rkg3: The 'rkg3' class object

In KGode: Kernel Based Gradient Matching for Parameter Inference in Ordinary Differential Equations

The 'rkg3' class object

Description

This class provides advanced gradient matching method by using the ode as a regularizer.

Format

R6Class object.

Value

an R6Class object which can be used for improving ode parameters estimation by using ode as a regularizer.

Methods

iterate(iter,innerloop,lamb)

Iteratively updating ode parameters and interpolation regression coefficients.

witerate(iter,innerloop,dtilda,lamb)

Iteratively updating ode parameters and the warped interpolation regression coefficients.

full(par,lam)

Updating ode parameters and rkhs interpolation regression coefficients simultaneously. This method is slow but guarantee convergence.

Public fields

rk

the 'rkhs' class object containing the interpolation information for each state of the ode.

ode_m

the 'ode' class object containing the information about the odes.

Active bindings

ode_m

the 'ode' class object containing the information about the odes.

Methods

Public methods

• rkg3$new()

• rkg3$greet()

• rkg3$add()

• rkg3$iterate()

• rkg3$witerate()

• rkg3$full()

• rkg3$wfull()

• rkg3$opfull()

• rkg3$wopfull()

• rkg3$cross()

• rkg3$fullos()

• rkg3$clone()

---

Method new()

Usage

rkg3$new(rk = NULL, odem = NULL)

---

Method greet()

Usage

rkg3$greet()

---

Method add()

Usage

rkg3$add(x)

---

Method iterate()

Usage

rkg3$iterate(iter, innerloop, lamb)

---

Method witerate()

Usage

rkg3$witerate(iter, innerloop, dtilda, lamb)

---

Method full()

Usage

rkg3$full(par, lam)

---

Method wfull()

Usage

rkg3$wfull(par, lam, dtilda)

---

Method opfull()

Usage

rkg3$opfull(lam)

---

Method wopfull()

Usage

rkg3$wopfull(lam, dtilda)

---

Method cross()

Usage

rkg3$cross(lam, testX, testY)

---

Method fullos()

Usage

rkg3$fullos(par)

---

Method clone()

The objects of this class are cloneable with this method.

Usage

rkg3$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Author(s)

Mu Niu, [e-mail]

KGode documentation built on Sept. 9, 2025, 5:25 p.m.