particle.est-class: Particle interaction estimation class

In FastGaSP: Fast and Exact Computation of Gaussian Stochastic Process

Particle interaction estimation class

Description

S4 class for storing estimated parameters and predictions for particle interaction models.

Objects from the Class

Objects of this class are created by the fit.particle.data (via fit) method when applied to particle.data objects to estimate interaction parameters and make predictions.

Slots

data_type:

Object of class character. Specifies the type of data ("simulation" or "experiment").

model:

Object of class characterOrNULL. Specifies the model type for simulation data (e.g., "Vicsek" or "two_interactions_Vicsek"). NULL for experimental data.

D_y:

Object of class numeric. Dimension of the output space.

num_interaction:

Object of class numeric. Number of interactions.

parameters:

Object of class numeric. Vector of estimated parameters with length 2*D_y + 1:

• First D_y elements: beta (inverse range parameters)

• Next D_y elements: tau (variance-noise ratios)

• Last element: interaction radius

sigma_2_0_est:

Object of class numeric. Estimated noise variance.

predictions:

object of class listOrNULL. Contains predicted means and 95% confidence intervals (lower and upper bounds) for the particle interactions if testing inputs are given.

training_data:

Object of class list. Contains the training data used in the GP model, obtained using the estimated interaction radius.

gp_weights:

Object of class matrix. Contains the weights from the GP computation (A^T_j Sigma_y^(-1) y) used for prediction, with each column corresponding to a type of interaction j.

Methods

show:

Method for displaying summary information about the estimated parameters.

References

Fang, X., & Gu, M. (2024). The inverse Kalman filter. arXiv:2407.10089.

See Also

fit.particle.data for more details about how to create a particle.est object. particle.data-class for the input data structure

FastGaSP documentation built on April 4, 2025, 5:16 a.m.