rk4_step: Runge Kutta solver for ordinary differential equations

In tfNeuralODE: Create Neural Ordinary Differential Equations with 'tensorflow'

Runge Kutta solver for ordinary differential equations

Description

Runge Kutta solver for ordinary differential equations

Usage

rk4_step(func, dt, state)

Arguments

func	The function to be numerically integrated.
dt	Time step.
state	A list describing the state of the function, with the first element being 1, and the second being a tensor that represents state

Value

A list containing a new time and the numerical integration of of the function across the time step to the new time.

Examples

reticulate::py_module_available("tensorflow")
# example code
library(tensorflow)
ode_fun<- function(u){
  r = u ^ 3
  true_A = rbind(c(-0.1, 2.0), c(-2.0, -0.1))
  du <- r %*% (true_A)
  return(as.matrix(du))
}
y<- tensorflow::tf$cast(t(as.matrix(c(2, 0))), dtype = tf$float32)
x<- rk4_step(ode_fun,  dt = 0.25,
            state = list(1.0, y))
x

tfNeuralODE documentation built on Oct. 17, 2023, 1:10 a.m.