Question

Is there a set of R functions analogous to intersect, union, etc. which act on vectors of coordinates? Take a two collections of locations in N dimensions, each of which could be viewed as N-row by M-location array, and find the common locations (i.e. columns), and so on. I suppose this could be done with a couple *apply lines to compare every column in one array to the columns of the other, but was hoping for something faster or neater.

Answer 1

There is a ‘sets’ package, but the simplest solution would be to use the base set operations: union, , intersect , and setdiff on the results of paste with a separator such as "_". If these are represented as ‘numeric” values you may need to make decsions about what really means “equality” and for that some round-ing or trunc-cation may be needed.

Another thought is that implementation might be more self-documenting in the columnwise arrangement you describe to use the “plyr” function:colwise. I’m not a dedicated user of plyr methods, but this appears to be how it might start:

colpastefn <- colwise( function(...){paste(... , sep="_") )

Although the description of the function suggests it should work as a standalone function on a dataframe, all of the illustrated uses are with ddply.