Simulate event times from a fitted Hawkes process

Draws a new event history from a chwkfit() result by its cluster representation: immigrant events from the baseline (a Poisson process, thinned from its upper bound for a sinusoidal baseline), then each event's offspring, a Poisson number with mean the branching ratio, at delays drawn from the fitted kernel, generation after generation, until none fall before the horizon.

Usage

chwksim(fit, hor = NULL, seed = 1L)

Arguments

fit

A chwkfit() result.

hor

Number, the end of the window; default the fit's horizon.

seed

Whole number, default 1, for this call only (the session's stream is left as it was).

Value

The sorted event times (numeric), with the attribute generation (0 for an immigrant, 1 for its children, ...).

See also

plothwk(), chwkfit().

Examples

set.seed(1)
times <- sort(stats::runif(300, 0, 100))
fit <- chwkfit(times, 100)
sim <- chwksim(fit, seed = 2)
length(sim)
#> [1] 286
table(attr(sim, "generation"))
#> 
#>   0   1   2   3   4 
#> 213  54  14   3   2 
try(chwksim(list(a = 1)))
#> Error : chwksim: `fit` must be a chwkfit() result