contextual do_parallel dosnt work on MRAN 3.5.1 about contextual

[g0ulash]

Can you show the complete code that you are trying to run? That would make debugging a tad easier.

[fhujiil]

Here it is :
prob_per_arm <- c(0.1, 0.1, 0.1, 0.1, 0.9)
horizon <- 250
simulations <- 1000

bandit <- BasicBernoulliBandit$new(weights = prob_per_arm)

agent <- list(Agent$new(EpsilonGreedyPolicy$new(0.1), bandit, "Epsilon = 0.1"),
Agent$new(EpsilonGreedyPolicy$new(0.2), bandit, "Epsilon = 0.2"),
Agent$new(EpsilonGreedyPolicy$new(0.3), bandit, "Epsilon = 0.3"),
Agent$new(EpsilonGreedyPolicy$new(0.4), bandit, "Epsilon = 0.4"),
Agent$new(EpsilonGreedyPolicy$new(0.5), bandit, "Epsilon = 0.5"))
history <- Simulator$new(agents = agent,
horizon = horizon,
simulations = 1000,do_parallel = T)$run()

[robinvanemden]

Hi @fhujiil -thanks for reporting this issue! It seems that using setMKLthreads() in MRO 3.5.3 in combination with RStudio crashes R. See also here.

I was able to resolve the issue by skipping "setMKLthreads(1)" when both RStudio and MRAN are detected .

This might degrade performance when running contextual on MRAN+RStudio though - as contextual's parallel processing and MRAN's multi-threaded linear algebra computations will now compete for the same processors.

This solution will be in contextual's next CRAN release, probably sometime next month. To use the updated code right away, just install from GitHub:

install.packages("devtools")
devtools::install_github('Nth-iteration-labs/contextual')