R Programming By Example

Object immutability

Improving the speed of R implementations does not necessarily involve advanced optimization techniques, such as parallelization. Indeed, there are a number of simple tweaks that, while not always obvious, can make R run significantly faster. The top bottleneck people encounter with R is a lack of understanding about its object immutability property, and the overhead incurred when making copies of such objects. Simply taking care of this can produce dramatic performance improvements, and it's not too difficult once you understand how to do so. This is a good candidate to start looking for optimizations.

As an example of some of the issues that can arise, suppose you have an array of numbers named a. Now, suppose you want ...