Here, we defined a "pure function" to be one that has no side effects – it is like the general map operation we saw in data parallel programming, where transformation function is applied individually to each input in a set of inputs to produce a set of modified outputs, with each element's computation happening fully independently from others (which allows for great parallelization behavior).
Here, we defined a "pure function" to be one that has no side effects – it is like the general
map
operation we saw in data parallel programming, where transformation function is applied individually to each input in a set of inputs to produce a set of modified outputs, with each element's computation happening fully independently from others (which allows for great parallelization behavior).