Package index • ngme2

ngme2 pacakge description

ngme2-package ngme2: Inference and prediction for mixed effects models with flexible non-Gaussian and Gaussian distributions.

Fit the linear mixed effects model

ngme(): Fit an additive linear mixed effect model over replicates

summary(<ngme>): Summary of ngme fit result

ngme_result(): Access the result of a ngme fitted model

ngme_cov_matrix(): variance of the data or the latent field

ngme_post_samples(): posterior samples of different latent models

Latent models

ngme_model_types(): Show ngme model types

f(): Specifying a latent process model (wrapper function for each model)

ar(): ngme AR(p) model specification

ar1(): ngme AR(1) model specification

arma(): ngme ARMA(p, q) model specification

arma11(): Convenience wrapper for ARMA(1,1)

matern(): ngme Matern SPDE model specification

rw1(): Random Walk Model of Order 1 (RW1)

rw2(): Random Walk Model of Order 2 (RW2)

ou(): Ornstein-Uhlenbeck Process Model

iid(): ngme iid model specification

tp(): ngme tensor-product model specification

bv(): Ngme bivariate model specification

bv_matern(): Ngme bivariate model with Matern sub_models

compute_score_given_pred(): Compute the scores given the prediction

re(): ngme random effect model

spacetime(): Ngme space-time non-separable model specification

generic(): Generic precision matrix operator

generic_ns(): Non-stationary precision matrix operator with custom matrix combinations

ngme2 noises

ngme_noise() noise_normal() noise_nig() noise_gal() noise_skew_t() noise_t() noise_normal_nig(): ngme noise specification

ngme_noise_types(): Show ngme noise types

compare_noise_kld(): Compare noise objects using Kullback-Leibler divergence

print(<noise_kld_comparison>): Print method for noise_kld_comparison

ngme2 optimization methods

sgd(): Vanilla SGD optimization

precond_sgd(): Preconditioner SGD optimization

momentum(): Momentum SGD optimization

adagrad(): AdaGrad SGD optimization

rmsprop(): Root Mean Square Propagation (RMSProp) SGD optimization

adam(): Adam SGD optimization

adamW(): AdamW SGD optimization

bfgs(): BFGS optimization

adaptive_gd(): Adaptive gradient descent From the paper: https://arxiv.org/pdf/1910.09529 The update rule for adaptive gradient descent is: $$\lambda_k = \min(\sqrt{1 + \theta_{k-1}} \lambda_{k-1}, \frac{||x_k - x_{k-1}||}{2 ||\nabla f(x_k) - \nabla f(x_{k-1})||} )$$ $$x_{k+1} = x_k - \lambda_k \nabla f(x_k)$$ $$\theta_k = \lambda_k / \lambda_{k-1}$$

Prediction and Cross validation

predict(<ngme>): Predict function of ngme2 predict using ngme after estimation

cross_validation(): Compute the cross-validation for the ngme model Perform cross-validation for ngme model first into sub_groups (a list of target, and train data)

Data visualization

plot(<ngme_noise>): Plot the density of one or more stationary noise objects

traceplot(): Trace plot of ngme fitting

Simulation functions

simulate(<ngme>): Simulate from a ngme object (possibly with replicates)

simulate(<ngme_model>): Simulate latent process with noise

simulate(<ngme_noise>): Simulate ngme noise object

Auxiliary functions

ngme_update(): Update ngme2 to the latest stable version

ngme_optimizers(): List supported optimizers

get_data_from_formula(): Extracts design matrix from a formula and data.

compute_log_like(): Compute Gaussian log-likelihood

make_time_series_cv_index(): Create Time Series Cross-Validation Indices

precision_matrix_multivariate(): Compute the precision matrix for multivariate model

precision_matrix_multivariate_spde(): Compute the precision matrix for multivariate spde Matern model

compute_index_corr_from_map(): Helper function to compute the index_corr vector

mean_list(): taking mean over a list of nested lists

ngme_as_sparse(): Convert sparse matrix into sparse dgCMatrix

control_opt(): Generate control specifications for ngme() function.

control_ngme(): Generate control specifications for the ngme model

ngme_parse_formula(): Parse the formula for ngme function

ngme_ts_make_A(): Make observation matrix for time series

print(<ngme>): Print an ngme model

print(<ngme_model>): Print ngme model

print(<ngme_operator>): Print ngme operator

print(<ngme_replicate>): Print ngme object

print(<ngme_noise>): Print ngme noise

print(<ngme_trajectories>): Print method for ngme_trajectories

print(<parameter_distance>): Print method for parameter_distance

get_parameter_distance(): Calculate parameter distance from true values

get_trace_trajectories(): Get trace trajectories from ngme fitting

plot(<parameter_distance>): Plot method for parameter_distance

ngme_prior(): ngme prior specification

ngme_prior_types(): Show ngme priors

test_ngme(): Test ngme function

ngme_make_mesh_repls(): ngme make mesh for different replicates

merge_noise(): Merge 2 noise into 1 noise

merge_replicates(): Merge model of replicates into model of 1 replicate given train_idx and test_idx, the merged model contains all the information of train_idx from different replicates.

get_trajectories(): get the trajectories of parameters of the model

name2fun(): Convert transformation name to function

create_paired_cv_splits(): Create paired indices for bivariate cross-validation Ensures that paired observations (e.g., u_wind and v_wind at same location) are kept together in the same fold

openmp_test(): Test OpenMP availability and report the number of threads.

Distributions

dig() rig() pig() qig(): The Inverse-Gaussian (IG) Distribution

dnig() rnig() pnig() qnig(): The Normal Inverse-Gaussian (NIG) Distribution

digam() rigam() pigam() qigam(): The Inverse-Gamma (IGam) Distribution

dgig() rgig() pgig() qgig(): The Generalised Inverse-Gaussian (GIG) Distribution

dgal() rgal() pgal() qgal(): The Generalized Asymmetric Laplace (GAL) Distribution

Datasets

argo_float: Argo float dataset

cienaga: The swamp of Cienaga Grande in Santa Marta, Colombia

cienaga.border: The x y location of the border of the swamp of Cienaga Grande in Santa Marta, Colombia