I am translating some code from Python to R, and am finding it hard to find the corresponding functions in each. In this particular case, the code I'm having trouble with is:
x_sol_best = solve_ivp(
fun=model_covid,
y0=x_0_cases,
t_span=[t_predictions[0], t_predictions[-1]],
t_eval=t_predictions,
args=tuple(optimal_params),
).y
From the scipy.integrate.solve_ivp documentation, I see that the default integration method used in this function is : ‘RK45’ (default): Explicit Runge-Kutta method of order 5(4)
What packages / functions in R would be equivalent to this?
From the R documentation of the ode function in R, I see that there are a number of RK 4(5) methods available (pasted below) - but the Python documentation states that RK45 is order 5(4)...
Can anyone offer any clarification? TIA
"rk45ck" | Runge-Kutta Cash-Karp, order 4(5)
"rk45f" | Runge-Kutta-Fehlberg, order 4(5); Octave: ode45, pair=1
"rk45e" | Runge-Kutta-England, order 4(5)
"rk45dp6" | Dormand-Prince, order 4(5), local order 6
"rk45dp7", "ode45" | Dormand-Prince 4(5), local order 7
According to the documentation, the default solver in solve_ivp() is Dormand-Prince. This called ode45 in the ode() function of the deSolve package.
x_sol_best = deSolve::ode(
y = x_0_cases,
times = t_predictions,
func = model_covid,
parms = c(...), # vector of parameter values
method = "ode45"
)[ , -1] # drop the t column