It would be useful to be able to specify the simulated innovations. Often I'm combining ARIMA models for some variables with different models for others and using a copula framework to simulated the system.
Currently, to simulate 1 step ahead, 1000 times:
ar1 = SARIMA(y,order=(1,0,0),include_mean=true)
StateSpaceModels.fit!(ar1)
ar_sim = simulate_scenarios(ar1,1,1000)[1,1,:]
Would be helpful to allow something like:
ar_sim = simulate_scenarios(ar1,1,1000; innovations=randn(1000))
generically:
ar_sim = simulate_scenarios(ar1,n,m; innovations=r)
where r is (n,m)