In cboettig/multiple_uncertainty: Multiple Uncertainty

library("dplyr")
library("tidyr")
library("ggplot2")
library("multipleuncertainty")
library("parallel")

With parallel BLAS libraries enabled, parts of the multiple_uncertainty calculation are already parallelized.
On machines with small numbers of cores, it might make sense to run the multiple_uncertainty

fig3 <- function(noise){
  grid <- seq(0,150, 1)
  o <- mclapply(
    list(small = c(g = 0.1, m = 0.1, i = 0.1),
         growth = c(g = 0.5, m = 0.1, i = 0.1),
         measure = c(g = 0.1, m = 0.5, i = 0.1),
         implement = c(g = 0.1, m = 0.1, i = 0.5)), 
    function(s)
      multiple_uncertainty(f = "logistic", x_grid = grid, sigma_g = s[["g"]], sigma_m = s[["m"]], sigma_i = s[["i"]], noise_dist = noise),
    mc.cores = parallel::detectCores())

  df <- data.frame(y_grid = grid, small = o$small, growth = o$growth, 
                   measure = o$measure, implement = o$implement) %>%
    tidyr::gather(scenario, value, -y_grid)
}

system.time(
df <- 
data.frame(noise = c("uniform", "lognormal")) %>%
  dplyr::group_by(noise) %>%
  dplyr::do(fig3(.$noise))
)

df %>% ggplot(aes(x = y_grid, y = value, col = scenario)) + 
    geom_line()  + 
    facet_wrap(~ noise) + 
    xlab("Stock") + 
    ylab("Escapement") + 
    coord_cartesian(xlim = c(0, 150), ylim = c(0,100)) + 
    theme_bw()

One option is to parallelize the do loop with multidplyr, however, this requires essentially ncores times as much memory, which is very inefficient.

cl <- get_default_cluster()
cl %>% cluster_assign_value("fig3", fig3) %>% cluster_library("multipleuncertainty")

expand.grid(cost = c(0, 0.02, 0.2),
            dr = c(0, 0.1),
            noise = c("uniform", "lognormal")) %>%
  multidplyr::partition() %>%
  dplyr::group_by(cost, dr, noise) %>%
  dplyr::do(fig3(.$cost, .$dr, .$noise)) -> df

df %>% 
  dplyr::filter(noise == "uniform") %>%
  ggplot(aes(x = y_grid, y = value, col = scenario)) + 
    geom_line()  + 
    facet_grid(cost ~ dr) + 
    xlab("Stock") + 
    ylab("Escapement") + 
    coord_cartesian(xlim = c(0, 150), ylim = c(0,100)) + 
    theme_bw()

Profiling

#library("profvis")
Rprof("multiple_uncertainty.out")
#profvis::profvis({
#  devtools::load_all()

grid <- seq(0, 200, by = 0.5)
growth <- multiple_uncertainty(x_grid = grid, sigma_g = 0.5, sigma_m = 0.1, sigma_i = 0.1)

#})
Rprof(NULL)  
pd <- readProfileData("multiple_uncertainty.out")