Whereas with mapreduce, the time and allocated memory are much larger. Consider the same N and elems as above, then the performance I get for the two variants is begin
The performance of the variant shown above is unexpectedly much better than that using mapreduce(). J = reduce(vcat, map(e -> e], elems)) įor this simplified example (in reality elems may not have this simple structure), the output should be elems =, ,.
I want to create an array of indices (for creation of a sparse matrix in a later stage) based on a list of element connectivity as follows: N = 10000 # Number of elementsĮlems = collect( for i in 1:N) # List of element connectivity
