Prepare Reusable Numeric Exponential Kernel Calculations

Validate and capture a numeric exponential kernel and baseline once for repeated deterministic calculations.

Usage

exp_kernel_plan(kernel, mu)

Arguments

kernel

An exp_kernel_spec or convertible numeric hspec.

mu

Finite nonnegative baseline, scalar or one value per event type.

Value

A locked environment with two functions: step(component, dt, type = NULL, mark = 1, integrate = TRUE) and wait(component, residual). Their results match exp_kernel_step and exp_kernel_wait.

Details

Later edits to the original inputs do not change the plan. Every call still checks the current component and changing arguments. Components are passed explicitly; no current state, histories or random generator are stored or advanced. Create a new plan when kernel parameters or baseline change. This experimental numeric interface requires equal beta values within each row. The likelihood engine option and existing event loops are unchanged.

Examples

kernel <- exp_kernel_spec(.2, 1)
plan <- exp_kernel_plan(kernel, mu = .3)
component <- matrix(.1)
dt <- plan$wait(component, residual = .7)
next_state <- plan$step(component, dt, type = 1)
next_state$after
#>           [,1]
#> [1,] 0.2129613