Fastet way to position a double in an existing vector-CodePudding

Let's say I have a double like 3.5 and I would like to find out where to sort it in an existing sorted vector say seq(1, 10), put differently, which index the new number would take in the vector. Of course it sits somewhere between 3 and 4 and hence between the third and fourth index, but what would be the fastet way to arrive at this result?

CodePudding user response：

As mentioned in the comments, findInterval is fastest for this purpose. Even a very simple loop in C that does the same thing is a little slower on average.

library(Rcpp)

cppFunction("int find_index(double x, NumericVector v) {
            int len = v.size();
            for(int i = 0; i < len;   i) {
             if(x <= v[i]) return i   1;
            }
            return NA_INTEGER;
}")

microbenchmark::microbenchmark(
  findInterval = findInterval(453993.5, 1:1000000),
  find_index   = find_index(453993.5, 1:1000000)
)
#> Unit: milliseconds
#>          expr    min     lq      mean  median     uq      max neval
#>  findInterval 1.9646 2.1739  2.996931 2.32375 2.4846  37.4218   100
#>    find_index 2.2151 2.4502 11.319199 2.60925 2.9800 337.9229   100

CodePudding user response：

Something like this?

First define dbl and my_seq then concatenate both with c(dbl, my_seq) and wrap it with sort

then define the index with which(my_vec == dbl):

dbl <- 3.5
my_seq <- seq(1,10)
my_vec <- sort(c(dbl, my_seq))
index <- which(my_vec == dbl)
index

output:

[1] 4