, which_parts='all', full_cov=False):
""" """
Predict the function(s) at the new point(s) Xnew. Predict the function(s) at the new point(s) Xnew.
Arguments **Arguments**
---------
:param Xnew: The points at which to make a prediction :param Xnew: The points at which to make a prediction
:type Xnew: np.ndarray, Nnew x self.input_dim :type Xnew: np.ndarray, Nnew x self.input_dim
:param X_variance_new: The uncertainty in the prediction points :param X_variance_new: The uncertainty in the prediction points

14
GPy/core/svigp.py
View file

@ -14,6 +14,7 @@ import sys
class SVIGP(GPBase): class SVIGP(GPBase):
""" """
Stochastic Variational inference in a Gaussian Process Stochastic Variational inference in a Gaussian Process
:param X: inputs :param X: inputs
@ -22,25 +23,26 @@ class SVIGP(GPBase):
:type Y: np.ndarray of observations (N x D) :type Y: np.ndarray of observations (N x D)
:param batchsize: the size of a h :param batchsize: the size of a h
Additional kwargs are used as for a sparse GP. They include Additional kwargs are used as for a sparse GP. They include:
:param q_u: canonical parameters of the distribution squasehd into a 1D array :param q_u: canonical parameters of the distribution squasehd into a 1D array
:type q_u: np.ndarray :type q_u: np.ndarray
:param M : Number of inducing points (optional, default 10. Ignored if Z is not None) :param M: Number of inducing points (optional, default 10. Ignored if Z is not None)
:type M: int :type M: int
:param kernel : the kernel/covariance function. See link kernels :param kernel: the kernel/covariance function. See link kernels
:type kernel: a GPy kernel :type kernel: a GPy kernel
:param Z: inducing inputs (optional, see note) :param Z: inducing inputs (optional, see note)
:type Z: np.ndarray (M x Q) | None :type Z: np.ndarray (M x Q) | None
:param X_uncertainty: The uncertainty in the measurements of X (Gaussian variance) :param X_uncertainty: The uncertainty in the measurements of X (Gaussian variance)
:type X_uncertainty: np.ndarray (N x Q) | None :type X_uncertainty: np.ndarray (N x Q) | None
:param Zslices: slices for the inducing inputs (see slicing TODO: link) :param Zslices: slices for the inducing inputs (see slicing TODO: link)
:param M : Number of inducing points (optional, default 10. Ignored if Z is not None) :param M: Number of inducing points (optional, default 10. Ignored if Z is not None)
:type M: int :type M: int
:param beta: noise precision. TODO> ignore beta if doing EP :param beta: noise precision. TODO: ignore beta if doing EP
:type beta: float :type beta: float
:param normalize_(X|Y) : whether to normalize the data before computing (predictions will be in original scales) :param normalize_(X|Y): whether to normalize the data before computing (predictions will be in original scales)
:type normalize_(X|Y): bool :type normalize_(X|Y): bool
""" """

31
GPy/examples/classification.py
View file

@ -11,13 +11,15 @@ import GPy
default_seed = 10000 default_seed = 10000
def crescent_data(seed=default_seed, kernel=None): # FIXME def crescent_data(seed=default_seed, kernel=None): # FIXME
"""Run a Gaussian process classification on the crescent data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood. """
Run a Gaussian process classification on the crescent data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood.
:param model_type: type of model to fit ['Full', 'FITC', 'DTC']. :param model_type: type of model to fit ['Full', 'FITC', 'DTC'].
:param seed : seed value for data generation. :param seed: seed value for data generation.
:type seed: int :type seed: int
:param inducing : number of inducing variables (only used for 'FITC' or 'DTC'). :param inducing: number of inducing variables (only used for 'FITC' or 'DTC').
:type inducing: int :type inducing: int
""" """
data = GPy.util.datasets.crescent_data(seed=seed) data = GPy.util.datasets.crescent_data(seed=seed)
@ -35,6 +37,7 @@ def crescent_data(seed=default_seed, kernel=None): # FIXME
def oil(num_inducing=50, max_iters=100, kernel=None): def oil(num_inducing=50, max_iters=100, kernel=None):
""" """
Run a Gaussian process classification on the three phase oil data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood. Run a Gaussian process classification on the three phase oil data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood.
""" """
data = GPy.util.datasets.oil() data = GPy.util.datasets.oil()
X = data['X'] X = data['X']
@ -64,8 +67,10 @@ def oil(num_inducing=50, max_iters=100, kernel=None):
def toy_linear_1d_classification(seed=default_seed): def toy_linear_1d_classification(seed=default_seed):
""" """
Simple 1D classification example Simple 1D classification example
:param seed : seed value for data generation (default is 4).
:param seed: seed value for data generation (default is 4).
:type seed: int :type seed: int
""" """
data = GPy.util.datasets.toy_linear_1d_classification(seed=seed) data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)
@ -92,8 +97,10 @@ def toy_linear_1d_classification(seed=default_seed):
def sparse_toy_linear_1d_classification(num_inducing=10,seed=default_seed): def sparse_toy_linear_1d_classification(num_inducing=10,seed=default_seed):
""" """
Sparse 1D classification example Sparse 1D classification example
:param seed : seed value for data generation (default is 4).
:param seed: seed value for data generation (default is 4).
:type seed: int :type seed: int
""" """
data = GPy.util.datasets.toy_linear_1d_classification(seed=seed) data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)
@ -123,10 +130,11 @@ def sparse_crescent_data(num_inducing=10, seed=default_seed, kernel=None):
Run a Gaussian process classification with DTC approxiamtion on the crescent data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood. Run a Gaussian process classification with DTC approxiamtion on the crescent data. The demonstration calls the basic GP classification model and uses EP to approximate the likelihood.
:param model_type: type of model to fit ['Full', 'FITC', 'DTC']. :param model_type: type of model to fit ['Full', 'FITC', 'DTC'].
:param seed : seed value for data generation. :param seed: seed value for data generation.
:type seed: int :type seed: int
:param inducing : number of inducing variables (only used for 'FITC' or 'DTC'). :param inducing: number of inducing variables (only used for 'FITC' or 'DTC').
:type inducing: int :type inducing: int
""" """
data = GPy.util.datasets.crescent_data(seed=seed) data = GPy.util.datasets.crescent_data(seed=seed)
@ -147,10 +155,11 @@ def FITC_crescent_data(num_inducing=10, seed=default_seed):
Run a Gaussian process classification with FITC approximation on the crescent data. The demonstration uses EP to approximate the likelihood. Run a Gaussian process classification with FITC approximation on the crescent data. The demonstration uses EP to approximate the likelihood.
:param model_type: type of model to fit ['Full', 'FITC', 'DTC']. :param model_type: type of model to fit ['Full', 'FITC', 'DTC'].
:param seed : seed value for data generation. :param seed: seed value for data generation.
:type seed: int :type seed: int
:param inducing : number of inducing variables (only used for 'FITC' or 'DTC'). :param inducing: number of inducing variables (only used for 'FITC' or 'DTC').
:type num_inducing: int :type num_inducing: int
""" """
data = GPy.util.datasets.crescent_data(seed=seed) data = GPy.util.datasets.crescent_data(seed=seed)
@ -171,8 +180,10 @@ def FITC_crescent_data(num_inducing=10, seed=default_seed):
def toy_heaviside(seed=default_seed): def toy_heaviside(seed=default_seed):
""" """
Simple 1D classification example using a heavy side gp transformation Simple 1D classification example using a heavy side gp transformation
:param seed : seed value for data generation (default is 4).
:param seed: seed value for data generation (default is 4).
:type seed: int :type seed: int
""" """
data = GPy.util.datasets.toy_linear_1d_classification(seed=seed) data = GPy.util.datasets.toy_linear_1d_classification(seed=seed)

19
GPy/inference/conjugate_gradient_descent.py
View file

@ -233,7 +233,7 @@ class CGD(Async_Optimize):
""" """
opt_async(self, f, df, x0, callback, update_rule=FletcherReeves, opt_async(self, f, df, x0, callback, update_rule=FletcherReeves,
messages=0, maxiter=5e3, max_f_eval=15e3, gtol=1e-6, messages=0, maxiter=5e3, max_f_eval=15e3, gtol=1e-6,
report_every=10, *args, **kwargs) report_every=10, \*args, \*\*kwargs)
callback gets called every `report_every` iterations callback gets called every `report_every` iterations
@ -244,16 +244,14 @@ class CGD(Async_Optimize):
f, and df will be called with f, and df will be called with
f(xi, *args, **kwargs) f(xi, \*args, \*\*kwargs)
df(xi, *args, **kwargs) df(xi, \*args, \*\*kwargs)
**returns** **Returns:**
-----------
Started `Process` object, optimizing asynchronously Started `Process` object, optimizing asynchronously
**calls** **Calls:**
---------
callback(x_opt, f_opt, g_opt, iteration, function_calls, gradient_calls, status_message) callback(x_opt, f_opt, g_opt, iteration, function_calls, gradient_calls, status_message)
@ -265,7 +263,7 @@ class CGD(Async_Optimize):
""" """
opt(self, f, df, x0, callback=None, update_rule=FletcherReeves, opt(self, f, df, x0, callback=None, update_rule=FletcherReeves,
messages=0, maxiter=5e3, max_f_eval=15e3, gtol=1e-6, messages=0, maxiter=5e3, max_f_eval=15e3, gtol=1e-6,
report_every=10, *args, **kwargs) report_every=10, \*args, \*\*kwargs)
Minimize f, calling callback every `report_every` iterations with following syntax: Minimize f, calling callback every `report_every` iterations with following syntax:
@ -276,11 +274,10 @@ class CGD(Async_Optimize):
f, and df will be called with f, and df will be called with
f(xi, *args, **kwargs) f(xi, \*args, \*\*kwargs)
df(xi, *args, **kwargs) df(xi, \*args, \*\*kwargs)
**returns** **returns**
---------
x_opt, f_opt, g_opt, iteration, function_calls, gradient_calls, status_message x_opt, f_opt, g_opt, iteration, function_calls, gradient_calls, status_message

66
GPy/kern/constructors.py
View file

@ -17,6 +17,7 @@ def rbf_inv(input_dim,variance=1., inv_lengthscale=None,ARD=False):
:type lengthscale: float :type lengthscale: float
:param ARD: Auto Relevance Determination (one lengthscale per dimension) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.rbf_inv.RBFInv(input_dim,variance,inv_lengthscale,ARD) part = parts.rbf_inv.RBFInv(input_dim,variance,inv_lengthscale,ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -33,6 +34,7 @@ def rbf(input_dim,variance=1., lengthscale=None,ARD=False):
:type lengthscale: float :type lengthscale: float
:param ARD: Auto Relevance Determination (one lengthscale per dimension) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.rbf.RBF(input_dim,variance,lengthscale,ARD) part = parts.rbf.RBF(input_dim,variance,lengthscale,ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -41,11 +43,13 @@ def linear(input_dim,variances=None,ARD=False):
""" """
Construct a linear kernel. Construct a linear kernel.
Arguments :param input_dim: dimensionality of the kernel, obligatory
--------- :type input_dim: int
input_dimD (int), obligatory :param variances:
variances (np.ndarray) :type variances: np.ndarray
ARD (boolean) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean
""" """
part = parts.linear.Linear(input_dim,variances,ARD) part = parts.linear.Linear(input_dim,variances,ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -64,12 +68,14 @@ def mlp(input_dim,variance=1., weight_variance=None,bias_variance=100.,ARD=False
:type bias_variance: float :type bias_variance: float
:param ARD: Auto Relevance Determination (allows for ARD version of covariance) :param ARD: Auto Relevance Determination (allows for ARD version of covariance)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.mlp.MLP(input_dim,variance,weight_variance,bias_variance,ARD) part = parts.mlp.MLP(input_dim,variance,weight_variance,bias_variance,ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
def gibbs(input_dim,variance=1., mapping=None): def gibbs(input_dim,variance=1., mapping=None):
""" """
Gibbs and MacKay non-stationary covariance function. Gibbs and MacKay non-stationary covariance function.
.. math:: .. math::
@ -124,6 +130,7 @@ def poly(input_dim,variance=1., weight_variance=None,bias_variance=1.,degree=2,
:type degree: int :type degree: int
:param ARD: Auto Relevance Determination (allows for ARD version of covariance) :param ARD: Auto Relevance Determination (allows for ARD version of covariance)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.poly.POLY(input_dim,variance,weight_variance,bias_variance,degree,ARD) part = parts.poly.POLY(input_dim,variance,weight_variance,bias_variance,degree,ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -132,10 +139,11 @@ def white(input_dim,variance=1.):
""" """
Construct a white kernel. Construct a white kernel.
Arguments :param input_dim: dimensionality of the kernel, obligatory
--------- :type input_dim: int
input_dimD (int), obligatory :param variance: the variance of the kernel
variance (float) :type variance: float
""" """
part = parts.white.White(input_dim,variance) part = parts.white.White(input_dim,variance)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -153,6 +161,7 @@ def exponential(input_dim,variance=1., lengthscale=None, ARD=False):
:type lengthscale: float :type lengthscale: float
:param ARD: Auto Relevance Determination (one lengthscale per dimension) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.exponential.Exponential(input_dim,variance, lengthscale, ARD) part = parts.exponential.Exponential(input_dim,variance, lengthscale, ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -169,6 +178,7 @@ def Matern32(input_dim,variance=1., lengthscale=None, ARD=False):
:type lengthscale: float :type lengthscale: float
:param ARD: Auto Relevance Determination (one lengthscale per dimension) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.Matern32.Matern32(input_dim,variance, lengthscale, ARD) part = parts.Matern32.Matern32(input_dim,variance, lengthscale, ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -185,6 +195,7 @@ def Matern52(input_dim, variance=1., lengthscale=None, ARD=False):
:type lengthscale: float :type lengthscale: float
:param ARD: Auto Relevance Determination (one lengthscale per dimension) :param ARD: Auto Relevance Determination (one lengthscale per dimension)
:type ARD: Boolean :type ARD: Boolean
""" """
part = parts.Matern52.Matern52(input_dim, variance, lengthscale, ARD) part = parts.Matern52.Matern52(input_dim, variance, lengthscale, ARD)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -193,10 +204,11 @@ def bias(input_dim, variance=1.):
""" """
Construct a bias kernel. Construct a bias kernel.
Arguments :param input_dim: dimensionality of the kernel, obligatory
--------- :type input_dim: int
input_dim (int), obligatory :param variance: the variance of the kernel
variance (float) :type variance: float
""" """
part = parts.bias.Bias(input_dim, variance) part = parts.bias.Bias(input_dim, variance)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -204,10 +216,15 @@ def bias(input_dim, variance=1.):
def finite_dimensional(input_dim, F, G, variances=1., weights=None): def finite_dimensional(input_dim, F, G, variances=1., weights=None):
""" """
Construct a finite dimensional kernel. Construct a finite dimensional kernel.
input_dim: int - the number of input dimensions
F: np.array of functions with shape (n,) - the n basis functions :param input_dim: the number of input dimensions
G: np.array with shape (n,n) - the Gram matrix associated to F :type input_dim: int
variances : np.ndarray with shape (n,) :param F: np.array of functions with shape (n,) - the n basis functions
:type F: np.array
:param G: np.array with shape (n,n) - the Gram matrix associated to F
:type G: np.array
:param variances: np.ndarray with shape (n,)
:type: np.ndarray
""" """
part = parts.finite_dimensional.FiniteDimensional(input_dim, F, G, variances, weights) part = parts.finite_dimensional.FiniteDimensional(input_dim, F, G, variances, weights)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -220,6 +237,7 @@ def spline(input_dim, variance=1.):
:type input_dim: int :type input_dim: int
:param variance: the variance of the kernel :param variance: the variance of the kernel
:type variance: float :type variance: float
""" """
part = parts.spline.Spline(input_dim, variance) part = parts.spline.Spline(input_dim, variance)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -232,6 +250,7 @@ def Brownian(input_dim, variance=1.):
:type input_dim: int :type input_dim: int
:param variance: the variance of the kernel :param variance: the variance of the kernel
:type variance: float :type variance: float
""" """
part = parts.Brownian.Brownian(input_dim, variance) part = parts.Brownian.Brownian(input_dim, variance)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -285,6 +304,7 @@ def periodic_exponential(input_dim=1, variance=1., lengthscale=None, period=2 *
:type period: float :type period: float
:param n_freq: the number of frequencies considered for the periodic subspace :param n_freq: the number of frequencies considered for the periodic subspace
:type n_freq: int :type n_freq: int
""" """
part = parts.periodic_exponential.PeriodicExponential(input_dim, variance, lengthscale, period, n_freq, lower, upper) part = parts.periodic_exponential.PeriodicExponential(input_dim, variance, lengthscale, period, n_freq, lower, upper)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -303,6 +323,7 @@ def periodic_Matern32(input_dim, variance=1., lengthscale=None, period=2 * np.pi
:type period: float :type period: float
:param n_freq: the number of frequencies considered for the periodic subspace :param n_freq: the number of frequencies considered for the periodic subspace
:type n_freq: int :type n_freq: int
""" """
part = parts.periodic_Matern32.PeriodicMatern32(input_dim, variance, lengthscale, period, n_freq, lower, upper) part = parts.periodic_Matern32.PeriodicMatern32(input_dim, variance, lengthscale, period, n_freq, lower, upper)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -321,6 +342,7 @@ def periodic_Matern52(input_dim, variance=1., lengthscale=None, period=2 * np.pi
:type period: float :type period: float
:param n_freq: the number of frequencies considered for the periodic subspace :param n_freq: the number of frequencies considered for the periodic subspace
:type n_freq: int :type n_freq: int
""" """
part = parts.periodic_Matern52.PeriodicMatern52(input_dim, variance, lengthscale, period, n_freq, lower, upper) part = parts.periodic_Matern52.PeriodicMatern52(input_dim, variance, lengthscale, period, n_freq, lower, upper)
return kern(input_dim, [part]) return kern(input_dim, [part])
@ -334,6 +356,7 @@ def prod(k1,k2,tensor=False):
:param tensor: The kernels are either multiply as functions defined on the same input space (default) or on the product of the input spaces :param tensor: The kernels are either multiply as functions defined on the same input space (default) or on the product of the input spaces
:type tensor: Boolean :type tensor: Boolean
:rtype: kernel object :rtype: kernel object
""" """
part = parts.prod.Prod(k1, k2, tensor) part = parts.prod.Prod(k1, k2, tensor)
return kern(part.input_dim, [part]) return kern(part.input_dim, [part])
@ -349,10 +372,12 @@ def symmetric(k):
def coregionalize(num_outputs,W_columns=1, W=None, kappa=None): def coregionalize(num_outputs,W_columns=1, W=None, kappa=None):
""" """
Coregionlization matrix B, of the form: Coregionlization matrix B, of the form:
.. math:: .. math::
\mathbf{B} = \mathbf{W}\mathbf{W}^\top + kappa \mathbf{I} \mathbf{B} = \mathbf{W}\mathbf{W}^\top + kappa \mathbf{I}
An intrinsic/linear coregionalization kernel of the form An intrinsic/linear coregionalization kernel of the form:
.. math:: .. math::
k_2(x, y)=\mathbf{B} k(x, y) k_2(x, y)=\mathbf{B} k(x, y)
@ -422,7 +447,7 @@ def independent_outputs(k):
def hierarchical(k): def hierarchical(k):
""" """
TODO THis can't be right! Construct a kernel with independent outputs from an existing kernel TODO This can't be right! Construct a kernel with independent outputs from an existing kernel
""" """
# for sl in k.input_slices: # for sl in k.input_slices:
# assert (sl.start is None) and (sl.stop is None), "cannot adjust input slices! (TODO)" # assert (sl.start is None) and (sl.stop is None), "cannot adjust input slices! (TODO)"
@ -440,7 +465,8 @@ def build_lcm(input_dim, num_outputs, kernel_list = [], W_columns=1,W=None,kappa
:param W_columns: number tuples of the corregionalization parameters 'coregion_W' :param W_columns: number tuples of the corregionalization parameters 'coregion_W'
:type W_columns: integer :type W_columns: integer
..Note the kernels dimensionality is overwritten to fit input_dim ..note the kernels dimensionality is overwritten to fit input_dim
""" """
for k in kernel_list: for k in kernel_list:

11
GPy/kern/kern.py
View file

@ -78,13 +78,15 @@ class kern(Parameterized):
def plot_ARD(self, fignum=None, ax=None, title='', legend=False): def plot_ARD(self, fignum=None, ax=None, title='', legend=False):
"""If an ARD kernel is present, it bar-plots the ARD parameters, """If an ARD kernel is present, it bar-plots the ARD parameters.
:param fignum: figure number of the plot :param fignum: figure number of the plot
:param ax: matplotlib axis to plot on :param ax: matplotlib axis to plot on
:param title: :param title:
title of the plot, title of the plot,
pass '' to not print a title pass '' to not print a title
pass None for a generic title pass None for a generic title
""" """
if ax is None: if ax is None:
fig = pb.figure(fignum) fig = pb.figure(fignum)
@ -175,8 +177,10 @@ class kern(Parameterized):
def add(self, other, tensor=False): def add(self, other, tensor=False):
""" """
Add another kernel to this one. Both kernels are defined on the same _space_ Add another kernel to this one. Both kernels are defined on the same _space_
:param other: the other kernel to be added :param other: the other kernel to be added
:type other: GPy.kern :type other: GPy.kern
""" """
if tensor: if tensor:
D = self.input_dim + other.input_dim D = self.input_dim + other.input_dim
@ -223,6 +227,7 @@ class kern(Parameterized):
:type other: GPy.kern :type other: GPy.kern
:param tensor: whether or not to use the tensor space (default is false). :param tensor: whether or not to use the tensor space (default is false).
:type tensor: bool :type tensor: bool
""" """
K1 = self.copy() K1 = self.copy()
K2 = other.copy() K2 = other.copy()
@ -321,6 +326,7 @@ class kern(Parameterized):
:type X: np.ndarray (num_samples x input_dim) :type X: np.ndarray (num_samples x input_dim)
:param X2: Observed data inputs (optional, defaults to X) :param X2: Observed data inputs (optional, defaults to X)
:type X2: np.ndarray (num_inducing x input_dim) :type X2: np.ndarray (num_inducing x input_dim)
""" """
assert X.shape[1] == self.input_dim assert X.shape[1] == self.input_dim
target = np.zeros(self.num_params) target = np.zeros(self.num_params)
@ -340,6 +346,7 @@ class kern(Parameterized):
:type X: np.ndarray (num_samples x input_dim) :type X: np.ndarray (num_samples x input_dim)
:param X2: Observed data inputs (optional, defaults to X) :param X2: Observed data inputs (optional, defaults to X)
:type X2: np.ndarray (num_inducing x input_dim)""" :type X2: np.ndarray (num_inducing x input_dim)"""
target = np.zeros_like(X) target = np.zeros_like(X)
if X2 is None: if X2 is None:
[p.dK_dX(dL_dK, X[:, i_s], None, target[:, i_s]) for p, i_s in zip(self.parts, self.input_slices)] [p.dK_dX(dL_dK, X[:, i_s], None, target[:, i_s]) for p, i_s in zip(self.parts, self.input_slices)]
@ -413,6 +420,7 @@ class kern(Parameterized):
:param Z: np.ndarray of inducing inputs (num_inducing x input_dim) :param Z: np.ndarray of inducing inputs (num_inducing x input_dim)
:param mu, S: np.ndarrays of means and variances (each num_samples x input_dim) :param mu, S: np.ndarrays of means and variances (each num_samples x input_dim)
:returns psi2: np.ndarray (num_samples,num_inducing,num_inducing) :returns psi2: np.ndarray (num_samples,num_inducing,num_inducing)
""" """
target = np.zeros((mu.shape[0], Z.shape[0], Z.shape[0])) target = np.zeros((mu.shape[0], Z.shape[0], Z.shape[0]))
[p.psi2(Z[:, i_s], mu[:, i_s], S[:, i_s], target) for p, i_s in zip(self.parts, self.input_slices)] [p.psi2(Z[:, i_s], mu[:, i_s], S[:, i_s], target) for p, i_s in zip(self.parts, self.input_slices)]
@ -657,6 +665,7 @@ def kern_test(kern, X=None, X2=None, verbose=False):
:type X: ndarray :type X: ndarray
:param X2: X2 input values to test the covariance function. :param X2: X2 input values to test the covariance function.
:type X2: ndarray :type X2: ndarray
""" """
pass_checks = True pass_checks = True
if X==None: if X==None:

8
GPy/kern/parts/coregionalize.py
View file

@ -11,12 +11,14 @@ class Coregionalize(Kernpart):
""" """
Covariance function for intrinsic/linear coregionalization models Covariance function for intrinsic/linear coregionalization models
This covariance has the form This covariance has the form:
.. math:: .. math::
\mathbf{B} = \mathbf{W}\mathbf{W}^\top + kappa \mathbf{I} \mathbf{B} = \mathbf{W}\mathbf{W}^\top + kappa \mathbf{I}
An intrinsic/linear coregionalization covariance function of the form An intrinsic/linear coregionalization covariance function of the form:
.. math:: .. math::
k_2(x, y)=\mathbf{B} k(x, y) k_2(x, y)=\mathbf{B} k(x, y)
it is obtained as the tensor product between a covariance function it is obtained as the tensor product between a covariance function
@ -31,7 +33,7 @@ class Coregionalize(Kernpart):
:param kappa: a vector which allows the outputs to behave independently :param kappa: a vector which allows the outputs to behave independently
:type kappa: numpy array of dimensionality (num_outputs,) :type kappa: numpy array of dimensionality (num_outputs,)
.. Note: see coregionalization examples in GPy.examples.regression for some usage. .. note: see coregionalization examples in GPy.examples.regression for some usage.
""" """
def __init__(self,num_outputs,W_columns=1, W=None, kappa=None): def __init__(self,num_outputs,W_columns=1, W=None, kappa=None):
self.input_dim = 1 self.input_dim = 1

7
GPy/kern/parts/mlp.py
View file

@ -7,11 +7,13 @@ four_over_tau = 2./np.pi
class MLP(Kernpart): class MLP(Kernpart):
""" """
multi layer perceptron kernel (also known as arc sine kernel or neural network kernel)
Multi layer perceptron kernel (also known as arc sine kernel or neural network kernel)
.. math:: .. math::
k(x,y) = \sigma^2 \frac{2}{\pi} \text{asin} \left(\frac{\sigma_w^2 x^\top y+\sigma_b^2}{\sqrt{\sigma_w^2x^\top x + \sigma_b^2 + 1}\sqrt{\sigma_w^2 y^\top y \sigma_b^2 +1}} \right) k(x,y) = \\sigma^{2}\\frac{2}{\\pi } \\text{asin} \\left ( \\frac{ \\sigma_w^2 x^\\top y+\\sigma_b^2}{\\sqrt{\\sigma_w^2x^\\top x + \\sigma_b^2 + 1}\\sqrt{\\sigma_w^2 y^\\top y \\sigma_b^2 +1}} \\right )
:param input_dim: the number of input dimensions :param input_dim: the number of input dimensions
:type input_dim: int :type input_dim: int
@ -24,6 +26,7 @@ class MLP(Kernpart):
:type ARD: Boolean :type ARD: Boolean
:rtype: Kernpart object :rtype: Kernpart object
""" """
def __init__(self, input_dim, variance=1., weight_variance=None, bias_variance=100., ARD=False): def __init__(self, input_dim, variance=1., weight_variance=None, bias_variance=100., ARD=False):

17
GPy/kern/parts/poly.py
View file

@ -7,19 +7,20 @@ four_over_tau = 2./np.pi
class POLY(Kernpart): class POLY(Kernpart):
""" """
polynomial kernel parameter initialisation. Included for completeness, but generally not recommended, is the polynomial kernel,
.. math::
k(x, y) = \sigma^2*(\sigma_w^2 x'y+\sigma_b^b)^d
The kernel parameters are \sigma^2 (variance), \sigma^2_w Polynomial kernel parameter initialisation. Included for completeness, but generally not recommended, is the polynomial kernel:
(weight_variance), \sigma^2_b (bias_variance) and d
.. math::
k(x, y) = \sigma^2\*(\sigma_w^2 x'y+\sigma_b^b)^d
The kernel parameters are :math:`\sigma^2` (variance), :math:`\sigma^2_w`
(weight_variance), :math:`\sigma^2_b` (bias_variance) and d
(degree). Only gradients of the first three are provided for (degree). Only gradients of the first three are provided for
kernel optimisation, it is assumed that polynomial degree would kernel optimisation, it is assumed that polynomial degree would
be set by hand. be set by hand.
The kernel is not recommended as it is badly behaved when the The kernel is not recommended as it is badly behaved when the
\sigma^2_w*x'*y + \sigma^2_b has a magnitude greater than one. For completeness :math:`\sigma^2_w\*x'\*y + \sigma^2_b` has a magnitude greater than one. For completeness
there is an automatic relevance determination version of this there is an automatic relevance determination version of this
kernel provided. kernel provided.
@ -32,7 +33,7 @@ class POLY(Kernpart):
:param bias_variance: the variance of the prior over bias parameters :math:`\sigma^2_b` :param bias_variance: the variance of the prior over bias parameters :math:`\sigma^2_b`
:param degree: the degree of the polynomial. :param degree: the degree of the polynomial.
:type degree: int :type degree: int
:param ARD: Auto Relevance Determination. If equal to "False", the kernel is isotropic (ie. one weight variance parameter \sigma^2_w), otherwise there is one weight variance parameter per dimension. :param ARD: Auto Relevance Determination. If equal to "False", the kernel is isotropic (ie. one weight variance parameter :math:`\sigma^2_w`), otherwise there is one weight variance parameter per dimension.
:type ARD: Boolean :type ARD: Boolean
:rtype: Kernpart object :rtype: Kernpart object

39
GPy/likelihoods/noise_models/gp_transformations.py
View file

@ -10,19 +10,23 @@ from GPy.util.univariate_Gaussian import std_norm_pdf,std_norm_cdf,inv_std_norm_
class GPTransformation(object): class GPTransformation(object):
""" """
Link function class for doing non-Gaussian likelihoods approximation Link function class for doing non-Gaussian likelihoods approximation
:param Y: observed output (Nx1 numpy.darray) :param Y: observed output (Nx1 numpy.darray)
..Note:: Y values allowed depend on the likelihood_function used
.. note:: Y values allowed depend on the likelihood_function used
""" """
def __init__(self): def __init__(self):
pass pass
class Identity(GPTransformation): class Identity(GPTransformation):
""" """
$$ .. math::
g(f) = f
$$ g(f) = f
""" """
#def transf(self,mu): #def transf(self,mu):
# return mu # return mu
@ -39,9 +43,10 @@ class Identity(GPTransformation):
class Probit(GPTransformation): class Probit(GPTransformation):
""" """
$$ .. math::
g(f) = \\Phi^{-1} (mu)
$$ g(f) = \\Phi^{-1} (mu)
""" """
#def transf(self,mu): #def transf(self,mu):
# return inv_std_norm_cdf(mu) # return inv_std_norm_cdf(mu)
@ -57,9 +62,9 @@ class Probit(GPTransformation):
class Log(GPTransformation): class Log(GPTransformation):
""" """
$$ .. math::
g(f) = \log(\mu) g(f) = \\log(\\mu)
$$
""" """
#def transf(self,mu): #def transf(self,mu):
# return np.log(mu) # return np.log(mu)
@ -75,9 +80,9 @@ class Log(GPTransformation):
class Log_ex_1(GPTransformation): class Log_ex_1(GPTransformation):
""" """
$$ .. math::
g(f) = \log(\exp(\mu) - 1) g(f) = \\log(\\exp(\\mu) - 1)
$$
""" """
#def transf(self,mu): #def transf(self,mu):
# """ # """
@ -110,9 +115,11 @@ class Reciprocal(GPTransformation):
class Heaviside(GPTransformation): class Heaviside(GPTransformation):
""" """
$$
g(f) = I_{x \in A} .. math::
$$
g(f) = I_{x \\in A}
""" """
def transf(self,f): def transf(self,f):
#transformation goes here #transformation goes here

46
GPy/likelihoods/noise_models/noise_distributions.py
View file

@ -16,7 +16,8 @@ class NoiseDistribution(object):
Likelihood class for doing Expectation propagation Likelihood class for doing Expectation propagation
:param Y: observed output (Nx1 numpy.darray) :param Y: observed output (Nx1 numpy.darray)
..Note:: Y values allowed depend on the LikelihoodFunction used
.. note:: Y values allowed depend on the LikelihoodFunction used
""" """
def __init__(self,gp_link,analytical_mean=False,analytical_variance=False): def __init__(self,gp_link,analytical_mean=False,analytical_variance=False):
#assert isinstance(gp_link,gp_transformations.GPTransformation), "gp_link is not a valid GPTransformation."#FIXME #assert isinstance(gp_link,gp_transformations.GPTransformation), "gp_link is not a valid GPTransformation."#FIXME
@ -51,6 +52,7 @@ class NoiseDistribution(object):
In case it is needed, this function assess the output values or makes any pertinent transformation on them. In case it is needed, this function assess the output values or makes any pertinent transformation on them.
:param Y: observed output (Nx1 numpy.darray) :param Y: observed output (Nx1 numpy.darray)
""" """
return Y return Y
@ -62,18 +64,21 @@ class NoiseDistribution(object):
:param obs: observed output :param obs: observed output
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return stats.norm.pdf(gp,loc=mu,scale=sigma) * self._mass(gp,obs) return stats.norm.pdf(gp,loc=mu,scale=sigma) * self._mass(gp,obs)
def _nlog_product_scaled(self,gp,obs,mu,sigma): def _nlog_product_scaled(self,gp,obs,mu,sigma):
""" """
Negative log-product between the cavity distribution and a likelihood factor. Negative log-product between the cavity distribution and a likelihood factor.
..Note:: The constant term in the Gaussian distribution is ignored.
.. note:: The constant term in the Gaussian distribution is ignored.
:param gp: latent variable :param gp: latent variable
:param obs: observed output :param obs: observed output
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return .5*((gp-mu)/sigma)**2 + self._nlog_mass(gp,obs) return .5*((gp-mu)/sigma)**2 + self._nlog_mass(gp,obs)
@ -85,6 +90,7 @@ class NoiseDistribution(object):
:param obs: observed output :param obs: observed output
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return (gp - mu)/sigma**2 + self._dnlog_mass_dgp(gp,obs) return (gp - mu)/sigma**2 + self._dnlog_mass_dgp(gp,obs)
@ -96,6 +102,7 @@ class NoiseDistribution(object):
:param obs: observed output :param obs: observed output
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return 1./sigma**2 + self._d2nlog_mass_dgp2(gp,obs) return 1./sigma**2 + self._d2nlog_mass_dgp2(gp,obs)
@ -106,6 +113,7 @@ class NoiseDistribution(object):
:param obs: observed output :param obs: observed output
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return sp.optimize.fmin_ncg(self._nlog_product_scaled,x0=mu,fprime=self._dnlog_product_dgp,fhess=self._d2nlog_product_dgp2,args=(obs,mu,sigma),disp=False) return sp.optimize.fmin_ncg(self._nlog_product_scaled,x0=mu,fprime=self._dnlog_product_dgp,fhess=self._d2nlog_product_dgp2,args=(obs,mu,sigma),disp=False)
@ -122,6 +130,7 @@ class NoiseDistribution(object):
:param obs: observed output :param obs: observed output
:param tau: cavity distribution 1st natural parameter (precision) :param tau: cavity distribution 1st natural parameter (precision)
:param v: cavity distribution 2nd natural paramenter (mu*precision) :param v: cavity distribution 2nd natural paramenter (mu*precision)
""" """
mu = v/tau mu = v/tau
mu_hat = self._product_mode(obs,mu,np.sqrt(1./tau)) mu_hat = self._product_mode(obs,mu,np.sqrt(1./tau))
@ -137,7 +146,8 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
..Note:: This function helps computing E(Y_star) = E(E(Y_star|f_star)) .. note:: This function helps computing E(Y_star) = E(E(Y_star|f_star))
""" """
return .5*((gp - mu)/sigma)**2 - np.log(self._mean(gp)) return .5*((gp - mu)/sigma)**2 - np.log(self._mean(gp))
@ -148,6 +158,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return (gp - mu)/sigma**2 - self._dmean_dgp(gp)/self._mean(gp) return (gp - mu)/sigma**2 - self._dmean_dgp(gp)/self._mean(gp)
@ -158,6 +169,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return 1./sigma**2 - self._d2mean_dgp2(gp)/self._mean(gp) + (self._dmean_dgp(gp)/self._mean(gp))**2 return 1./sigma**2 - self._d2mean_dgp2(gp)/self._mean(gp) + (self._dmean_dgp(gp)/self._mean(gp))**2
@ -169,7 +181,8 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
..Note:: This function helps computing E(V(Y_star|f_star)) .. note:: This function helps computing E(V(Y_star|f_star))
""" """
return .5*((gp - mu)/sigma)**2 - np.log(self._variance(gp)) return .5*((gp - mu)/sigma)**2 - np.log(self._variance(gp))
@ -180,6 +193,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return (gp - mu)/sigma**2 - self._dvariance_dgp(gp)/self._variance(gp) return (gp - mu)/sigma**2 - self._dvariance_dgp(gp)/self._variance(gp)
@ -190,6 +204,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return 1./sigma**2 - self._d2variance_dgp2(gp)/self._variance(gp) + (self._dvariance_dgp(gp)/self._variance(gp))**2 return 1./sigma**2 - self._d2variance_dgp2(gp)/self._variance(gp) + (self._dvariance_dgp(gp)/self._variance(gp))**2
@ -201,7 +216,8 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
..Note:: This function helps computing E( E(Y_star|f_star)**2 ) .. note:: This function helps computing E( E(Y_star|f_star)**2 )
""" """
return .5*((gp - mu)/sigma)**2 - 2*np.log(self._mean(gp)) return .5*((gp - mu)/sigma)**2 - 2*np.log(self._mean(gp))
@ -212,6 +228,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return (gp - mu)/sigma**2 - 2*self._dmean_dgp(gp)/self._mean(gp) return (gp - mu)/sigma**2 - 2*self._dmean_dgp(gp)/self._mean(gp)
@ -222,6 +239,7 @@ class NoiseDistribution(object):
:param gp: latent variable :param gp: latent variable
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
return 1./sigma**2 - 2*( self._d2mean_dgp2(gp)/self._mean(gp) - (self._dmean_dgp(gp)/self._mean(gp))**2 ) return 1./sigma**2 - 2*( self._d2mean_dgp2(gp)/self._mean(gp) - (self._dmean_dgp(gp)/self._mean(gp))**2 )
@ -243,6 +261,7 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
maximum = sp.optimize.fmin_ncg(self._nlog_conditional_mean_scaled,x0=self._mean(mu),fprime=self._dnlog_conditional_mean_dgp,fhess=self._d2nlog_conditional_mean_dgp2,args=(mu,sigma),disp=False) maximum = sp.optimize.fmin_ncg(self._nlog_conditional_mean_scaled,x0=self._mean(mu),fprime=self._dnlog_conditional_mean_dgp,fhess=self._d2nlog_conditional_mean_dgp2,args=(mu,sigma),disp=False)
mean = np.exp(-self._nlog_conditional_mean_scaled(maximum,mu,sigma))/(np.sqrt(self._d2nlog_conditional_mean_dgp2(maximum,mu,sigma))*sigma) mean = np.exp(-self._nlog_conditional_mean_scaled(maximum,mu,sigma))/(np.sqrt(self._d2nlog_conditional_mean_dgp2(maximum,mu,sigma))*sigma)
@ -266,6 +285,7 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
""" """
maximum = sp.optimize.fmin_ncg(self._nlog_exp_conditional_mean_sq_scaled,x0=self._mean(mu),fprime=self._dnlog_exp_conditional_mean_sq_dgp,fhess=self._d2nlog_exp_conditional_mean_sq_dgp2,args=(mu,sigma),disp=False) maximum = sp.optimize.fmin_ncg(self._nlog_exp_conditional_mean_sq_scaled,x0=self._mean(mu),fprime=self._dnlog_exp_conditional_mean_sq_dgp,fhess=self._d2nlog_exp_conditional_mean_sq_dgp2,args=(mu,sigma),disp=False)
mean_squared = np.exp(-self._nlog_exp_conditional_mean_sq_scaled(maximum,mu,sigma))/(np.sqrt(self._d2nlog_exp_conditional_mean_sq_dgp2(maximum,mu,sigma))*sigma) mean_squared = np.exp(-self._nlog_exp_conditional_mean_sq_scaled(maximum,mu,sigma))/(np.sqrt(self._d2nlog_exp_conditional_mean_sq_dgp2(maximum,mu,sigma))*sigma)
@ -278,6 +298,7 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
:predictive_mean: output's predictive mean, if None _predictive_mean function will be called. :predictive_mean: output's predictive mean, if None _predictive_mean function will be called.
""" """
# E( V(Y_star|f_star) ) # E( V(Y_star|f_star) )
maximum = sp.optimize.fmin_ncg(self._nlog_exp_conditional_variance_scaled,x0=self._variance(mu),fprime=self._dnlog_exp_conditional_variance_dgp,fhess=self._d2nlog_exp_conditional_variance_dgp2,args=(mu,sigma),disp=False) maximum = sp.optimize.fmin_ncg(self._nlog_exp_conditional_variance_scaled,x0=self._variance(mu),fprime=self._dnlog_exp_conditional_variance_dgp,fhess=self._d2nlog_exp_conditional_variance_dgp2,args=(mu,sigma),disp=False)
@ -310,6 +331,7 @@ class NoiseDistribution(object):
:param mu: cavity distribution mean :param mu: cavity distribution mean
:param sigma: cavity distribution standard deviation :param sigma: cavity distribution standard deviation
:predictive_mean: output's predictive mean, if None _predictive_mean function will be called. :predictive_mean: output's predictive mean, if None _predictive_mean function will be called.
""" """
qf = stats.norm.ppf(p,mu,sigma) qf = stats.norm.ppf(p,mu,sigma)
return self.gp_link.transf(qf) return self.gp_link.transf(qf)
@ -321,6 +343,7 @@ class NoiseDistribution(object):
:param x: tuple (latent variable,output) :param x: tuple (latent variable,output)
:param mu: latent variable's predictive mean :param mu: latent variable's predictive mean
:param sigma: latent variable's predictive standard deviation :param sigma: latent variable's predictive standard deviation
""" """
return self._nlog_product_scaled(x[0],x[1],mu,sigma) return self._nlog_product_scaled(x[0],x[1],mu,sigma)
@ -331,7 +354,9 @@ class NoiseDistribution(object):
:param x: tuple (latent variable,output) :param x: tuple (latent variable,output)
:param mu: latent variable's predictive mean :param mu: latent variable's predictive mean
:param sigma: latent variable's predictive standard deviation :param sigma: latent variable's predictive standard deviation
..Note: Only avilable when the output is continuous
.. note: Only available when the output is continuous
""" """
assert not self.discrete, "Gradient not available for discrete outputs." assert not self.discrete, "Gradient not available for discrete outputs."
return np.array((self._dnlog_product_dgp(gp=x[0],obs=x[1],mu=mu,sigma=sigma),self._dnlog_mass_dobs(obs=x[1],gp=x[0]))) return np.array((self._dnlog_product_dgp(gp=x[0],obs=x[1],mu=mu,sigma=sigma),self._dnlog_mass_dobs(obs=x[1],gp=x[0])))
@ -343,7 +368,9 @@ class NoiseDistribution(object):
:param x: tuple (latent variable,output) :param x: tuple (latent variable,output)
:param mu: latent variable's predictive mean :param mu: latent variable's predictive mean
:param sigma: latent variable's predictive standard deviation :param sigma: latent variable's predictive standard deviation
..Note: Only avilable when the output is continuous
.. note: Only available when the output is continuous
""" """
assert not self.discrete, "Hessian not available for discrete outputs." assert not self.discrete, "Hessian not available for discrete outputs."
cross_derivative = self._d2nlog_mass_dcross(gp=x[0],obs=x[1]) cross_derivative = self._d2nlog_mass_dcross(gp=x[0],obs=x[1])
@ -356,14 +383,17 @@ class NoiseDistribution(object):
:param x: tuple (latent variable,output) :param x: tuple (latent variable,output)
:param mu: latent variable's predictive mean :param mu: latent variable's predictive mean
:param sigma: latent variable's predictive standard deviation :param sigma: latent variable's predictive standard deviation
""" """
return sp.optimize.fmin_ncg(self._nlog_joint_predictive_scaled,x0=(mu,self.gp_link.transf(mu)),fprime=self._gradient_nlog_joint_predictive,fhess=self._hessian_nlog_joint_predictive,args=(mu,sigma),disp=False) return sp.optimize.fmin_ncg(self._nlog_joint_predictive_scaled,x0=(mu,self.gp_link.transf(mu)),fprime=self._gradient_nlog_joint_predictive,fhess=self._hessian_nlog_joint_predictive,args=(mu,sigma),disp=False)
def predictive_values(self,mu,var): def predictive_values(self,mu,var):
""" """
Compute mean, variance and conficence interval (percentiles 5 and 95) of the prediction Compute mean, variance and conficence interval (percentiles 5 and 95) of the prediction.
:param mu: mean of the latent variable :param mu: mean of the latent variable
:param var: variance of the latent variable :param var: variance of the latent variable
""" """
if isinstance(mu,float) or isinstance(mu,int): if isinstance(mu,float) or isinstance(mu,int):
mu = [mu] mu = [mu]

8
GPy/likelihoods/noise_models/poisson_noise.py
View file

@ -13,10 +13,10 @@ class Poisson(NoiseDistribution):
""" """
Poisson likelihood Poisson likelihood
Y is expected to take values in {0,1,2,...} Y is expected to take values in {0,1,2,...}
-----
$$ .. math::
L(x) = \exp(\lambda) * \lambda**Y_i / Y_i! L(x) = \\exp(\\lambda) * \\frac{\\lambda^Y_i}{Y_i!}
$$
""" """
def __init__(self,gp_link=None,analytical_mean=False,analytical_variance=False): def __init__(self,gp_link=None,analytical_mean=False,analytical_variance=False):
#self.discrete = True #self.discrete = True

9
GPy/mappings/mlp.py
View file

@ -10,11 +10,13 @@ class MLP(Mapping):
.. math:: .. math::
f(\mathbf{x}*) = \mathbf{W}^0\boldsymbol{\phi}(\mathbf{W}^1\mathbf{x}+\mathb{b}^1)^* + \mathbf{b}^0 f(\\mathbf{x}*) = \\mathbf{W}^0\\boldsymbol{\\phi}(\\mathbf{W}^1\\mathbf{x}+\\mathbf{b}^1)^* + \\mathbf{b}^0
where where
..math::
\phi(\cdot) = \text{tanh}(\cdot) .. math::
\\phi(\\cdot) = \\text{tanh}(\\cdot)
:param X: input observations :param X: input observations
:type X: ndarray :type X: ndarray
@ -22,6 +24,7 @@ class MLP(Mapping):
:type output_dim: int :type output_dim: int
:param hidden_dim: dimension of hidden layer. If it is an int, there is one hidden layer of the given dimension. If it is a list of ints there are as manny hidden layers as the length of the list, each with the given number of hidden nodes in it. :param hidden_dim: dimension of hidden layer. If it is an int, there is one hidden layer of the given dimension. If it is a list of ints there are as manny hidden layers as the length of the list, each with the given number of hidden nodes in it.
:type hidden_dim: int or list of ints. :type hidden_dim: int or list of ints.
""" """
def __init__(self, input_dim=1, output_dim=1, hidden_dim=3): def __init__(self, input_dim=1, output_dim=1, hidden_dim=3):

8
GPy/models/mrd.py
View file

@ -39,6 +39,7 @@ class MRD(Model):
:param num_inducing: number of inducing inputs to use :param num_inducing: number of inducing inputs to use
:param kernels: list of kernels or kernel shared for all BGPLVMS :param kernels: list of kernels or kernel shared for all BGPLVMS
:type kernels: [GPy.kern.kern] | GPy.kern.kern | None (default) :type kernels: [GPy.kern.kern] | GPy.kern.kern | None (default)
""" """
def __init__(self, likelihood_or_Y_list, input_dim, num_inducing=10, names=None, def __init__(self, likelihood_or_Y_list, input_dim, num_inducing=10, names=None,
kernels=None, initx='PCA', kernels=None, initx='PCA',
@ -338,8 +339,11 @@ class MRD(Model):
def plot_scales(self, fignum=None, ax=None, titles=None, sharex=False, sharey=True, *args, **kwargs): def plot_scales(self, fignum=None, ax=None, titles=None, sharex=False, sharey=True, *args, **kwargs):
""" """
:param:`titles` :
titles for axes of datasets TODO: Explain other parameters
:param titles: titles for axes of datasets
""" """
if titles is None: if titles is None:
titles = [r'${}$'.format(name) for name in self.names] titles = [r'${}$'.format(name) for name in self.names]

54
GPy/util/linalg.py
View file

@ -27,48 +27,56 @@ except:
_blas_available = False _blas_available = False
def dtrtrs(A, B, lower=0, trans=0, unitdiag=0): def dtrtrs(A, B, lower=0, trans=0, unitdiag=0):
"""Wrapper for lapack dtrtrs function """
Wrapper for lapack dtrtrs function
:param A: Matrix A :param A: Matrix A
:param B: Matrix B :param B: Matrix B
:param lower: is matrix lower (true) or upper (false) :param lower: is matrix lower (true) or upper (false)
:returns: :returns:
""" """
return lapack.dtrtrs(A, B, lower=lower, trans=trans, unitdiag=unitdiag) return lapack.dtrtrs(A, B, lower=lower, trans=trans, unitdiag=unitdiag)
def dpotrs(A, B, lower=0): def dpotrs(A, B, lower=0):
"""Wrapper for lapack dpotrs function """
Wrapper for lapack dpotrs function
:param A: Matrix A :param A: Matrix A
:param B: Matrix B :param B: Matrix B
:param lower: is matrix lower (true) or upper (false) :param lower: is matrix lower (true) or upper (false)
:returns: :returns:
""" """
return lapack.dpotrs(A, B, lower=lower) return lapack.dpotrs(A, B, lower=lower)
def dpotri(A, lower=0): def dpotri(A, lower=0):
"""Wrapper for lapack dpotri function """
Wrapper for lapack dpotri function
:param A: Matrix A :param A: Matrix A
:param lower: is matrix lower (true) or upper (false) :param lower: is matrix lower (true) or upper (false)
:returns: A inverse :returns: A inverse
""" """
return lapack.dpotri(A, lower=lower) return lapack.dpotri(A, lower=lower)
def trace_dot(a, b): def trace_dot(a, b):
""" """
efficiently compute the trace of the matrix product of a and b Efficiently compute the trace of the matrix product of a and b
""" """
return np.sum(a * b) return np.sum(a * b)
def mdot(*args): def mdot(*args):
"""Multiply all the arguments using matrix product rules. """
Multiply all the arguments using matrix product rules.
The output is equivalent to multiplying the arguments one by one The output is equivalent to multiplying the arguments one by one
from left to right using dot(). from left to right using dot().
Precedence can be controlled by creating tuples of arguments, Precedence can be controlled by creating tuples of arguments,
for instance mdot(a,((b,c),d)) multiplies a (a*((b*c)*d)). for instance mdot(a,((b,c),d)) multiplies a (a*((b*c)*d)).
Note that this means the output of dot(a,b) and mdot(a,b) will differ if Note that this means the output of dot(a,b) and mdot(a,b) will differ if
a or b is a pure tuple of numbers. a or b is a pure tuple of numbers.
""" """
if len(args) == 1: if len(args) == 1:
return args[0] return args[0]
@ -119,10 +127,12 @@ def jitchol_old(A, maxtries=5):
:rval L: the Cholesky decomposition of A :rval L: the Cholesky decomposition of A
.. Note: .. note:
Adds jitter to K, to enforce positive-definiteness Adds jitter to K, to enforce positive-definiteness
if stuff breaks, please check: if stuff breaks, please check:
np.allclose(sp.linalg.cholesky(XXT, lower = True), np.triu(sp.linalg.cho_factor(XXT)[0]).T) np.allclose(sp.linalg.cholesky(XXT, lower = True), np.triu(sp.linalg.cho_factor(XXT)[0]).T)
""" """
try: try:
return linalg.cholesky(A, lower=True) return linalg.cholesky(A, lower=True)
@ -142,6 +152,7 @@ def jitchol_old(A, maxtries=5):
def pdinv(A, *args): def pdinv(A, *args):
""" """
:param A: A DxD pd numpy array :param A: A DxD pd numpy array
:rval Ai: the inverse of A :rval Ai: the inverse of A
@ -152,6 +163,7 @@ def pdinv(A, *args):
:rtype Li: np.ndarray :rtype Li: np.ndarray
:rval logdet: the log of the determinant of A :rval logdet: the log of the determinant of A
:rtype logdet: float64 :rtype logdet: float64
""" """
L = jitchol(A, *args) L = jitchol(A, *args)
logdet = 2.*np.sum(np.log(np.diag(L))) logdet = 2.*np.sum(np.log(np.diag(L)))
@ -177,14 +189,13 @@ def chol_inv(L):
def multiple_pdinv(A): def multiple_pdinv(A):
""" """
Arguments
---------
:param A: A DxDxN numpy array (each A[:,:,i] is pd) :param A: A DxDxN numpy array (each A[:,:,i] is pd)
Returns :rval invs: the inverses of A
------- :rtype invs: np.ndarray
invs : the inverses of A :rval hld: 0.5* the log of the determinants of A
hld: 0.5* the log of the determinants of A :rtype hld: np.array
""" """
N = A.shape[-1] N = A.shape[-1]
chols = [jitchol(A[:, :, i]) for i in range(N)] chols = [jitchol(A[:, :, i]) for i in range(N)]
@ -198,15 +209,13 @@ def PCA(Y, input_dim):
""" """
Principal component analysis: maximum likelihood solution by SVD Principal component analysis: maximum likelihood solution by SVD
Arguments
---------
:param Y: NxD np.array of data :param Y: NxD np.array of data
:param input_dim: int, dimension of projection :param input_dim: int, dimension of projection
Returns
-------
:rval X: - Nxinput_dim np.array of dimensionality reduced data :rval X: - Nxinput_dim np.array of dimensionality reduced data
W - input_dimxD mapping from X to Y :rval W: - input_dimxD mapping from X to Y
""" """
if not np.allclose(Y.mean(axis=0), 0.0): if not np.allclose(Y.mean(axis=0), 0.0):
print "Y is not zero mean, centering it locally (GPy.util.linalg.PCA)" print "Y is not zero mean, centering it locally (GPy.util.linalg.PCA)"
@ -273,11 +282,10 @@ def DSYR_blas(A, x, alpha=1.):
Performs a symmetric rank-1 update operation: Performs a symmetric rank-1 update operation:
A <- A + alpha * np.dot(x,x.T) A <- A + alpha * np.dot(x,x.T)
Arguments
---------
:param A: Symmetric NxN np.array :param A: Symmetric NxN np.array
:param x: Nx1 np.array :param x: Nx1 np.array
:param alpha: scalar :param alpha: scalar
""" """
N = c_int(A.shape[0]) N = c_int(A.shape[0])
LDA = c_int(A.shape[0]) LDA = c_int(A.shape[0])
@ -295,11 +303,10 @@ def DSYR_numpy(A, x, alpha=1.):
Performs a symmetric rank-1 update operation: Performs a symmetric rank-1 update operation:
A <- A + alpha * np.dot(x,x.T) A <- A + alpha * np.dot(x,x.T)
Arguments
---------
:param A: Symmetric NxN np.array :param A: Symmetric NxN np.array
:param x: Nx1 np.array :param x: Nx1 np.array
:param alpha: scalar :param alpha: scalar
""" """
A += alpha * np.dot(x[:, None], x[None, :]) A += alpha * np.dot(x[:, None], x[None, :])
@ -363,8 +370,9 @@ def cholupdate(L, x):
""" """
update the LOWER cholesky factor of a pd matrix IN PLACE update the LOWER cholesky factor of a pd matrix IN PLACE
if L is the lower chol. of K, then this function computes L_ if L is the lower chol. of K, then this function computes L\_
where L_ is the lower chol of K + x*x^T where L\_ is the lower chol of K + x*x^T
""" """
support_code = """ support_code = """
#include <math.h> #include <math.h>

38
GPy/util/mocap.py
View file

@ -92,13 +92,15 @@ class tree:
def swap_vertices(self, i, j): def swap_vertices(self, i, j):
"""Swap two vertices in the tree structure array. """
Swap two vertices in the tree structure array.
swap_vertex swaps the location of two vertices in a tree structure array. swap_vertex swaps the location of two vertices in a tree structure array.
ARG tree : the tree for which two vertices are to be swapped.
ARG i : the index of the first vertex to be swapped. :param tree: the tree for which two vertices are to be swapped.
ARG j : the index of the second vertex to be swapped. :param i: the index of the first vertex to be swapped.
RETURN tree : the tree structure with the two vertex locations :param j: the index of the second vertex to be swapped.
swapped. :rval tree: the tree structure with the two vertex locations swapped.
""" """
store_vertex_i = self.vertices[i] store_vertex_i = self.vertices[i]
store_vertex_j = self.vertices[j] store_vertex_j = self.vertices[j]
@ -117,12 +119,16 @@ class tree:
def rotation_matrix(xangle, yangle, zangle, order='zxy', degrees=False): def rotation_matrix(xangle, yangle, zangle, order='zxy', degrees=False):
"""Compute the rotation matrix for an angle in each direction. """
Compute the rotation matrix for an angle in each direction.
This is a helper function for computing the rotation matrix for a given set of angles in a given order. This is a helper function for computing the rotation matrix for a given set of angles in a given order.
ARG xangle : rotation for x-axis.
ARG yangle : rotation for y-axis. :param xangle: rotation for x-axis.
ARG zangle : rotation for z-axis. :param yangle: rotation for y-axis.
ARG order : the order for the rotations.""" :param zangle: rotation for z-axis.
:param order: the order for the rotations.
"""
if degrees: if degrees:
xangle = math.radians(xangle) xangle = math.radians(xangle)
yangle = math.radians(yangle) yangle = math.radians(yangle)
@ -301,10 +307,14 @@ class acclaim_skeleton(skeleton):
def load_skel(self, file_name): def load_skel(self, file_name):
"""Loads an ASF file into a skeleton structure. """
Loads an ASF file into a skeleton structure.
loads skeleton structure from an acclaim skeleton file. loads skeleton structure from an acclaim skeleton file.
ARG file_name : the file name to load in.
RETURN skel : the skeleton for the file.""" :param file_name: the file name to load in.
:rval skel: the skeleton for the file. - TODO isn't returning this?
"""
fid = open(file_name, 'r') fid = open(file_name, 'r')
self.read_skel(fid) self.read_skel(fid)

2
GPy/util/plot_latent.py
View file

@ -15,7 +15,7 @@ def most_significant_input_dimensions(model, which_indices):
try: try:
input_1, input_2 = np.argsort(model.input_sensitivity())[::-1][:2] input_1, input_2 = np.argsort(model.input_sensitivity())[::-1][:2]
except: except:
raise ValueError, "cannot Atomatically determine which dimensions to plot, please pass 'which_indices'" raise ValueError, "cannot automatically determine which dimensions to plot, please pass 'which_indices'"
else: else:
input_1, input_2 = which_indices input_1, input_2 = which_indices
return input_1, input_2 return input_1, input_2

2
doc/Makefile
View file

@ -2,7 +2,7 @@
# #
# You can set these variables from the command line. # You can set these variables from the command line.
SPHINXOPTS = SPHINXOPTS = -a -w log.txt -E
SPHINXBUILD = sphinx-build SPHINXBUILD = sphinx-build
PAPER = PAPER =
BUILDDIR = _build BUILDDIR = _build

2
doc/conf.py
View file

@ -106,7 +106,7 @@ class Mock(object):
print "Mocking" print "Mocking"
MOCK_MODULES = ['sympy', MOCK_MODULES = ['sympy',
'sympy.utilities', 'sympy.utilities.codegen', 'sympy.core.cache', 'sympy.utilities', 'sympy.utilities.codegen', 'sympy.core.cache',
'sympy.core', 'sympy.parsing', 'sympy.parsing.sympy_parser' 'sympy.core', 'sympy.parsing', 'sympy.parsing.sympy_parser', 'Tango', 'numdifftools'
] ]
for mod_name in MOCK_MODULES: for mod_name in MOCK_MODULES:
sys.modules[mod_name] = Mock() sys.modules[mod_name] = Mock()

2
doc/tuto_interacting_with_models.rst
View file

@ -107,7 +107,7 @@ inputs: ::
m['iip'] = np.arange(-5,0) m['iip'] = np.arange(-5,0)
Getting the model's likelihood and gradients Getting the model's likelihood and gradients
=========================================== =============================================
Appart form the printing the model, the marginal Appart form the printing the model, the marginal
log-likelihood can be obtained by using the function log-likelihood can be obtained by using the function
``log_likelihood()``. Also, the log-likelihood gradients ``log_likelihood()``. Also, the log-likelihood gradients