GP

class fvgp.GP(x_data, y_data, init_hyperparameters=None, hyperparameter_bounds=None, noise_variances=None, compute_device='cpu', gp_kernel_function=None, gp_kernel_function_grad=None, gp_noise_function=None, gp_noise_function_grad=None, gp_mean_function=None, gp_mean_function_grad=None, gp2Scale=False, gp2Scale_dask_client=None, gp2Scale_batch_size=10000, calc_inv=False, online=False, ram_economy=False, args=None, info=False)

This class provides all the tools for a single-task Gaussian Process (GP). Use fvGP for multitask GPs. However, the fvGP class inherits all methods from this class. This class allows for full HPC support for training via the HGDL package.

V … number of input points

D … input space dimensionality

N … arbitrary integers (N1, N2,…)

Parameters:

x_data (np.ndarray or list of tuples) – The input point positions. Shape (V x D), where D is the input_space_dim. If dealing with non-Euclidean inputs x_data should be a list, not a numpy array.
y_data (np.ndarray) – The values of the data points. Shape (V,1) or (V).
init_hyperparameters (np.ndarray, optional) – Vector of hyperparameters used by the GP initially. This class provides methods to train hyperparameters. The default is a random draw from a uniform distribution within hyperparameter_bounds, with a shape appropriate for the default kernel (D + 1), which is an anisotropic Matern kernel with automatic relevance determination (ARD). If sparse_node or gp2Scale is enabled, the default kernel changes to the anisotropic Wendland kernel.
hyperparameter_bounds (np.ndarray, optional) – A 2d numpy array of shape (N x 2), where N is the number of needed hyperparameters. The default is None, in which case the hyperparameter_bounds are estimated from the domain size and the initial y_data. If the data set changes significantly, the hyperparameters and the bounds should be changed/retrained. Initial hyperparameters and bounds can also be set in the train calls. The default only works for the default kernels.
noise_variances (np.ndarray, optional) – An numpy array defining the uncertainties/noise in the data y_data in form of a point-wise variance. Shape (len(y_data), 1) or (len(y_data)). Note: if no noise_variances are provided here, the gp_noise_function callable will be used; if the callable is not provided, the noise variances will be set to abs(np.mean(y_data) / 100.0. If noise covariances are required, also make use of the gp_noise_function.
compute_device (str, optional) – One of “cpu” or “gpu”, determines how linear system solves are run. The default is “cpu”. For “gpu”, pytorch has to be installed manually. If gp2Scale is enabled but no kernel is provided, the choice of the compute_device becomes much more important. In that case, the default kernel will be computed on the cpu or the gpu which will significantly change the compute time depending on the compute architecture.
gp_kernel_function (Callable, optional) – A symmetric positive semi-definite covariance function (a kernel) that calculates the covariance between data points. It is a function of the form k(x1,x2,hyperparameters, obj). The input x1 is a N1 x D array of positions, x2 is a N2 x D array of positions, the hyperparameters argument is a 1d array of length D+1 for the default kernel and of a different user-defined length for other kernels obj is an fvgp.GP instance. The default is a stationary anisotropic kernel (fvgp.GP.default_kernel) which performs automatic relevance determination (ARD). The output is a covariance matrix, an N1 x N2 numpy array.
gp_kernel_function_grad (Callable, optional) – A function that calculates the derivative of the gp_kernel_function with respect to the hyperparameters. If provided, it will be used for local training (optimization) and can speed up the calculations. It accepts as input x1 (a N1 x D array of positions), x2 (a N2 x D array of positions), hyperparameters (a 1d array of length D+1 for the default kernel), and a fvgp.GP instance. The default is a finite difference calculation. If ‘ram_economy’ is True, the function’s input is x1, x2, direction (int), hyperparameters (numpy array), and a fvgp.GP instance, and the output is a numpy array of shape (len(hps) x N). If ‘ram economy’ is False,the function’s input is x1, x2, hyperparameters, and a fvgp.GP instance. The output is a numpy array of shape (len(hyperparameters) x N1 x N2). See ‘ram_economy’.
gp_mean_function (Callable, optional) – A function that evaluates the prior mean at a set of input position. It accepts as input an array of positions (of shape N1 x D), hyperparameters (a 1d array of length D+1 for the default kernel) and a fvgp.GP instance. The return value is a 1d array of length N1. If None is provided, fvgp.GP._default_mean_function is used.
gp_mean_function_grad (Callable, optional) – A function that evaluates the gradient of the gp_mean_function at a set of input positions with respect to the hyperparameters. It accepts as input an array of positions (of size N1 x D), hyperparameters (a 1d array of length D+1 for the default kernel) and a fvgp.GP instance. The return value is a 2d array of shape (len(hyperparameters) x N1). If None is provided, either zeros are returned since the default mean function does not depend on hyperparameters, or a finite-difference approximation is used if gp_mean_function is provided.
gp_noise_function (Callable optional) – The noise function is a callable f(x,hyperparameters,obj) that returns a positive symmetric definite matrix of shape(len(x),len(x)). The input x is a numpy array of shape (N x D). The hyperparameter array is the same that is communicated to mean and kernel functions. The obj is a fvgp.GP instance.
gp_noise_function_grad (Callable, optional) – A function that evaluates the gradient of the gp_noise_function at an input position with respect to the hyperparameters. It accepts as input an array of positions (of size N x D), hyperparameters (a 1d array of length D+1 for the default kernel) and a fvgp.GP instance. The return value is a 3-D array of shape (len(hyperparameters) x N x N). If None is provided, either zeros are returned since the default noise function does not depend on hyperparameters. If gp_noise_function is provided but no gradient function, a finite-difference approximation will be used. The same rules regarding ram economy as for the kernel definition apply here.
gp2Scale (bool, optional) – Turns on gp2Scale. This will distribute the covariance computations across multiple workers. This is an advanced feature for HPC GPs up to 10 million data points. If gp2Scale is used, the default kernel is an anisotropic Wendland kernel which is compactly supported. The noise function will have to return a scipy.sparse matrix instead of a numpy array. There are a few more things to consider (read on); this is an advanced option. If no kernel is provided, the compute_device option should be revisited. The kernel will use the specified device to compute covariances. The default is False.
gp2Scale_dask_client (dask.distributed.Client, optional) – A dask client for gp2Scale to distribute covariance computations over. Has to contain at least 3 workers. On HPC architecture, this client is provided by the job script. Please have a look at the examples. A local client is used as default.
gp2Scale_batch_size (int, optional) – Matrix batch size for distributed computing in gp2Scale. The default is 10000.
calc_inv (bool, optional) – If True, the algorithm calculates and stores the inverse of the covariance matrix after each training or update of the dataset or hyperparameters, which makes computing the posterior covariance faster (5-10 times). For larger problems (>2000 data points), the use of inversion should be avoided due to computational instability and costs. The default is False. Note, the training will always use Cholesky or LU decomposition instead of the inverse for stability reasons. Storing the inverse is a good option when the dataset is not too large and the posterior covariance is heavily used.
online (bool, optional) – A new setting that allows optimization for online applications. Default=False. If True, calc_inv will be set to true, and the inverse and the logdet() of full dataset will only be computed once in the beginning and after that only updated. This leads to a significant speedup because the most costly aspects of a GP are entirely avoided.
ram_economy (bool, optional) – Only of interest if the gradient and/or Hessian of the marginal log_likelihood is/are used for the training. If True, components of the derivative of the marginal log-likelihood are calculated subsequently, leading to a slow-down but much less RAM usage. If the derivative of the kernel (or noise function) with respect to the hyperparameters (gp_kernel_function_grad) is going to be provided, it has to be tailored: for ram_economy=True it should be of the form f(x1[, x2], direction, hyperparameters, obj) and return a 2d numpy array of shape len(x1) x len(x2). If ram_economy=False, the function should be of the form f(x1[, x2,] hyperparameters, obj) and return a numpy array of shape H x len(x1) x len(x2), where H is the number of hyperparameters. CAUTION: This array will be stored and is very large.
args (any, optional) – args will be a class attribute and therefore available to kernel, noise and prior mean functions.
info (bool, optional) – Provides a way how to see the progress of gp2Scale, Default is False

x_data

Datapoint positions

Type:: np.ndarray

y_data

Datapoint values

Type:: np.ndarray

noise_variances

Datapoint observation (co)variances

Type:: np.ndarray

prior.hyperparameters

Current hyperparameters in use.

Type:: np.ndarray

prior.K

Current prior covariance matrix of the GP

Type:: np.ndarray

prior.m

Current prior mean vector.

Type:: np.ndarray

marginal_density.KVinv

If enabled, the inverse of the prior covariance + nose matrix V inv(K+V)

Type:: np.ndarray

marginal_density.KVlogdet

logdet(K+V)

Type:: float

likelihood.V

the noise covariance matrix

Type:: np.ndarray

update_gp_data(x_new, y_new, noise_variances_new=None, append=True)

This function updates the data in the gp object instance. The data will only be overwritten of overwrite = True, otherwise the data will be appended. This is a change from earlier versions. Now, the default is not to overwrite the existing data.

Parameters:

x_new (np.ndarray) – The point positions. Shape (V x D), where D is the input_space_dim.
y_new (np.ndarray) – The values of the data points. Shape (V,1) or (V).
noise_variances_new (np.ndarray, optional) – An numpy array defining the uncertainties in the data y_data in form of a point-wise variance. Shape (len(y_data), 1) or (len(y_data)). Note: if no variances are provided here, the noise_covariance callable will be used; if the callable is not provided the noise variances will be set to abs(np.mean(y_data)) / 100.0. If you provided a noise function, the noise_variances_new will be ignored.
append (bool, optional) – Indication whether to append to or overwrite the existing dataset. Default = True. In the default case, data will be appended.

train(objective_function=None, objective_function_gradient=None, objective_function_hessian=None, hyperparameter_bounds=None, init_hyperparameters=None, method='global', pop_size=20, tolerance=0.0001, max_iter=120, local_optimizer='L-BFGS-B', global_optimizer='genetic', constraints=(), dask_client=None)

This function finds the maximum of the log marginal likelihood and therefore trains the GP (synchronously). This can be done on a remote cluster/computer by specifying the method to be ‘hgdl’ and providing a dask client. However, in that case fvgp.GP.train_async() is preferred. The GP prior will automatically be updated with the new hyperparameters after the training.

Parameters:

objective_function (callable, optional) – The function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a scalar. This function can be used to train via non-standard user-defined objectives. The default is the negative log marginal likelihood.
objective_function_gradient (callable, optional) – The gradient of the function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a vector of len(hps). This function can be used to train via non-standard user-defined objectives. The default is the gradient of the negative log marginal likelihood.
objective_function_hessian (callable, optional) – The hessian of the function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a matrix of shape(len(hps),len(hps)). This function can be used to train via non-standard user-defined objectives. The default is the hessian of the negative log marginal likelihood.
hyperparameter_bounds (np.ndarray, optional) – A numpy array of shape (D x 2), defining the bounds for the optimization. A 2d numpy array of shape (N x 2), where N is the number of hyperparameters. The default is None, in which case the hyperparameter_bounds are estimated from the domain size and the y_data. If the data set changes significantly, the hyperparameters and the bounds should be changed/retrained. The default only works for the default kernels.
init_hyperparameters (np.ndarray, optional) – Initial hyperparameters used as starting location for all optimizers with local component. The default is a random draw from a uniform distribution within the bounds.
method (str or Callable, optional) – The method used to train the hyperparameters. The options are global, local, hgdl, mcmc, and a callable. The callable gets a gp.GP instance and has to return a 1d np.ndarray of hyperparameters. The default is global (scipy’s differential evolution). If method = “mcmc”, the attribute fvgp.GP.mcmc_info is updated and contains convergence and distribution information.
pop_size (int, optional) – A number of individuals used for any optimizer with a global component. Default = 20.
tolerance (float, optional) – Used as termination criterion for local optimizers. Default = 0.0001.
max_iter (int, optional) – Maximum number of iterations for global and local optimizers. Default = 120.
local_optimizer (str, optional) – Defining the local optimizer. Default = L-BFGS-B, most scipy.optimize.minimize functions are permissible.
global_optimizer (str, optional) – Defining the global optimizer. Only applicable to method = hgdl. Default = genetic
constraints (tuple of object instances, optional) – Equality and inequality constraints for the optimization. If the optimizer is hgdl see hgdl.readthedocs.io. If the optimizer is a scipy optimizer, see the scipy documentation.
dask_client (distributed.client.Client, optional) – A Dask Distributed Client instance for distributed training if HGDL is used. If None is provided, a new dask.distributed.Client instance is constructed.

train_async(objective_function=None, objective_function_gradient=None, objective_function_hessian=None, hyperparameter_bounds=None, init_hyperparameters=None, max_iter=10000, local_optimizer='L-BFGS-B', global_optimizer='genetic', constraints=(), dask_client=None)

This function asynchronously finds the maximum of the log marginal likelihood and therefore trains the GP. This can be done on a remote cluster/computer by providing a dask client. This function submits the training and returns an object which can be given to fvgp.GP.update_hyperparameters(), which will automatically update the GP prior with the new hyperparameters.

Parameters:

objective_function (callable, optional) – The function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a scalar. This function can be used to train via non-standard user-defined objectives. The default is the negative log marginal likelihood.
objective_function_gradient (callable, optional) – The gradient of the function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a vector of len(hps). This function can be used to train via non-standard user-defined objectives. The default is the gradient of the negative log marginal likelihood.
objective_function_hessian (callable, optional) – The hessian of the function that will be MINIMIZED for training the GP. The form of the function is f(hyperparameters=hps) and returns a matrix of shape(len(hps),len(hps)). This function can be used to train via non-standard user-defined objectives. The default is the hessian of the negative log marginal likelihood.
hyperparameter_bounds (np.ndarray, optional) – A numpy array of shape (D x 2), defining the bounds for the optimization. A 2d numpy array of shape (N x 2), where N is the number of hyperparameters. The default is None, in which case the hyperparameter_bounds are estimated from the domain size and the y_data. If the data set changes significantly, the hyperparameters and the bounds should be changed/retrained. The default only works for the default kernels.
init_hyperparameters (np.ndarray, optional) – Initial hyperparameters used as starting location for all optimizers with local component. The default is a random draw from a uniform distribution within the bounds.
max_iter (int, optional) – Maximum number of epochs for HGDL. Default = 10000.
local_optimizer (str, optional) – Defining the local optimizer. Default = L-BFGS-B, most scipy.optimize.minimize functions are permissible.
global_optimizer (str, optional) – Defining the global optimizer. Only applicable to method = hgdl. Default = genetic
constraints (tuple of hgdl.NonLinearConstraint instances, optional) – Equality and inequality constraints for the optimization. See hgdl.readthedocs.io
dask_client (distributed.client.Client, optional) – A Dask Distributed Client instance for distributed training if HGDL is used. If None is provided, a new dask.distributed.Client instance is constructed.

Returns:

Optimization object that can be given to fvgp.GP.update_hyperparameters()
to update the prior GP (object instance)

stop_training(opt_obj)

This function stops the training if HGDL is used. It leaves the dask client alive.

Parameters:: opt_obj (HGDL object instance) – HGDL object instance returned by fvgp.GP.train_async()

kill_training(opt_obj)

This function stops the training if HGDL is used, and kills the dask client.

Parameters:: opt_obj (HGDL object instance) – HGDL object instance returned by fvgp.GP.train_async()

update_hyperparameters(opt_obj)

This function asynchronously finds the maximum of the marginal log_likelihood and therefore trains the GP. This can be done on a remote cluster/computer by providing a dask client. This function just submits the training and returns an object which can be given to fvgp.GP.update_hyperparameters(), which will automatically update the GP prior with the new hyperparameters.

Parameters:: opt_obj (HGDL object instance) – HGDL object instance returned by fvgp.GP.train_async()
Returns:: The current hyperparameters
Return type:: np.ndarray

set_hyperparameters(hps)

Function to set hyperparameters.

Parameters:: hps (np.ndarray) – A 1-d numpy array of hyperparameters.

get_hyperparameters()

Function to get the current hyperparameters.

Returns:: hyperparameters
Return type:: np.ndarray

get_prior_pdf()

Function to get the current prior covariance matrix.

Returns:: A dictionary containing information about the GP prior distribution
Return type:: dict

log_likelihood(hyperparameters=None)

Function that computes the marginal log-likelihood

Parameters:: hyperparameters (np.ndarray, optional) – Vector of hyperparameters of shape (N). If not provided, the covariance will not be recomputed.
Returns:: log marginal likelihood of the data
Return type:: float

test_log_likelihood_gradient(hyperparameters)

Function to test your gradient of the log-likelihood and therefore of the kernel function.

Parameters:: hyperparameters (np.ndarray, optional) – Vector of hyperparameters of shape (N).
Return type:: analytical and finite difference gradient to compare

posterior_mean(x_pred, hyperparameters=None, x_out=None)

This function calculates the posterior mean for a set of input points.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
hyperparameters (np.ndarray, optional) – A numpy array of the correct size depending on the kernel. This is optional in case the posterior mean has to be computed with given hyperparameters, which is, for instance, the case if the posterior mean is a constraint during training. The default is None which means the initialized or trained hyperparameters are used.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space (most often 1).

Returns:

Solution points and function values

Return type:

dict

posterior_mean_grad(x_pred, hyperparameters=None, x_out=None, direction=None)

This function calculates the gradient of the posterior mean for a set of input points.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
hyperparameters (np.ndarray, optional) – A numpy array of the correct size depending on the kernel. This is optional in case the posterior mean has to be computed with given hyperparameters, which is, for instance, the case if the posterior mean is a constraint during training. The default is None which means the initialized or trained hyperparameters are used.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.
direction (int, optional) – Direction of derivative, If None (default) the whole gradient will be computed.

Returns:

Solution

Return type:

dict

posterior_covariance(x_pred, x_out=None, variance_only=False, add_noise=False)

Function to compute the posterior covariance.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.
variance_only (bool, optional) – If True the computation of the posterior covariance matrix is avoided which can save compute time. In that case the return will only provide the variance at the input points. Default = False.
add_noise (bool, optional) – If True the noise variances will be added to the posterior variances. Default = False.

Returns:

Solution

Return type:

dict

posterior_covariance_grad(x_pred, x_out=None, direction=None)

Function to compute the gradient of the posterior covariance.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.
direction (int, optional) – Direction of derivative, If None (default) the whole gradient will be computed.

Returns:

Solution

Return type:

dict

joint_gp_prior(x_pred, x_out=None)

Function to compute the joint prior over f (at measured locations) and f_pred at x_pred.

Parameters:

x_pred (np.ndarray or list) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution

Return type:

dict

joint_gp_prior_grad(x_pred, direction, x_out=None)

Function to compute the gradient of the data-informed prior.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
direction (int) – Direction of derivative.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution

Return type:

dict

entropy(S)

Function computing the entropy of a normal distribution res = entropy(S); S is a 2d np.ndarray array, a covariance matrix which is non-singular.

Parameters:: S (np.ndarray) – A covariance matrix.
Returns:: Entropy
Return type:: float

gp_entropy(x_pred, x_out=None)

Function to compute the entropy of the gp prior probability distribution.

Parameters:: x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces. Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.
Returns:: Entropy
Return type:: float

gp_entropy_grad(x_pred, direction, x_out=None)

Function to compute the gradient of entropy of the prior in a given direction.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
direction (int) – Direction of derivative.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Entropy gradient in given direction

Return type:

float

kl_div_grad(mu1, dmu1dx, mu2, S1, dS1dx, S2): This function computes the gradient of the KL divergence between two normal distributions when the gradients of the mean and covariance are given. a = kl_div(mu1, dmudx,mu2, S1, dS1dx, S2); S1, S2 are 2d numpy arrays, matrices have to be non-singular, mu1, mu2 are mean vectors, given as 2d arrays

kl_div(mu1, mu2, S1, S2)

Function to compute the KL divergence between two Gaussian distributions.

Parameters:

mu1 (np.ndarray) – Mean vector of distribution 1.
mu2 (np.ndarray) – Mean vector of distribution 2.
S1 (np.ndarray) – Covariance matrix of distribution 1.
S2 (np.ndarray) – Covariance matrix of distribution 2.

Returns:

KL divergence

Return type:

float

gp_kl_div(x_pred, comp_mean, comp_cov, x_out=None)

Function to compute the kl divergence of a posterior at given points and a given normal distribution.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
comp_mean (np.ndarray) – Comparison mean vector for KL divergence. len(comp_mean) = len(x_pred)
comp_cov (np.ndarray) – Comparison covariance matrix for KL divergence. shape(comp_cov) = (len(x_pred),len(x_pred))
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution

Return type:

dict

gp_kl_div_grad(x_pred, comp_mean, comp_cov, direction, x_out=None)

Function to compute the gradient of the kl divergence of a posterior at given points.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
comp_mean (np.ndarray) – Comparison mean vector for KL divergence. len(comp_mean) = len(x_pred)
comp_cov (np.ndarray) – Comparison covariance matrix for KL divergence. shape(comp_cov) = (len(x_pred),len(x_pred))
direction (int) – The direction in which the gradient will be computed.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution

Return type:

dict

mutual_information(joint, m1, m2)

Function to calculate the mutual information between two normal distributions, which is equivalent to the KL divergence(joint, marginal1 * marginal1).

Parameters:

joint (np.ndarray) – The joint covariance matrix.
m1 (np.ndarray) – The first marginal distribution
m2 (np.ndarray) – The second marginal distribution

Returns:

Mutual information

Return type:

float

gp_mutual_information(x_pred, x_out=None)

Function to calculate the mutual information between the random variables f(x_data) and f(x_pred). The mutual information is always positive, as it is a KL divergence, and is bounded from below by 0. The maxima are expected at the data points. Zero is expected far from the data support. :param x_pred: A numpy array of shape (V x D), interpreted as an array of input point positions or a list for

GPs on non-Euclidean input spaces.

Parameters:: x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.
Returns:: Solution
Return type:: dict

gp_total_correlation(x_pred, x_out=None)

Function to calculate the interaction information between the random variables f(x_data) and f(x_pred). This is the mutual information of each f(x_pred) with f(x_data). It is also called the Multiinformation. It is best used when several prediction points are supposed to be mutually aware. The total correlation is always positive, as it is a KL divergence, and is bounded from below by 0. The maxima are expected at the data points. Zero is expected far from the data support.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution – Total correlation between prediction points, as a collective.

Return type:

dict

gp_relative_information_entropy(x_pred, x_out=None)

Function to compute the KL divergence and therefore the relative information entropy of the prior distribution defined over predicted function values and the posterior distribution. The value is a reflection of how much information is predicted to be gained through observing a set of data points at x_pred.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution – Relative information entropy of prediction points, as a collective.

Return type:

dict

gp_relative_information_entropy_set(x_pred, x_out=None)

Function to compute the KL divergence and therefore the relative information entropy of the prior distribution over predicted function values and the posterior distribution. The value is a reflection of how much information is predicted to be gained through observing each data point in x_pred separately, not all at once as in gp_relative_information_entropy.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution – Relative information entropy of prediction points, but not as a collective.

Return type:

dict

posterior_probability(x_pred, comp_mean, comp_cov, x_out=None)

Function to compute probability of a probabilistic quantity of interest, given the GP posterior at given points.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
comp_mean (np.ndarray) – A vector of mean values, same length as x_pred.
comp_cov (np.nparray) – Covariance matrix, in R^{len(x_pred) times len(x_pred)}
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution – The probability of a probabilistic quantity of interest, given the GP posterior at a given point.

Return type:

dict

posterior_probability_grad(x_pred, comp_mean, comp_cov, direction, x_out=None)

Function to compute the gradient of the probability of a probabilistic quantity of interest, given the GP posterior at a given point.

Parameters:

x_pred (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions or a list for GPs on non-Euclidean input spaces.
comp_mean (np.ndarray) – A vector of mean values, same length as x_pred.
comp_cov (np.nparray) – Covariance matrix, in R^{len(x_pred) times len(x_pred)}
direction (int) – The direction to compute the gradient in.
x_out (np.ndarray, optional) – Output coordinates in case of multitask GP use; a numpy array of size (N x L), where N is the number of output points, and L is the dimensionality of the output space.

Returns:

Solution – The gradient of the probability of a probabilistic quantity of interest, given the GP posterior at a given point.

Return type:

dict

normalize_y_data(y_data)

Function to normalize the y_data. The user is responsible to normalize the noise accordingly. This function will not update the object instance.

Parameters:: y_data (np.ndarray) – Numpy array of shape (U).

crps(x_test, y_test)

This function calculates the continuous rank probability score. Note that in the multitask setting the user should perform their input point transformation beforehand.

Parameters:

x_test (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions.
y_test (np.ndarray) – A numpy array of shape (V x 1). These are the y data to compare against.

Returns:

CRPS

Return type:

float

rmse(x_test, y_test)

This function calculates the root mean squared error. Note that in the multitask setting the user should perform their input point transformation beforehand.

Parameters:

x_test (np.ndarray) – A numpy array of shape (V x D), interpreted as an array of input point positions.
y_test (np.ndarray) – A numpy array of shape (V x 1). These are the y data to compare against

Returns:

RMSE

Return type:

float

make_2d_x_pred(bx, by, resx=100, resy=100)

This is a purely convenience-driven function calculating prediction points on a grid. :param bx: A numpy array of shape (2) defining lower and upper bounds in x direction. :type bx: np.ndarray :param by: A numpy array of shape (2) defining lower and upper bounds in y direction. :type by: np.ndarray :param resx: Resolution in x direction. Default = 100. :type resx: int, optional :param resy: Resolution in y direction. Default = 100. :type resy: int, optional

Returns:: prediction points
Return type:: np.ndarray

make_1d_x_pred(b, res=100)

This is a purely convenience-driven function calculating prediction points on a 1d grid.

Parameters:

b (np.ndarray) – A numpy array of shape (2) defineing lower and upper bounds
res (int, optional) – Resolution. Default = 100

Returns:

prediction points

Return type:

np.ndarray