Chunking migration application on worker's side

The migration application from the first chapter is an example of the third type of chunking, namely when the workers "see" the whole dataset and only get instructions which part of the data to work on. Here, you will parallelize the code. Earlier you defined a function ar1_multiple_blocks_of_trajectories() that takes a vector of row identifiers as argument and generates migration trajectories using the corresponding rows of the parameter set ar1est.

ar1_multiple_blocks_of_trajectories() depends on ar1_block_of_trajectories() which in turn depends on ar1_one_trajectory(). These functions along with the cluster object cl of size 4, function show_migration(), the dataset ar1est (reduced to 200 rows) and packages parallel and ggplot2 are available in your workspace.

Export the dataset ar1est and the global functions ar1_block_of_trajectories() and ar1_one_trajectory() to the workers.
Use clusterApply() to run ar1_multiple_blocks_of_trajectories() in parallel where each parallel task processes one row of ar1est. Thus, the x argument to clusterApply() should be a vector from 1 to the number of rows of ar1est. Use default values for the starting rate, length of trajectories and block size.
Combine results into a matrix using rbind and plot results using the show_migration() function.

Take Hint (-30 XP)

Exercise

Chunking migration application on worker's side

Instructions