Parallel Computing for Data Science

70 CHAPTER 3. SCHEDULING

The package has the user set up a for loop, as in serial code, but then use

the foreach() function instead of for(). One must also make one more

small change, adding an operator, %dopar%, but that’s all the user must

do to parallelize his/her serial code.

Thus foreach has a very appealing simplicity. However, in some cases, this

simplicity can mask major opportunities for achieving speedup, as will be

seen in the example in the next section.

3.10.1 Example: Mutual Outlinks Problem

Here is foreach code for the mutual outlinks problem.

mutoutfe <− function ( l i n k s ) {

require ( f o r e a c h )

nr <− nrow( l i n k s )

nc <− nc