Struct rkf45::ErrorControlOptions

pub struct ErrorControlOptions {
    pub absolute_magnitude: f64,
    pub relative_magnitude: f64,
    pub function_scale: f64,
    pub derivative_scale: f64,
}

Options for computing desired error when performing adaptive time-stepping.

rkf45_adaptive compares its estimate to a desired error, which is computed as

D = max(absolute_magnitude + relative_magnitude * (function_scale * y[i] + derivative_scale * dt * dydt[i]))

where [i] denotes the ith component, and the max is taken over all components.

This form supports a variety of ways of controlling the size of the desired error relative to y itself or to its increments.

Fields

absolute_magnitude: f64

Magnitude of the absolute component of desired error. Even if relative error is the primary feature of interest, this must be set to a nonzero value as a safety measure in case both y and dydt are near zero.

relative_magnitude: f64

Magnitude of the relative component of desired error.

function_scale: f64

Contribution of y to the relative component of desired error.

derivative_scale: f64

Contribution of y’s increment (dt * dydt) to the relative component of desired error.

Implementations

impl ErrorControlOptions

pub fn simple(scale: f64) -> ErrorControlOptions

A simple error control option that sets the desired error to

D = max(scale * (1 + y[i] + dt * dydt[i]))

This has a lower bound of scale, but it grows proportional to y or dydt as either one becomes large.