Skip to main content

StochasticProcess1D

rustyqlib::core::montecarlo::process

Trait StochasticProcess1D 

Source 
pub trait StochasticProcess1D: Sync {
    // Required methods
    fn drift(&self, t: f64, x: f64) -> f64;
    fn diffusion(&self, t: f64, x: f64) -> f64;

    // Provided methods
    fn diffusion_dx(&self, t: f64, x: f64) -> f64 { ... }
    fn exact_step(&self, _t: f64, _x: f64, _dt: f64, _dw: f64) -> Option<f64> { ... }
    fn constrain(&self, x: f64) -> f64 { ... }
    fn evolve(
        &self,
        scheme: DiscretizationScheme,
        t: f64,
        x: f64,
        dt: f64,
        dw: f64,
    ) -> f64 { ... }
}
Expand description

A scalar Itô process dX = a(t, X) dt + b(t, X) dW.

Implementations are shared across rayon path chunks, hence Sync.

Required Methods§

Source

fn drift(&self, t: f64, x: f64) -> f64

Drift coefficient a(t, x).

Source

fn diffusion(&self, t: f64, x: f64) -> f64

Diffusion coefficient b(t, x).

Provided Methods§

Source

fn diffusion_dx(&self, t: f64, x: f64) -> f64

∂b/∂x, the extra coefficient Milstein needs.

Source

fn exact_step(&self, _t: f64, _x: f64, _dt: f64, _dw: f64) -> Option<f64>

One draw from the closed-form transition X_{t+dt} | X_t = x, when the process has one. None (the default) makes the Exact scheme fall back to Euler.

Source

fn constrain(&self, x: f64) -> f64

State constraint applied after every step. Identity by default — only processes whose state space is genuinely bounded (lognormal equity at zero, truncated variance) should clamp.

Source

fn evolve( &self, scheme: DiscretizationScheme, t: f64, x: f64, dt: f64, dw: f64, ) -> f64

Advance the state by one step of scheme: the generic stepping, written against drift/diffusion alone. Override only to exploit model structure (e.g. one volatility lookup shared by the coefficients, or a scheme the enum cannot express).

Dyn Compatibility§

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors§