Allow dt > 1

[weshinsley]

Is there anything else that needs doing for this to work?
Anything I've missed?

[richfitz]

Thanks - this looks much simpler than expected. Now for the wrinkle (which we can put into another PR if you prefer)

The motivating reason for this is the malaria mixed ODE/discrete time model, where dt represents the time step for the discrete part of the model. In this case we want to allow stopping at any time. So this (nonexistant) test should pass:

test_that("can apply stochastic updates by setting dt", {
  sys <- dust_system_create(malaria(), list(), n_particles = 10, dt = 10)
  dust_system_set_state_initial(sys)
  t <- seq(0, 20)
  y <- dust_system_simulate(sys, t)
  i_beta <- dust_unpack_index(sys)$beta
  beta <- y[i_beta, , ]

  ## Equivalence to a deterinistic model:
  sys0 <- dust_system_create(malaria(), list(), n_particles = 10)
  for (i in seq_len(length(t) - 1)) {
    dust_system_set_state(sys0, y[, , i])
    dust_system_run_to_time(sys0, t[[i + 1]])
    y0 <- dust_system_state(sys0)
    expect_equal(y0[-i_beta, ], y[-i_beta, , i + 1])
  }
})

[weshinsley]

Well - that test does now work by just not calling the checks that would prevent it from doing so if dt > 1 - not sure if this is the answer. But also not sure what checks should/can be carried out at that point (interface.R:700) if dt > 1...