Estimate quasi-potential dynamics on decomposition coordinates

Fits \(dX/dt = -\nabla U(X;\,\text{layer}) + \text{noise}\). Critical points = commitment / irreversibility; barrier height = transition difficulty. Cross-sectional (destructive) sampling is the native data type, not a workaround.

Usage

estimate_dynamics(strategy, data, ...)

# S4 method for class 'DynamicsEstimator,StateTransitionData'
estimate_dynamics(strategy, data, ...)

# S4 method for class 'KdeLogDensityEstimator,StateTransitionData'
.estimate_dynamics_impl(strategy, data, ...)

# S4 method for class 'KdeLogDensityEstimator'
supported_sampling_designs(strategy)

Arguments

strategy

a DynamicsEstimator implementation

data

StateTransitionData with decomposition in metadata()

...

forwarded to the implementation

Value

StageResult whose value is data annotated with quasi-potential, critical points, and barrier heights in metadata()

Details

Boundary validation (validate_boundary) is enforced structurally, the same way as for decompose: the method below, dispatched on the VIRTUAL DynamicsEstimator class itself, runs before any strategy-specific code. Concrete strategies implement the internal .estimate_dynamics_impl generic instead of estimate_dynamics() directly, so validation cannot be skipped by omission.