C++ or Rcpp: comparison of two vectors without loop - c++

Question

I am a novice in C++ and Rcpp, and I am wondering how to compare each element of two different vectors without loop at one time.
My goal is to change the element of v1 by referencing other vector.`
Current code is
v1 = {6,7,8,9,10}
v2 = {2,4,6,8,10}
v3 = {a,b,a,b,c}
v4 = {0,0,0,0,0}
v5 = {a,b,c}
v6 = {1,2,3}
for (i in 1:5){
if (v1[i] > v2[i]){
for (j in 1:3){
if (v5[j] == v3[i]){
v4[i] = v2[i] + v6[j]
if (v1[i] > v4[i]){
v1[i] = v4[i]
}
}
}
}
}
The result sould be
v1 = {3,6,7,9,10}
In fact, v1, v2, v3, v4 and v5, v6 are the different dataframe in R. Each element of v1 is compared to v2. If an element i in v1 is larger than i element in v2, the element of v1 becomes a sum of i element of v1 and element of v6 by corresponding v3 & v5. Then the newly estimated value v4[i] is compared to v1[i].
I have ta large number of cases in v1~v5 and v5~v6. In this case, using loop takes a long time. Is it possible to compare the different vectors without loop? or how to estimate and reference the other vector's element?

Answer 1

I do not see the need to use Rcpp or C++ here. The way I understand your requirements, you are trying to manipulate two sets of equal length vectors. For a "set of equal length" vectors one normally uses a data.frame or one of its extensions. Here I am using base R, data.table and dplyr with tibble. See for yourself which syntax you prefer. Generally speaking, data.table will most likely be faster for large data sets.
Setup data:
v1 <- c(6,7,8,9,10)
v2 <- c(2,4,6,8,10)
v3 <- c("a","b","a","b","c")
v5 <- c("a","b","c")
v6 <- c(1,2,3)
Base R:
df1 <- data.frame(v1, v2, v3)
df2 <- data.frame(v5, v6)
df1 <- merge(df1, df2, by.x = "v3", by = "v5")
df1$v4 <- df1$v2 + df1$v6
df1$v1 <- ifelse(df1$v1 > df1$v2 & df1$v1 > df1$v4, df1[["v4"]], df1[["v1"]])
df1
#> v3 v1 v2 v6 v4
#> 1 a 3 2 1 3
#> 2 a 7 6 1 7
#> 3 b 6 4 2 6
#> 4 b 9 8 2 10
#> 5 c 10 10 3 13
data.table:
library(data.table)
dt1 <- data.table(v1, v2, v3, key = "v3")
dt2 <- data.table(v5, v6, key = "v5")
dt1[dt2, v4 := v2 + v6]
dt1[v1 > v2 & v1 > v4, v1 := v4]
dt1
#> v1 v2 v3 v4
#> 1: 3 2 a 3
#> 2: 7 6 a 7
#> 3: 6 4 b 6
#> 4: 9 8 b 10
#> 5: 10 10 c 13
dplyr:
suppressPackageStartupMessages(library(dplyr))
t1 <- tibble(v1, v2, v3)
t2 <- tibble(v5, v6)
t1 %>%
inner_join(t2, by = c("v3" = "v5")) %>%
mutate(v4 = v2 + v6) %>%
mutate(v1 = case_when(
v1 > v2 & v1 > v4 ~ v4,
TRUE ~ v1
))
#> # A tibble: 5 x 5
#> v1 v2 v3 v6 v4
#> <dbl> <dbl> <chr> <dbl> <dbl>
#> 1 3 2 a 1 3
#> 2 6 4 b 2 6
#> 3 7 6 a 1 7
#> 4 9 8 b 2 10
#> 5 10 10 c 3 13
Created on 2019-04-19 by the reprex package (v0.2.1)
The general idea is always the same:
join the two tables on the character column
create new column v4 as sum of v2 and v6
update v1 to the value of v4 where v1 > v2 and v1 > v4
Note that base R and data.table do not preserve the order, so it would make more sense to put the output into an additional column.