Add comparison function

[richfitz]

#Actual Model
pkgload::load_all()

data <- process_data(readRDS("tests/testthat/weekly.rds"))

pars <- cowflu_inputs(alpha = 0.05, beta = 0.35, gamma = 0.2, sigma = 0.5, asc_rate = 1,
                      dispersion = 1,
                      cowflu_fixed_inputs(p_region_export = movement$p_region_export,
                                          p_cow_export = movement$p_cow_export,
                                          movement_matrix = movement$movement_matrix,
                                          time_test = 136,
                                          n_herds_per_region = usda_data$n_herds_per_region,
                                          n_cows_per_herd = usda_data$n_cows_per_herd,
                                          start_herd = 26940, #26804 is where Texas starts.
                                          start_count = 5,
                                          condition_on_export = TRUE
                                          ))
#pars$time_test <- 136
#pars$n_test <- 0

#Texas starts at herd: usda_data$region_start[41] + 1  = 26804,
#this is with 4 cows in it.
#Currently, if you try and put more start cows in than there exists, it'll do it anyway and have negative in the S.
##TODO: Make a check to catch this.
set.seed(1)
n_particles <- 2
sys <- dust2::dust_system_create(cows(), pars, n_particles = n_particles, dt = 1) # n_threads for parallelism
dust2::dust_system_set_state_initial(sys)

## Check on doing comparison to data; this is just for debugging really.
## Here assuming that you are working with days:
dust2::dust_system_run_to_time(sys, data$day[[1]])
dust2::dust_system_compare_data(sys, list(positive_tests = data$positive_tests[[1]]))

## Run a particle filter
data_day <- dust2::dust_filter_data(data, time = "day")
filter <- dust2::dust_filter_create(cows(), 0, data_day, n_particles = 20, n_threads = 8)
ll <- dust2::dust_likelihood_run(filter, pars)

## or equivalently:
data$time <- data$week
filter <- dust2::dust_filter_create(cows(), 0, data, n_particles = 20, n_threads = 8)
ll <- dust2::dust_likelihood_run(filter, pars)

[richfitz]

#Actual Model
pkgload::load_all()

data <- process_data(readRDS("tests/testthat/weekly.rds"))

## WEEKLY RATES EVERYWHERE!
pars <- cowflu_inputs(alpha = 0.35,
                      beta = 2.45,
                      gamma = 1.4,
                      sigma = 3.5,
                      asc_rate = 1,
                      dispersion = 1,
                      cowflu_fixed_inputs(p_region_export = movement$p_region_export,
                                          p_cow_export = movement$p_cow_export,
                                          movement_matrix = movement$movement_matrix,
                                          time_test = 136,
                                          n_herds_per_region = usda_data$n_herds_per_region,
                                          n_cows_per_herd = usda_data$n_cows_per_herd,
                                          start_herd = 26940, #26804 is where Texas starts.
                                          start_count = 5,
                                          condition_on_export = TRUE
                                          ))
#pars$time_test <- 136
#pars$n_test <- 0

#Texas starts at herd: usda_data$region_start[41] + 1  = 26804,
#this is with 4 cows in it.
#Currently, if you try and put more start cows in than there exists, it'll do it anyway and have negative in the S.
##TODO: Make a check to catch this.
set.seed(1)
n_particles <- 2
sys <- dust2::dust_system_create(cows(), pars, n_particles = n_particles, dt = 1) # n_threads for parallelism
dust2::dust_system_set_state_initial(sys)

## Check on doing comparison to data; this is just for debugging really.
## Here assuming that you are working with weeks:
dust2::dust_system_run_to_time(sys, data$week[[1]])
dust2::dust_system_compare_data(sys, list(positive_tests = data$positive_tests[[1]]))

data_week <- dust2::dust_filter_data(data, time = "week")
filter <- dust2::dust_filter_create(cows(), 0, data_week,
                                    dt = 0.1, n_particles = 20, n_threads = 8)
ll <- dust2::dust_filter_run(filter, pars)