Module processes 

Equity model dynamics as StochasticProcess1D / StochasticProcess implementations — the bridge between the model layer (GBM, Dupire local vol, Heston) and the generic stepping in core::montecarlo::process.

The Monte Carlo engine consumes these; the same objects can feed any future coefficient-driven method (finite difference via the Feynman-Kac link, other samplers) without touching the models.

Structs

BlackScholesProcess

Risk-neutral lognormal dynamics dS = (r - q) S dt + sigma(S, t) S dW with constant or local volatility.

HestonProcess

Heston dynamics as a 2-state, 2-factor process, state [S, v]:

MultiAssetGbmProcess

N correlated lognormal assets as one N-state, N-factor process:

Enums

HestonScheme

Heston-specific stepping schemes — model-owned, because good variance stepping is genuinely model-specific and no generic Euler/Milstein switch covers it (the QuantLib / TF Quant Finance pattern).

VolDynamics

Volatility dynamics along a path: constant (GBM) or Dupire local vol.

Functions

qe_variance_step

One Andersen QE draw of the CIR variance transition v_{t+dt} | v_t = v from a standard normal z_v — the sampler matches the exact conditional mean and variance of the square-root process, switching between a squared-Gaussian and a mass-at-zero/exponential form. Public for consumers that step the variance leg on its own (the SLV engine pairs it with a leverage-adjusted spot step).